wtdgb2-mpi 分布式版本

.gitignore

@@ -0,0 +1,21 @@
+*.o
+*.a
+.*.swp
+*.dSYM
+.vscode
+F1.fasta
+euler.py
+kbm2
+map2dbgcns
+out
+pgzf
+reads.fa.gz
+reads.kbmap
+reads.kbmap2
+run.done
+run_wtdbg2.sh
+ttr_finder
+wtdbg-cns
+wtdbg2
+wtdmo
+wtpoa-cns

.travis.yml

@@ -0,0 +1,5 @@
+language: c
+compiler:
+  - gcc
+  - clang
+script: make

LICENSE.txt

@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <http://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<http://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<http://www.gnu.org/philosophy/why-not-lgpl.html>.

Makefile

@@ -0,0 +1,50 @@
+VERSION=2.5.1
+RELEASE=20200511
+
+CC  := mpicc
+BIN := /usr/local/bin
+DEBUG := 0
+
+ifeq (0, ${MAKELEVEL})
+TIMESTAMP=$(shell date)
+endif
+
+ifeq (1, ${DEBUG})
+CFLAGS=-g3 -W -Wall -Wno-unused-but-set-variable -O0 -DDEBUG=1 -DVERSION="$(VERSION)" -DRELEASE="$(RELEASE)" -D_FILE_OFFSET_BITS=64 -D_GNU_SOURCE -mpopcnt -msse4.2
+else
+CFLAGS=-W -Wall -Wno-unused-but-set-variable -O4 -DVERSION="$(VERSION)" -DRELEASE="$(RELEASE)" -D_FILE_OFFSET_BITS=64 -D_GNU_SOURCE -mpopcnt -msse4.2
+endif
+
+GLIBS=-lm -lrt -lpthread -lz
+GENERIC_SRC=mem_share.h chararray.h sort.h list.h pgzf.h sort.h list.h dna.h thread.h filereader.h filewriter.h bitvec.h bit2vec.h bitsvec.h hashset.h
+
+PROGS=kbm2 wtdbg2 wtdbg-cns wtpoa-cns pgzf
+
+all: $(PROGS)
+
+kbm2: $(GENERIC_SRC) kbm.c kbm.h kbmpoa.h wtpoa.h tripoa.h poacns.h kswx.h ksw.h ksw.c
+	$(CC) $(CFLAGS) -o $@ kbm.c ksw.c $(GLIBS)
+
+wtdbg2: $(GENERIC_SRC) wtdbg.c wtdbg-graph.h wtdbg.h kbm.h kswx.h ksw.h ksw.c kbmpoa.h wtpoa.h tripoa.h poacns.h
+	$(CC) $(CFLAGS) -o $@ wtdbg.c ksw.c $(GLIBS)
+
+wtdbg-cns: $(GENERIC_SRC) wtdbg-cns.c kswx.h ksw.h ksw.c dbgcns.h dagcns.h queue.h general_graph.h
+	$(CC) $(CFLAGS) -o wtdbg-cns wtdbg-cns.c ksw.c $(GLIBS)
+
+wtpoa-cns: $(GENERIC_SRC) wtpoa.h wtpoa-cns.c poacns.h tripoa.h ksw.h ksw.c
+	$(CC) $(CFLAGS) -o $@ wtpoa-cns.c ksw.c $(GLIBS)
+
+pgzf: mem_share.h sort.h list.h thread.h pgzf.h pgzf.c
+	$(CC) $(CFLAGS) -o $@ pgzf.c $(GLIBS)
+
+best_sam_hits4longreads: $(GENERIC_SRC) best_sam_hits4longreads.c
+	$(CC) $(CFLAGS) -o $@ best_sam_hits4longreads.c $(GLIBS)
+
+clean:
+	rm -f *.o *.gcda *.gcno *.gcov gmon.out $(PROGS)
+
+clear:
+	rm -f *.o *.gcda *.gcno *.gcov gmon.out
+
+install: $(PROGS)
+	mkdir -p $(BIN) && cp -fvu $(PROGS) $(BIN)

README-ori.md

@@ -0,0 +1,289 @@
+# NEWS
+* |2018-09-21| Rename wtdbg-1.2.8 to wtdbg2
+* |2018-09-19| GFA supports
+```sh
+wtdbg-dot2gfa.pl dbg.3.dot >dbg.3.gfa
+```
+* |2018-09-01| New consensus module: POACNS <br>
+```sh
+wtpoa-cns -t 0 -i dbg.ctg.lay -fo dbg.ctg.poacns.fa
+```
+1, wtpoa-cns implements POA to generate MSA from reads fragments within an edge. Used SSE instructions, but seems still can be further improved in speed <br>
+2, performs realignment on MSA. <br>
+3, recalibrates homopolymers, and make consensus sequence for an edge. Most reamining small errors come from homopolymer, more efforts are needed <br>
+4, joins edges' sequences into a contig. <br>
+
+Welcome to test POACNS and feedback.
+
+# WTDBG
+
+A fuzzy Bruijn graph (FBG) approach to long noisy reads assembly
+
+# Introduction
+A challenge in assembling long noisy reads from third generation sequencing (TGS) is reducing its requirement of computing resource, especially for large genomes.
+To address this issue, I developed a novel sequence alignment algorithm and a new assembly graph for efficiently assembling large genomes using TGS data.
+
+* Alignment <br>
+KBM: Kmer-BIN-Mapping.<br>
+KBM groups k-mers from each non-overlapped sliding 256 bp fragments in long reads into bins.
+Bins of which most k-mers are high frequency, are filtered as highly repetitive ones.
+Then, KBM searches synteny of matched bin pairs in sequences in a dynamic programming way.
+A matched bin pair in two sequences is defined as two bins different by original but share a set of k-mers.
+The result of alignments in KBM have the same features of traditional sequence alignment, excepting the unit of KBM alignments is 256 bp bin instead of single base.
+
+* Assembly <br>
+FBG: Fuzzy Bruijn Graph. <br>
+FBG is composed of vertices in length of 1024 bp from reads, and edges connecting vertices in their order on read paths.
+Comparing with DBG, the size of vertices in FBG are much bigger, thus won¡¯t be sensitive to small repeat.
+To tolerate high sequencing errors, FBG's vertices are found using gapped sequence alignments from KBM or other aligners, comparing with searching identical k-mers in DBG.
+
+* Why choose wtdbg in genome assembly <br>
+There are many assemblers for long noisy reads assembly, e.g. FALCON, CANU, miniasm, and SMARTdenovo (progenitor of wtdbg). If you have a genome of 10G bp or bigger in size,
+wtdbg is your first or even the only option. For small but complicated genomes (< 3 G), wtdbg was often reported to yield better assembly by my friends.
+Besides, KBM is easy to use when you are setting up a web-server for long reads mapping (see Example 2).
+
+* Limitation <br>
+Max read length is 0x0003FFFFU (256 Kb), longer reads will be split. <br>
+Max number of reads is 0x03FFFFFFU (64 M). If your data volume exceeds, please filter relative shorter reads. <br>
+In KBM, max read length is 0xFFFFFFFFU (4 Gb), max number of reads is 0x0FFFFFFFU (256 M). <br>
+Max number of threads is 4096. <br>
+Cannot parallelly run in multiple nodes. But you can implement it simplely using `kbm` and `wtdbg --load-alignments` <br>
+Developed and tested in Linux-GCC only. <br>
+Only accepts fasta/fastq format for input, '.gz' suffixed files will be piped by `gzip -dc`.
+
+# Installation
+```sh
+git clone https://github.com/ruanjue/wtdbg2.git
+cd wtdbg2
+make
+```
+
+# Long reads mapping
+
+Supposes you have `hg19.fa` as reference sequences, and `reads.fq.gz` as query sequences.
+## Example 1
+```sh
+kbm -t 64 -d hg19.fa -i reads.fa.gz -o reads.kbmap
+
+```
+### output format
+* COL1  `qry_name`
+* COL2  `qry_strand`
+* COL3  `qry_length`
+* COL4  `qry_beg`
+* COL5  `qry_end`
+* COL6  `ref_name`
+* COL7  `ref_strand` (always equals `+`)
+* COL8  `ref_length`
+* COL9  `ref_beg`
+* COL10 `ref_end`
+* COL11 `match_len` (length of matched k-mers)
+* COL12 `align_len` (length of aligned)
+* COL13 `#kcnt`     (number of matched k-mers)
+* COL14 `#gap`      (number of gapped BINs)
+* COL15 `cigar`     (256 x SAM's cigar)
+
+## Example 2
+Suitable for online tools, or frequently used references
+### Build KBM-INDEX
+```sh
+kbm -t 64 -d hg19.fa -i /dev/null -W hg19.kbmidx
+```
+### Set up KBM server
+```sh
+kbm -R hg19.kbmidx start
+```
+Now, hg19.kbmidx is cached in memory for further call.
+
+### Mapping with KBM-INDEX
+```sh
+kbm -R hg19.kbmidx -t 64 -i reads.fa -o reads.kbmap
+```
+Please note that, hg19.kbmidx can be multilple called by any processes in the same computer. <br>
+
+### Shutdown KBM server
+```sh
+kbm -R hg19.kbmidx stop
+```
+
+# Long reads assembling
+
+## Quick Start
+```sh
+echo "manual"
+run_wtdbg_assembly.sh -h
+echo "generating shell script"
+run_wtdbg_assembly.sh -t 0 -i reads.fa.gz -o dbg -T >run.sh
+```
+`run.sh` is ready for invoked
+
+## Play with wtdbg
+```sh
+wtdbg2 -h
+wtdbg2 --help
+```
+### options
+```sh
+ -t <int>    Number of threads, 0: all cores, [0]
+ -i <string> Long reads sequences file, + *
+ -I <string> Error-free sequences file, +
+ -o <string> Prefix of output files, *
+ -f          Force overwrite
+ -k <int>    Kmer fsize, 0 <= k <= 25, [0]
+ -p <int>    Kmer psize, 0 <= p <= 25, [21]
+             k + p <= 25, seed is <k-mer>+<p-homopolymer-compressed>
+ -K <float>  Filter high frequency kmers, maybe repetitive, [1000]
+             if K >= 1, take the integer value as cutoff
+             else, mask the top fraction part high frequency kmers
+ -E <int>    Min kmer frequency, [2]
+ -F          Filter low frequency kmers by a 4G-bytes array (max_occ=3 2-bits). Here, -E must greater than 1
+ -S <int>    Subsampling kmers, 1/(<-S>) kmers are indexed, [4]
+             -S is very useful in saving memeory and speeding up
+             please note that subsampling kmers will have less matched length
+ -X <int>    Max number of bin(256bp) in one gap, [4]
+ -Y <int>    Max number of bin(256bp) in one deviation, [4]
+ -x <int>    penalty for BIN gap, [-7]
+ -y <int>    penalty for BIN deviation, [-21]
+ -l <float>  Min length of alignment, [2048]
+ -m <float>  Min matched, [200]
+ -s <float>  Max length variation of two aligned fragments, [0.2]
+ -q          Quiet
+ -v          Verbose, +
+ --help      Show more options
+```
+
+### For higher error rate long sequences
+Decrease `-p`. Try `-p 19` or `-p 17` <br>
+Decrease `-S`. Try `-S 2` or `-S 1` <br>
+Both will increase computing time.
+
+### For very high coverage
+Increase `--edge-min`. Try `--edge-min 4`, or higher.
+
+### For low coverage
+Decrease `--edge-min`. Try `--edge-min 2 --rescue-low-cov-edges`.
+
+### Filter reads
+`--tidy-reads 5000`. Will filtered shorter sequences. If names in format of `\/\d+_\d+$`, will selected the longest subread.
+
+### output
+Suppose the prefix is `dbg`
+* dbg.1.dot <br>
+DOT file for initialized graph
+* dbg.1.nodes <br>
+nodes and their positions in reads
+* dbg.1.reads <br>
+reads and their nodes
+* dbg.2.dot <br>
+DOT file after transitive reduction
+* dbg.3.dot <br>
+DOT file after merging bubble and remove tips
+* dbg.alignments <br>
+KBMAP file, all vs all alignments
+* dbg.binkmer <br>
+Distribution of number of k-mers in a BIN
+* dbg.closed\_bins <br>
+Filtered BINs
+* dbg.clps
+Reads clip information. <br>
+COL1 read\_name <br>
+COL2 read\_length <br>
+COL3 keep\_offset <br>
+COL4 keep\_length
+* dbg.ctg.dot <br>
+DOT file for contigs
+* dbg.ctg.lay <br>
+Contigs layout file. Will be read by `wtdbg-cns`. This file is the main result of `wtdbg` <br>
+**Format**:<br>
+```
+>ctg(\d+) nodes=(\d+) len=(\d+)
+E <OFFSET> <NODE1> <STRAND1> <NODE2> <STRAND2>
+S <READ_NAME> <STRAND> <REG_OFFSET> <REG_LENGTH> <REG_SEQ>
+S <READ_NAME> <STRAND> <REG_OFFSET> <REG_LENGTH> <REG_SEQ>
+S <READ_NAME> <STRAND> <REG_OFFSET> <REG_LENGTH> <REG_SEQ>
+...
+E ...
+...
+
+```
+One contig contains many edges (starting with 'E'), each edge contains many regions inside reads. <br>
+Please note that one read often contains many REGs.
+
+* dbg.events <br>
+Log file of graph simplification
+* dbg.frg.dot <br>
+DOT file for unitigs
+* dbg.frg.nodes <br>
+unitigs and their nodes
+* dbg.kmerdep <br>
+Distribution of k-mer depth
+* STDERR stream <br>
+wtdbg print runtime information on progrom's STDERR stream. `--quiet` to disiable it
+
+## Consensus
+```sh
+wtdbg-cns -t 64 -i dbg.ctg.lay -o dbg.ctg.lay.fa
+```
+The output file `dbg.ctg.lay.fa` is ready for further polished by `PILON` or `QUIVER`.
+
+```sh
+wtpoa-cns -t 64 -i dbg.ctg.lay -o dbg.ctg.lay.fa
+```
+wtpoa-cns is slower than wtdbg-cns, but offer more accurate consensus sequences.
+I will update it in following development.
+
+# Performance
+## Human (3G) CHM1 PacBio P5C3 dataset, 65.5 core.hours
+* Data Source
+http://datasets.pacb.com/2014/Human54x/fasta.html
+* Command
+```sh
+wtdbg2 -t 96 -i pb.fa -fo dbg --tidy-reads 5000 --edge-min 2 --rescue-low-cov-edges
+```
+* Contigs
+`TOT 2978536704, CNT 8752, AVG 340327, MAX 11662848, N50 1925120, L50 453, N90 400128, L90 1727, Min 5120`
+* Runtime
+`real 6131.803 sec, user 201836.200 sec, sys 33956.790 sec, maxrss 117281672.0 kB, maxvsize 202422172.0 kB`
+
+## Human (3G) CHM1 PacBio P6C4 dataset, 211.3 core.hours
+* Data Source
+http://www.ebi.ac.uk/ena/data/view/PRJNA246220
+* Command
+```sh
+wtdbg2 -t 96 -i wt.fa -fo dbg --tidy-reads 5000 --edge-min 4 --rescue-low-cov-edges
+```
+* Contigs
+`TOT 2964872448, CNT 1909, AVG 1553103, MAX 105310208, N50 23586816, L50 34, N90 3326976, L90 158, Min 5120`
+* Runtime
+`real 16806.534 sec, user 681278.770 sec, sys 79371.630 sec, maxrss 264956752.0 kB, maxvsize 443356532.0 kB`
+
+## Axolotl (32G) PacBio dataset, 32 X, 3053 core.hours
+* Command
+```sh
+wtdbg2 -t 96 -i ../rawdata/pacbio.fa.gz -p 21 -S 2 --aln-noskip --rescue-low-cov-edges --tidy-reads 5000 -fo axolotl
+```
+* Contigs
+`TOT 27375160576, CNT 115355, AVG 237313, MAX 7812608, N50 606976, L50 12527, N90 144896, L90 47295, Min 5120`
+* Runtime
+`real 190237.591 sec, user 10994200.800 sec, sys 488715.030 sec, maxrss 1671005352.0 kB, maxvsize 2365400208.0 kB`
+
+## Human (3G) NA12878 ONT dataset, 197.5 core.hours
+* Data Source
+https://github.com/nanopore-wgs-consortium/NA12878
+* Command
+```sh
+wtdbg2 -t 64 -i NA12878-ONT.fa.gz -fo dbg -S 2 --edge-min 2 --rescue-low-cov-edges
+```
+* Contigs
+`TOT 2827644928, CNT 19473, AVG 145209, MAX 31366400, N50 4540672, L50 162, N90 172800, L90 1111, Min 5120`
+
+* Runtime
+`real 14992.925 sec, user 649202.270 sec, sys 61638.300 sec, maxrss 256840096.0 kB, maxvsize 356668088.0 kB`
+
+# Citation
+To be published. <br>
+URL **https://github.com/ruanjue/wtdbg2/** <br>
+
+# Contact
+Jue Ruan <ruanjue@gmail.com> <br>
+Jue Ruan <ruanjue@caas.cn>

README.md

@@ -0,0 +1,138 @@
+## <a name="start"></a>Getting Started
+```sh
+git clone https://github.com/ruanjue/wtdbg2
+cd wtdbg2 && make
+#quick start with wtdbg2.pl
+./wtdbg2.pl -t 16 -x rs -g 4.6m -o dbg reads.fa.gz
+
+# Step by step commandlines
+# assemble long reads
+./wtdbg2 -x rs -g 4.6m -i reads.fa.gz -t 16 -fo dbg
+
+# derive consensus
+./wtpoa-cns -t 16 -i dbg.ctg.lay.gz -fo dbg.raw.fa
+
+# polish consensus, not necessary if you want to polish the assemblies using other tools
+minimap2 -t16 -ax map-pb -r2k dbg.raw.fa reads.fa.gz | samtools sort -@4 >dbg.bam
+samtools view -F0x900 dbg.bam | ./wtpoa-cns -t 16 -d dbg.raw.fa -i - -fo dbg.cns.fa
+
+# Addtional polishment using short reads
+bwa mem -t 16 dbg.cns.fa sr.1.fa sr.2.fa | samtools sort -O SAM | ./wtpoa-cns -t 16 -x sam-sr -d dbg.cns.fa -i - -fo dbg.srp.fa
+```
+
+## <a name="intro"></a>Introduction
+
+Wtdbg2 is a *de novo* sequence assembler for long noisy reads produced by
+PacBio or Oxford Nanopore Technologies (ONT). It assembles raw reads without
+error correction and then builds the consensus from intermediate assembly
+output. Wtdbg2 is able to assemble the human and even the 32Gb
+[Axolotl][Axolotl] genome at a speed tens of times faster than [CANU][canu] and
+[FALCON][falcon] while producing contigs of comparable base accuracy.
+
+During assembly, wtdbg2 chops reads into 1024bp segments, merges similar
+segments into a vertex and connects vertices based on the segment adjacency on
+reads. The resulting graph is called fuzzy Bruijn graph (FBG). It is akin to De
+Bruijn graph but permits mismatches/gaps and keeps read paths when collapsing
+k-mers. The use of FBG distinguishes wtdbg2 from the majority of long-read
+assemblers.
+
+## <a name="install"></a>Installation
+
+Wtdbg2 only works on 64-bit Linux. To compile, please type `make` in the source
+code directory. You can then copy `wtdbg2` and `wtpoa-cns` to your `PATH`.
+
+Wtdbg2 also comes with an approxmimate read mapper `kbm`, a faster but less
+accurate consesus tool `wtdbg-cns` and many auxiliary scripts in the `scripts`
+directory.
+
+## <a name="use"></a>Usage
+
+Wtdbg2 has two key components: an assembler **wtdbg2** and a consenser
+**wtpoa-cns**. Executable **wtdbg2** assembles raw reads and generates the
+contig layout and edge sequences in a file "*prefix*.ctg.lay.gz". Executable
+**wtpoa-cns** takes this file as input and produces the final consensus in
+FASTA. A typical workflow looks like this:
+```sh
+./wtdbg2 -x rs -g 4.6m -t 16 -i reads.fa.gz -fo prefix
+./wtpoa-cns -t 16 -i prefix.ctg.lay.gz -fo prefix.ctg.fa
+```
+where `-g` is the estimated genome size and `-x` specifies the sequencing
+technology, which could take value "rs" for PacBio RSII, "sq" for PacBio
+Sequel, "ccs" for PacBio CCS reads and "ont" for Oxford Nanopore. This option
+sets multiple parameters and should be **applied before other parameters**.
+When you are unable to get a good assembly, you may need to tune other
+parameters as follows.
+
+Wtdbg2 combines normal k-mers and homopolymer-compressed (HPC) k-mers to find
+read overlaps. Option `-k` specifies the length of normal k-mers, while `-p`
+specifies the length of HPC k-mers. By default, wtdbg2 samples a fourth of all
+k-mers by their hashcodes. For data of relatively low coverage, you may
+increase this sampling rate by reducing `-S`. This will greatly increase the
+peak memory as a cost, though. Option `-e`, which defaults to 3, specifies the
+minimum read coverage of an edge in the assembly graph. You may adjust this
+option according to the overall sequencing depth, too. Option `-A` also helps
+relatively low coverage data at the cost of performance. For PacBio data,
+`-L5000` often leads to better assemblies emperically, so is recommended.
+Please run `wtdbg2 --help` for a complete list of available options or consult
+[README-ori.md](README-ori.md) for more help.
+
+The following table shows various command lines and their resource usage for
+the assembly step:
+
+|Dataset                 |GSize |Cov     |Asm options        |CPU asm |CPU cns |Real tot|     RAM|
+|:-----------------------|-----:|-------:|:------------------|-------:|-------:|-------:|-------:|
+|[E. coli][pbcr]         |4.6Mb |PB x20  |-x rs -g4.6m -t16  |     53s|   8m54s|     42s|    1.0G|
+|[C. elegans][ce]        |100Mb |PB x80  |-x rs -g100m -t32  |   1h07m|   5h06m|  13m42s|   11.6G|
+|[D. melanogaster A4][dm2]| 144m|PB x120 |-x rs -g144m -t32  |   2h06m|   5h11m|  26m17s|   19.4G|
+|[D. melanogaster ISO1][dm1]|144m|ONT x32|-xont -g144m -t32  |   5h12m|   4h30m|  25m59s|   17.3G|
+|[A. thaliana][at]       |125Mb |PB x75  |-x sq -g125m -t32  |  11h26m|   4h57m|  49m35s|   25.7G|
+|[Human NA12878][na12878]|3Gb   |ONT x36 |-x ont -g3g -t31   | 793h11m|  97h46m|  31h03m|  221.8G|
+|[Human NA19240][na19240]|3Gb   |ONT x35 |-x ont -g3g -t31   | 935h31m|  89h17m|  35h20m|  215.0G|
+|[Human HG00733][hg00733]|3Gb   |PB x93  |-x sq -g3g -t47    |2114h26m| 152h24m|  52h22m|  338.1G|
+|[Human NA24385][na24385]|3Gb   |CCS x28 |-x ccs -g3g -t31   | 231h25m|  58h48m|  10h14m|  112.9G|
+|[Human CHM1][chm1]      |3Gb   |PB x60  |-x rs -g3g -t96    | 105h33m| 139h24m|   5h17m|  225.1G|
+|[Axolotl][axosra]       |32Gb  |PB x32  |-x rs -g32g -t96   |2806h40m|1456h13m| 110h16m| 1788.1G|
+
+The timing was obtained on three local servers with different hardware
+configurations. There are also run-to-run fluctuations. Exact timing on your
+machines may differ. The assembled contigs can be found at the following FTP:
+```txt
+ftp://ftp.dfci.harvard.edu/pub/hli/wtdbg/
+```
+
+## Limitations
+
+* For Nanopore data, wtdbg2 may produce an assembly smaller than the true
+  genome.
+
+* When inputing multiple files of both fasta and fastq format, please put fastq first, then fasta.
+  Otherwise, program cannot find '>' in fastq, and append all fastq in one read.
+
+## Citing wtdbg2
+
+If you use wtdbg2, please cite:
+
+> Ruan, J. and Li, H. (2019) Fast and accurate long-read assembly with wtdbg2. *bioRxiv*. doi:10.1101/530972
+
+
+## Getting Help
+
+Please use the [GitHub's Issues page][issue] if you have questions. You may
+also directly contact Jue Ruan at ruanjue@gmail.com.
+
+[miniasm]: https://github.com/lh3/miniasm
+[canu]: https://github.com/marbl/canu
+[falcon]: https://github.com/PacificBiosciences/FALCON
+[Axolotl]: https://www.nature.com/articles/nature25458
+[chm1]: https://trace.ncbi.nlm.nih.gov/Traces/sra/?study=SRP044331
+[na12878]: https://github.com/nanopore-wgs-consortium/NA12878/blob/master/rel5.md
+[na19240]: https://www.ebi.ac.uk/ena/data/view/PRJEB26791
+[pbcr]: http://www.cbcb.umd.edu/software/PBcR/data/selfSampleData.tar.gz
+[ce]: https://github.com/PacificBiosciences/DevNet/wiki/C.-elegans-data-set
+[axosra]: https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA378970
+[issue]: https://github.com/ruanjue/wtdbg2/issues
+[at]: https://downloads.pacbcloud.com/public/SequelData/ArabidopsisDemoData/
+[dm1]: https://www.ebi.ac.uk/ena/data/view/SRR6702603
+[dm2]: https://www.ebi.ac.uk/ena/data/view/SRR5439404
+[hg00733]: https://www.ebi.ac.uk/ena/data/view/SRR7615963
+[na24385]: https://ftp-trace.ncbi.nlm.nih.gov/giab/ftp/data/AshkenazimTrio/HG002_NA24385_son/PacBio_CCS_15kb/

best_sam_hits4longreads.c

@@ -0,0 +1,217 @@
+
+#include "mem_share.h"
+#include "filereader.h"
+#include "list.h"
+#include "hashset.h"
+
+typedef struct {
+	u4i stroff, strlen;
+	u4i taglen, flag;
+	u4i qlen, qb, qe;
+	u4i refidx, reflen;
+} lr_hit_t;
+define_list(lrhitv, lr_hit_t);
+
+int select_best_hit(lrhitv *hits, String *lines, u4i minlen, float mincov, FILE *out){
+	lr_hit_t *h1, *h2;
+	u4i i, j, pass;
+	int x, y, ret;
+	sort_array(hits->buffer, hits->size, lr_hit_t, num_cmpgt(b.qe - b.qb, a.qe - a.qb));
+	ret = 0;
+	for(i=0;i<hits->size;i++){
+		h1 = ref_lrhitv(hits, i);
+		if(h1->qe - h1->qb < minlen) break;
+		if(h1->qe - h1->qb < UInt(mincov * h1->qlen)) break;
+		pass = 1;
+		for(j=0;j<i;j++){
+			h2 = ref_lrhitv(hits, j);
+			x = num_max(h1->qb, h2->qb);
+			y = num_min(h1->qe, h2->qe);
+			if(y - x >= (1 - mincov) * (h1->qe - h1->qb)){
+				pass = 0;
+				break;
+			}
+		}
+		if(pass){
+			fprintf(out, "%s\n", lines->string + h1->stroff);
+			ret ++;
+		}
+	}
+	clear_lrhitv(hits);
+	clear_string(lines);
+	return ret;
+}
+
+int usage(char *prog){
+	printf(
+	"Usage: %s [-h] [-v] [-B:retain secondary aligment] [-l min_map_len:100] [-f min_map_cov:0.70]\n"
+	, prog
+	);
+	return 1;
+}
+
+int main(int argc, char **argv){
+	FileReader *fr;
+	FILE *out;
+	cuhash *refs;
+	cplist *reftags;
+	u4v    *reflens;
+	String *lines;
+	lrhitv *hits;
+	lr_hit_t *hit, HIT;
+	char *str, *reftag;
+	u4i minlen, i, reflen;
+	float mincov;
+	int c, primary_hit, verbose;
+	u1i movs[256];
+	minlen = 100;
+	mincov = 0.70;
+	primary_hit = 1;
+	verbose = 0;
+	out = stdout;
+	while((c = getopt(argc, argv, "hvBl:f:")) != -1){
+		switch(c){
+			case 'l': minlen = atoi(optarg); break;
+			case 'f': mincov = atof(optarg); break;
+			case 'B': primary_hit = 0; break;
+			case 'v': verbose = 1; break;
+			default: return usage(argv[0]);
+		}
+	}
+	fr = open_filereader(NULL, 1);
+	refs = init_cuhash(13);
+	reftags = init_cplist(8);
+	reflens = init_u4v(8);
+	hits = init_lrhitv(4);
+	lines = init_string(1024);
+	memset(movs, 0, 256);
+	movs[(int)'M'] = 0b11;
+	movs[(int)'I'] = 0b10;
+	movs[(int)'D'] = 0b01;
+	movs[(int)'N'] = 0b01;
+	movs[(int)'S'] = 0b10;
+	movs[(int)'H'] = 0b10;
+	movs[(int)'P'] = 0b00;
+	movs[(int)'='] = 0b11;
+	movs[(int)'X'] = 0b11;
+	while((c = readline_filereader(fr))){
+		if(fr->line->string[0] == '@'){
+			fprintf(out, "%s\n", fr->line->string);
+			if(fr->line->string[1] == 'S' && fr->line->string[2] == 'Q'){
+				if((c = split_line_filereader(fr, '\t')) > 2){
+					reftag = NULL;
+					reflen = 0;
+					for(i=1;i<3;i++){
+						if(get_col_len(fr, i) <= 3){
+							continue;
+						}
+						str = get_col_str(fr, i);
+						if(str[0] == 'S' && str[1] == 'N' && str[2] == ':'){
+							reftag = strdup(str + 3);
+						} else if(str[0] == 'L' && str[1] == 'N' && str[2] == ':'){
+							reflen = atol(str + 3);
+						}
+					}
+					if(strlen(reftag) && reflen){
+						push_cplist(reftags, reftag);
+						push_u4v(reflens, reflen);
+						put_cuhash(refs, (cuhash_t){reftag, reftags->size - 1});
+					}
+				}
+			}
+		} else {
+			hit = &HIT;
+			str = index(fr->line->string, '\t');
+			if(str == NULL){
+				fprintf(stderr, "[WARNNING:too_few_column] %s\n", fr->line->string);
+				continue;
+			}
+			hit->taglen = str - fr->line->string;
+			if(hits->size && (hits->buffer[0].taglen != hit->taglen || strncmp(lines->string + hits->buffer[0].stroff, fr->line->string, hit->taglen))){
+				select_best_hit(hits, lines, minlen, mincov, out);
+			}
+			hit->stroff = lines->size;
+			hit->strlen = fr->line->size;
+			append_string(lines, fr->line->string, fr->line->size);
+			add_char_string(lines, '\0');
+			if((c = split_line_filereader(fr, '\t')) < 11){
+				fprintf(stderr, "[WARNNING:too_few_columns] %s\n", lines->string + hit->stroff);
+				continue;
+			}
+			hit->taglen = get_col_len(fr, 0);
+			hit->flag = atol(get_col_str(fr, 1));
+			if(primary_hit && (hit->flag & 0x900)){
+				continue;
+			}
+			if(get_col_str(fr, 2)[0] == '*'){
+				continue;
+			}
+			hit->refidx = getval_cuhash(refs, get_col_str(fr, 2));
+			if(hit->refidx == MAX_U4){
+				fprintf(stderr, "[WARNNING:unknown_refname] %s\n", lines->string + hit->stroff);
+				continue;
+			}
+			hit->reflen = reflens->buffer[hit->refidx];
+			hit->qlen = 0;
+			u4i tb, te, qb, qe, len, cnt, tmp;
+			u4i ln;
+			char op;
+			tb = atoi(get_col_str(fr, 3));
+			te = tb;
+			qb = qe = 0;
+			len = 0;
+			tmp = cnt = 0;
+			str = get_col_str(fr, 5); // CIGAR
+			ln = 0;
+			op = 0;
+			while(str[0]){
+				if(str[0] >= '0' && str[0] <= '9'){
+					ln = ln * 10 + str[0] - '0';
+				} else {
+					op = movs[(int)str[0]];
+					if(op & 0b01){
+						te += ln;
+						qe += tmp;
+						tmp = 0;
+						if(cnt == 0){
+							qb = qe;
+							cnt = 1;
+						}
+						if(op & 0b10){
+							qe += ln;
+							len += ln;
+						}
+					} else if(op & 0b10){
+						tmp += ln;
+						len += ln;
+					}
+					ln = 0;
+				}
+				str ++;
+			}
+			if(hit->flag & 0x10){
+				tmp = len - qb;
+				qb = len - qe;
+				qe = tmp;
+			}
+			hit->qlen = len;
+			hit->qb = qb;
+			hit->qe = qe;
+			if(verbose){
+				fprintf(out, "#%s\t%u\t+\t%u\t%u\t%s\t%c\t%u\t%u\n", get_col_str(fr, 0), len, qb, qe, get_col_str(fr, 2), "+-"[(hit->flag & 0x10) >> 4], tb, te);
+			}
+			push_lrhitv(hits, HIT);
+		}
+	}
+	select_best_hit(hits, lines, minlen, mincov, out);
+	free_string(lines);
+	free_lrhitv(hits);
+	free_cuhash(refs);
+	for(i=0;i<reftags->size;i++){
+		free(reftags->buffer[i]);
+	}
+	free_cplist(reftags);
+	free_u4v(reflens);
+	close_filereader(fr);
+	return 0;
+}

bit2vec.h

@@ -0,0 +1,97 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __BIT2_VEC_RJ_H
+#define __BIT2_VEC_RJ_H
+
+#include <stdio.h>
+#include <stdint.h>
+#include <string.h>
+#include <stdlib.h>
+#include "mem_share.h"
+
+typedef struct {
+	uint8_t *bits;
+	uint64_t size;
+	uint64_t cap;
+} Bit2Vec;
+
+static inline Bit2Vec* init_bit2vec(uint64_t size){
+	Bit2Vec *vec;
+	vec = malloc(sizeof(Bit2Vec));
+	vec->size  = 0;
+	vec->cap   = size;
+	vec->bits  = calloc(1, (size * 2 + 7) / 8);
+	return vec;
+}
+
+static inline size_t bit2vec_obj_desc_cnt(void *obj, int idx){
+	return (((Bit2Vec*)obj)->cap * 2 + 7) / 8;
+	idx = idx;
+}
+
+static const obj_desc_t bit2vec_obj_desc = {"bit2vec_obj_desc", sizeof(Bit2Vec), 1, {1}, {offsetof(Bit2Vec, bits)}, {(obj_desc_t*)&OBJ_DESC_DATA}, bit2vec_obj_desc_cnt, NULL};
+
+static inline void clear_bit2vec(Bit2Vec *vec){
+	memset(vec->bits, 0, (vec->cap * 2 + 7) / 8);
+	vec->size = 0;
+}
+
+static inline void free_bit2vec(Bit2Vec *vec){
+	free(vec->bits);
+	free(vec);
+}
+
+static inline int encap_bit2vec(Bit2Vec *vec, uint32_t n){
+	uint64_t cap;
+	if(vec->size + n <= vec->cap) return 0;
+	cap = vec->cap;
+	while(vec->size + n > vec->cap){
+		if(vec->cap < 1024 * 1024){
+			vec->cap <<= 1;
+		} else {
+			vec->cap += 1024 * 1024;
+		}
+	}
+	vec->bits = realloc(vec->bits, (vec->cap * 2 + 7) / 8);
+	memset(vec->bits + (cap * 2 + 7) / 8, 0, (vec->cap * 2 + 7) / 8 - (cap * 2 + 7) / 8);
+	return 1;
+}
+
+static inline void set_bit2vec(Bit2Vec *vec, uint64_t idx, uint8_t dat){
+	vec->bits[idx >> 2] = (vec->bits[idx >> 2] & (~(3U << ((idx & 0x03U) << 1)))) | ((dat & 0x03) << ((idx & 0x03U) << 1));
+}
+
+static inline void push_bit2vec(Bit2Vec *vec, uint8_t dat){
+	encap_bit2vec(vec, 1);
+	set_bit2vec(vec, vec->size, dat);
+	vec->size ++;
+}
+
+static inline uint8_t get_bit2vec(Bit2Vec *vec, uint64_t idx){
+	return (vec->bits[idx >> 2] >> ((idx & 0x03U) << 1)) & 0x03;
+}
+
+static inline int pop_bit2vec(Bit2Vec *vec){
+	if(vec->size == 0) return -1;
+	vec->size --;
+	return get_bit2vec(vec, vec->size);
+}
+
+#endif

bitsvec.h

@@ -0,0 +1,200 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __BITS_VEC_RJ_H
+#define __BITS_VEC_RJ_H
+
+#include <stdio.h>
+#include <stdint.h>
+#include <string.h>
+#include <stdlib.h>
+#include "mem_share.h"
+
+/* Useful functions when n_bit > 8 */
+
+static inline void u8byte2bits(uint64_t val, uint8_t *dat, uint64_t offset, uint8_t size){
+	uint8_t i, *v;
+	v = (uint8_t*)&val;
+#if __BYTE_ORDER == 1234
+	for(i=0;i<size;i++){
+		if(v[i >> 3] & (1U << (i & 0x7U))) dat[offset >> 3] |= 1U << (offset & 0x7U);
+		else dat[offset >> 3] &= ~(1U << (offset & 0x7U));
+		offset ++;
+	}
+#else
+	for(i=0;i<size;i++){
+		if(v[7 - (i >> 3)] & (1U << (i & 0x7U))) dat[offset >> 3] |= 1U << (offset & 0x7U);
+		else dat[offset >> 3] &= ~(1U << (offset & 0x7U));
+		offset ++;
+	}
+#endif
+}
+
+static inline uint64_t bits2u8byte(uint8_t *dat, uint64_t offset, uint8_t size){
+	uint64_t ret;
+	uint8_t i, *v;
+	ret = 0;
+	v = (uint8_t*)&ret;
+#if __BYTE_ORDER == 1234
+	for(i=0;i<size;i++){
+		if(dat[offset >> 3] & (1U << (offset & 0x7U))) v[i >> 3] |= 1U << (i & 0x7U);
+		offset ++;
+	}
+#else
+	for(i=0;i<size;i++){
+		if(dat[offset >> 3] & (1U << (offset & 0x7U))) v[7 - (i >> 3)] |= 1U << (i & 0x7U);
+		offset ++;
+	}
+#endif
+	return ret;
+}
+
+typedef struct {
+	uint8_t *bits;
+	uint64_t size;
+	uint64_t cap;
+	uint8_t n_bit;
+	uint32_t mask;
+} BitsVec;
+
+static inline BitsVec* init_bitsvec(uint64_t size, uint32_t n_bit){
+	BitsVec *vec;
+	if(n_bit == 0) n_bit = 1;
+	else if(n_bit > 8) n_bit = 8;
+	if(size < 8) size = 8;
+	vec = calloc(1, sizeof(BitsVec));
+	vec->n_bit = n_bit;
+	vec->mask  = (1U << n_bit) - 1U;
+	vec->size  = 0;
+	vec->cap   = size;
+	vec->bits  = calloc((size * vec->n_bit + 15) / 8, 1);
+	return vec;
+}
+
+static inline size_t bitsvec_obj_desc_cnt(void *obj, int idx){
+	return (((BitsVec*)obj)->cap * ((BitsVec*)obj)->n_bit + 15) / 8;
+	idx = idx;
+}
+
+static const obj_desc_t bitsvec_obj_desc = {"bitsvec_obj_desc", sizeof(BitsVec), 1, {1}, {offsetof(BitsVec, bits)}, {(obj_desc_t*)&OBJ_DESC_DATA}, bitsvec_obj_desc_cnt, NULL};
+
+static inline void clear_bitsvec(BitsVec *vec){
+	vec->size = 0;
+}
+
+static inline void free_bitsvec(BitsVec *vec){
+	free(vec->bits);
+	free(vec);
+}
+
+static inline int encap_bitsvec(BitsVec *vec, u8i n){
+	if(vec->size + n <= vec->cap) return 0;
+	if(vec->size + n < 0x3FFFFFFFLLU){
+		vec->cap = roundup_power2(vec->size + n);
+	} else {
+		vec->cap = (vec->size + n + 0x3FFFFFFFLLU) & (MAX_U8 << 30);
+	}
+	vec->bits = realloc(vec->bits, (vec->cap * vec->n_bit + 15) / 8);
+	return 1;
+}
+
+static inline void set_bitsvec(BitsVec *vec, u8i idx, u1i dat){
+	register u8i off;
+	register u2i x, d;
+	off = (idx * vec->n_bit);
+	d = off & 0x07;
+	off >>= 3;
+	x = (((u2i)vec->bits[off + 1]) << 8) | vec->bits[off + 0];
+	x = (x & (~(vec->mask << d))) | ((UInt(dat) & vec->mask) << d);
+	vec->bits[off] = x;
+	vec->bits[off + 1] = x >> 8;
+}
+
+static inline void push_bitsvec(BitsVec *vec, u1i dat){
+	encap_bitsvec(vec, 1);
+	set_bitsvec(vec, vec->size, dat);
+	vec->size ++;
+}
+
+// TODO: need to be optimized
+static inline void pushs_bitsvec(BitsVec *vec, u1i dat, u8i len){
+	u8i i;
+	encap_bitsvec(vec, len);
+	for(i=0;i<len;i++){
+		set_bitsvec(vec, vec->size + i, dat);
+	}
+	vec->size += len;
+}
+
+static inline u2i gets_bitsvec(BitsVec *vec, u8i idx){
+	u8i off;
+	off = (idx * vec->n_bit);
+	return ((((u2i)vec->bits[(off >> 3) + 1]) << 8) | vec->bits[(off >> 3) + 0]) >> (off & 0x07);
+}
+
+static inline uint8_t get_bitsvec(BitsVec *vec, u8i idx){
+	return gets_bitsvec(vec, idx) & vec->mask;
+}
+
+static inline void append_bitsvec(BitsVec *dst, BitsVec *src, u8i off, u8i len){
+	u8i i, di, si, se;
+	u2i x;
+	u1i p, n, sd;
+	if(0){ // Assume dst->n_bit == src->n_bit
+		if(dst->n_bit != src->n_bit){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			abort();
+		}
+	}
+	encap_bitsvec(dst, len);
+	p = (dst->n_bit & 0x01)? 8 : (8 / dst->n_bit);
+	n = dst->size % p;
+	if(n) n = p - n;
+	if(len <= n){
+		for(i=0;i<len;i++){
+			set_bitsvec(dst, dst->size + i, get_bitsvec(src, off + i));
+		}
+		dst->size += len;
+		return;
+	} else {
+		for(i=0;i<n;i++){
+			set_bitsvec(dst, dst->size + i, get_bitsvec(src, off + i));
+		}
+		dst->size += n;
+	}
+	di = (dst->size * dst->n_bit) >> 3;
+	si = ((off + i) * src->n_bit);
+	sd = si & 0x07;
+	si >>= 3;
+	se = ((off + len) * src->n_bit + 7) >> 3;
+	while(si < se){
+		x = ((src->bits[si + 1] << 8) | src->bits[si]) >> sd;
+		dst->bits[di++] = x;
+		si ++;
+	}
+	dst->size += len - i;
+}
+
+static inline int pop_bitsvec(BitsVec *vec){
+	if(vec->size == 0) return -1;
+	vec->size --;
+	return get_bitsvec(vec, vec->size);
+}
+
+#endif

bitvec.h

@@ -0,0 +1,530 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __BIT_VEC_RJ_H
+#define __BIT_VEC_RJ_H
+
+#include <stdio.h>
+#include <stdint.h>
+#include <string.h>
+#include <stdlib.h>
+#include "mem_share.h"
+
+static const u1i byte_ones_table[256] = {
+	0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4,
+	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
+	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
+	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
+	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
+	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
+	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
+	4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8
+};
+
+static inline unsigned int _bitvec_roundup_power2(unsigned int v){
+	if(v == 0) return 0;
+	v--;
+	v |= v >> 1;
+	v |= v >> 2;
+	v |= v >> 4;
+	v |= v >> 8;
+	v |= v >> 16;
+	return v + 1;
+}
+
+typedef struct {
+	u8i *bits;
+	u8i n_bit;
+	u8i n_cap;
+
+	u8i *sums;
+	u8i sum_size;
+	u8i n_ones;
+	u8i *hash;
+	u8i hash_size;
+	u8i hash_mod;
+	int64_t iter_idx;
+} BitVec;
+
+#if 0
+
+static inline u4i count_ones_bit32(u4i v){
+	v = v - ((v >> 1) & 0x55555555U);                        // reuse input as temporary
+	v = (v & 0x33333333U) + ((v >> 2) & 0x33333333U);        // temp
+	return (((v + (v >> 4)) & 0xF0F0F0FU) * 0x1010101U) >> 24; // count
+}
+
+#define ONES_STEP_4 0x1111111111111111ULL
+#define ONES_STEP_8 0x0101010101010101ULL
+
+static inline int count_ones_bit64(const u8i x){
+	register u8i byte_sums = x - ((x & 0xa * ONES_STEP_4) >> 1);
+	byte_sums = (byte_sums & 3 * ONES_STEP_4) + ((byte_sums >> 2) & 3 * ONES_STEP_4);
+	byte_sums = (byte_sums + (byte_sums >> 4)) & 0x0f * ONES_STEP_8;
+	return byte_sums * ONES_STEP_8 >> 56;
+}
+
+#else
+
+#define count_ones_bit32(v) __builtin_popcount(v)
+#define count_ones_bit64(v) __builtin_popcountll(v)
+
+#endif
+
+#define reverse_u1i(v) (((((u1i)v) * 0x0202020202ULL) & 0x010884422010ULL) % 1023)
+
+static inline size_t bitvec_obj_desc_cnt(void *bitv, int idx){
+	switch(idx){
+		case 0: return ((BitVec*)bitv)->n_cap / 64 * 8;
+		case 1: return ((BitVec*)bitv)->sums? (((BitVec*)bitv)->sum_size * 2 + 1) * 8 : 0;
+		case 2: return ((BitVec*)bitv)->hash? (((BitVec*)bitv)->hash_size) * 8 : 0;
+		default: return 0;
+	}
+}
+
+static const obj_desc_t bitvec_obj_desc = {"bitvec_obj_desc", sizeof(BitVec), 3, {1, 1, 1}, {offsetof(BitVec, bits), offsetof(BitVec, sums), offsetof(BitVec, hash)}, {(obj_desc_t*)&OBJ_DESC_DATA, (obj_desc_t*)&OBJ_DESC_DATA, (obj_desc_t*)&OBJ_DESC_DATA}, bitvec_obj_desc_cnt, NULL};
+
+static inline BitVec* init_bitvec(u8i n_bit){
+	BitVec *bitv;
+	if(n_bit == 0) n_bit = 64 * 8;
+	bitv = (BitVec*)malloc(sizeof(BitVec));
+	bitv->n_bit = 0;
+	bitv->n_cap = (((n_bit + 63) / 64) + 7) / 8 * 64 * 8;
+	bitv->bits  = (u8i*)calloc((bitv->n_cap / 64) + 1, 8);
+	bitv->bits[bitv->n_cap / 64] = 0x0000000000000001LLU;
+	//memset(bitv->bits, 0, bitv->n_cap / 8);
+	bitv->sums = NULL;
+	bitv->hash = NULL;
+	bitv->sum_size = 0;
+	bitv->n_ones = 0;
+	bitv->hash_size = 0;
+	bitv->hash_mod = 0;
+	bitv->iter_idx = 0;
+	return bitv;
+}
+
+static inline size_t dump_bitvec(BitVec *bitv, FILE *out){
+	fwrite(&bitv->n_bit, sizeof(u8i), 1, out);
+	fwrite(&bitv->n_cap, sizeof(u8i), 1, out);
+	fwrite(bitv->bits, sizeof(u8i), bitv->n_cap / 64, out);
+	return sizeof(u8i) * (2 + bitv->n_cap / 64);
+}
+
+static inline BitVec* load_bitvec(FILE *inp){
+	BitVec *bitv;
+	size_t n;
+	bitv = (BitVec*)malloc(sizeof(BitVec));
+	if((n = fread(&bitv->n_bit, sizeof(u8i), 1, inp)) != 1){
+		free(bitv); return NULL;
+	}
+	if((n = fread(&bitv->n_cap, sizeof(u8i), 1, inp)) != 1){
+		free(bitv); return NULL;
+	}
+	bitv->bits = (u8i*)malloc(bitv->n_cap / 8);
+	if(bitv->bits == NULL){
+		fprintf(stderr, " Out of memeory in load_bitvec\n "); fflush(stderr); exit(1);
+	}
+	if((n = fread(bitv->bits, sizeof(u8i), bitv->n_cap / 64, inp)) != bitv->n_cap / 64){
+		free(bitv); free(bitv->bits); return NULL;
+	}
+	bitv->sums = NULL;
+	bitv->hash = NULL;
+	bitv->hash_size = 0;
+	return bitv;
+}
+
+#if 0
+static inline BitVec* mem_load_bitvec(void *mem, FILE *inp){
+	BitVec *bitv;
+	size_t off, n;
+	bitv = mem;
+	off = ((sizeof(BitVec) + 7) / 8) * 8;
+	if((n = fread(&bitv->n_bit, sizeof(u8i), 1, inp)) != 1) return NULL;
+	if((n = fread(&bitv->n_cap, sizeof(u8i), 1, inp)) != 1) return NULL;
+	bitv->sums = NULL;
+	bitv->hash = NULL;
+	bitv->hash_size = 0;
+	bitv->bits = mem + off;
+	off += (bitv->n_cap / 64) * 8;
+	if((n = fread(bitv->bits, sizeof(u8i), bitv->n_cap / 64, inp)) != bitv->n_cap / 64) return NULL;
+	return bitv;
+}
+#endif
+
+static inline void clear_bitvec(BitVec *bitv){ bitv->n_bit = 0; }
+
+static inline void zeros_bitvec(BitVec *bitv){ memset(bitv->bits, 0, bitv->n_cap / 8); }
+
+// exclusive end
+static inline void reg_zeros_bitvec(BitVec *bitv, u8i beg, u8i end){
+	u8i b, e;
+	if(beg >= end) return;
+	b = beg >> 6;
+	e = end >> 6;
+	if(b == e){
+		bitv->bits[b] &= (MAX_U8 << (beg & 0x3FU)) ^ (MAX_U8 >> (64 - (end & 0x3FU)));
+	} else {
+		bitv->bits[b] &= ~(MAX_U8 << (beg & 0x3FU));
+		while(++b < e){ bitv->bits[b] = 0; }
+		bitv->bits[b] &= MAX_U8 << (end & 0x3FU);
+	}
+}
+
+static inline void ones_bitvec(BitVec *bitv){ memset(bitv->bits, 0xFFU, bitv->n_cap / 8); }
+
+// exclusive end
+static inline void reg_ones_bitvec(BitVec *bitv, u8i beg, u8i end){
+	u8i b, e;
+	if(beg >= end) return;
+	b = beg >> 6;
+	e = end >> 6;
+	if(b == e){
+		bitv->bits[b] |= (MAX_U8 << (beg & 0x3FU)) & (MAX_U8 >> (64 - (end & 0x3FU)));
+	} else {
+		bitv->bits[b] |= MAX_U8 << (beg & 0x3FU);
+		while(++b < e){ bitv->bits[b] = MAX_U8; }
+		bitv->bits[b] |= ~(MAX_U8 << (end & 0x3FU));
+	}
+}
+
+static inline void flip_bitvec(BitVec *bitv, u8i idx){ bitv->bits[idx>>6] ^= 1LLU << (idx&0x3FU); }
+
+static inline void one_bitvec(BitVec *bitv, u8i idx){ bitv->bits[idx>>6] |= 1LLU << (idx&0x3FU); }
+
+static inline void zero_bitvec(BitVec *bitv, u8i idx){ bitv->bits[idx>>6] &= ~(1LLU << (idx&0x3FU)); }
+
+static inline void set_bitvec(BitVec *bitv, u8i idx, int v){
+	if(v){
+		one_bitvec(bitv, idx);
+	} else {
+		zero_bitvec(bitv, idx);
+	}
+}
+
+static inline u8i get_bitvec(BitVec *bitv, u8i idx){ return (bitv->bits[idx>>6] >> (idx&0x3FU)) & 0x01LLU; }
+
+static inline u8i get64_bitvec(BitVec *bitv, u8i off){
+	u8i m, n;
+	m = off >> 6;
+	n = off & 0x3F;
+	if(n){
+		return (bitv->bits[m] >> (64 - n)) | (bitv->bits[m + 1] << n);
+	} else {
+		return bitv->bits[m];
+	}
+}
+
+static inline void set64_bitvec(BitVec *bitv, u8i off, u8i val){
+	u8i m, n;
+	m = off >> 6;
+	n = off & 0x3F;
+	if(n){
+		bitv->bits[m] = ((bitv->bits[m] << (64 - n)) >> (64 - n)) | (val << (64 - n));
+		m ++;
+		bitv->bits[m] = ((bitv->bits[m] >> n) << n) | (val >> (64 - n));
+	} else {
+		bitv->bits[m] = val;
+	}
+}
+
+static inline void encap_bitvec(BitVec *bitv, u8i num){
+	u8i cap;
+	if(bitv->n_bit + num < bitv->n_cap) return;
+	cap = bitv->n_cap;
+	while(bitv->n_bit + num >= bitv->n_cap){
+		if(bitv->n_cap < 1024 * 1024 * 8){
+			bitv->n_cap <<= 1;
+		} else bitv->n_cap += 1024 * 1024 * 8;
+	}
+	bitv->bits = (u8i*)realloc(bitv->bits, bitv->n_cap / 8 + 8);
+	memset(((void*)bitv->bits) + cap / 8, 0, (bitv->n_cap - cap) / 8 + 8);
+	bitv->bits[cap / 64] = 0x0000000000000001LLU;
+}
+
+static inline void recap_bitvec(BitVec *bitv, u8i new_cap){
+	if(new_cap & 0x3FU) new_cap = (new_cap & 0xFFFFFFFFFFFFFFC0LLU) + 0x40U;
+	if(bitv->n_cap == new_cap) return;
+	bitv->bits = (u8i*)realloc(bitv->bits, new_cap / 8 + 8);
+	if(new_cap > bitv->n_cap){
+		memset(((void*)bitv->bits) + bitv->n_cap / 8, 0, (new_cap - bitv->n_cap) / 8 + 8);
+	}
+	bitv->bits[new_cap / 64] = 0x0000000000000001LLU;
+	bitv->n_cap = new_cap;
+}
+
+static inline void one2bitvec(BitVec *bitv){ encap_bitvec(bitv, 1); one_bitvec(bitv, bitv->n_bit); bitv->n_bit ++; }
+
+static inline void zero2bitvec(BitVec *bitv){ encap_bitvec(bitv, 1); zero_bitvec(bitv, bitv->n_bit); bitv->n_bit ++; }
+
+static inline u8i get_2bitvec(BitVec *bitv, u8i idx){ return (bitv->bits[idx>>5] >> ((idx&0x1FU) << 1)) & 0x03LLU; }
+
+static inline void set_2bitvec(BitVec *bitv, u8i idx, u8i v){
+	bitv->bits[idx>>5] = (bitv->bits[idx>>5] & (~(0x03LLU << ((idx&0x1FU) << 1)))) | ((v&0x03LLU) << ((idx&0x1FU) << 1));
+}
+
+static inline void push_2bitvec(BitVec *bitv, u8i v){
+	encap_bitvec(bitv, 2);
+	set_2bitvec(bitv, bitv->n_bit >> 1, v);
+	bitv->n_bit = ((bitv->n_bit >> 1) + 1) << 1;
+}
+
+static inline void end_bitvec(BitVec *bitv){ encap_bitvec(bitv, 1); one_bitvec(bitv, bitv->n_bit); }
+
+static inline u8i next_one_bitvec(BitVec *bitv, u8i idx){
+	register u8i p, v;
+	register u4i s;
+	p = idx >> 6;
+	s = idx & 0x3F;
+	while(!(bitv->bits[p] >> s)){ p ++; s = 0; }
+	v = bitv->bits[p] >> s;
+	s += __builtin_ctzll(v);
+	return (p << 6) + s;
+}
+
+static inline u8i reg_count_bitvec(BitVec *bitv, u8i beg, u8i end){
+	u8i cnt, b, e, t;
+	if(beg >= end) return 0;
+	b = beg >> 6;
+	e = end >> 6;
+	if(b == e){
+		t = (bitv->bits[b] & (MAX_U8 >> (64 - (end & 0x3F)))) >> (beg & 0x3F);
+		cnt = count_ones_bit64(t);
+	} else {
+		cnt = count_ones_bit64(bitv->bits[b] >> (beg & 0x3F));
+		while(++b < e){
+			cnt += count_ones_bit64(bitv->bits[b]);
+		}
+		if(end & 0x3F){
+			cnt += count_ones_bit64(bitv->bits[b] & (MAX_U8 >> (64 - (end & 0x3F))));
+		}
+	}
+	return cnt;
+}
+
+static const int Mod37BitPosition[] = // map a bit value mod 37 to its position
+{
+ 32,  0,  1, 26,  2, 23, 27,  0,  3, 16,
+ 24, 30, 28, 11,  0, 13,  4,  7, 17,  0,
+ 25, 22, 31, 15, 29, 10, 12,  6,  0, 21,
+ 14,  9,  5, 20,  8, 19, 18
+};
+
+static inline u8i next_one_bitvec2(BitVec *bitv, u8i idx){
+	register u8i p;
+	register u4i s, v;
+	p = idx >> 6;
+	s = idx & 0x3F;
+	while(!(bitv->bits[p] >> s)){ p ++; s = 0; }
+	if(!((bitv->bits[p] >> s) & 0xFFFFFFFFU)) s += 32;
+	v = bitv->bits[p] >> s;
+	s += Mod37BitPosition[(-v & v) % 37];
+	return (p << 6) + s;
+}
+
+static inline u8i next_one_bitvec3(BitVec *bitv, u8i idx){
+	register u8i p;
+	register u4i s;
+	p = idx >> 6;
+	s = idx & 0x3F;
+	while(!(bitv->bits[p] >> s)){ p ++; s = 0; }
+	while(!((bitv->bits[p] >> s) & 0xFFU)) s += 8;
+	while(!((bitv->bits[p] >> s) & 0x01U)) s ++;
+	return (p << 6) + s;
+}
+
+//n_cap MUST be times of 64 * 8
+static inline void index_bitvec_core(BitVec *bitv, size_t n_cap){
+	u8i i, k, s, t, m;
+	m = ((n_cap + 63) / 64 + 7) / 8;
+	if(bitv->sums) free(bitv->sums);
+	bitv->sums = (u8i*)calloc((m * 2 + 1), 8);
+	t = 0;
+	for(i=0;i<n_cap;i+=64*8){
+		k = ((i>>6) >> 3) << 1;
+		bitv->sums[k] = t;
+		s = 0;
+		s += count_ones_bit64(bitv->bits[(i>>6)+0]);
+		bitv->sums[k+1] |= s << 0;
+		s += count_ones_bit64(bitv->bits[(i>>6)+1]);
+		bitv->sums[k+1] |= s << 9;
+		s += count_ones_bit64(bitv->bits[(i>>6)+2]);
+		bitv->sums[k+1] |= s << 18;
+		s += count_ones_bit64(bitv->bits[(i>>6)+3]);
+		bitv->sums[k+1] |= s << 27;
+		s += count_ones_bit64(bitv->bits[(i>>6)+4]);
+		bitv->sums[k+1] |= s << 36;
+		s += count_ones_bit64(bitv->bits[(i>>6)+5]);
+		bitv->sums[k+1] |= s << 45;
+		s += count_ones_bit64(bitv->bits[(i>>6)+6]);
+		bitv->sums[k+1] |= s << 54;
+		s += count_ones_bit64(bitv->bits[(i>>6)+7]);
+		t += s;
+	}
+	bitv->sums[((i>>6) >> 3) << 1] = t;
+	bitv->n_ones = t;
+	bitv->sum_size = m;
+	bitv->hash_size = (n_cap / 64 / 8) / 2;
+	if(bitv->hash_size == 0) bitv->hash_size = 1;
+	bitv->hash_mod = (t + bitv->hash_size) / bitv->hash_size;
+	if(bitv->hash_mod == 0) bitv->hash_mod = 1;
+	if(bitv->hash) free(bitv->hash);
+	bitv->hash = (u8i*)malloc(sizeof(u8i) * bitv->hash_size);
+	s = 0;
+	t = 0;
+	for(i=0;i<=m;i++){
+		k = bitv->sums[i*2] / bitv->hash_mod;
+		if(s < k){
+			while(s < k){ bitv->hash[s] = t; s ++; }
+			t = i? i - 1 : 0;
+		}
+	}
+	bitv->hash[bitv->sums[m*2] / bitv->hash_mod] = t;
+}
+
+static inline void index_bitvec(BitVec *bitv){
+	index_bitvec_core(bitv, bitv->n_cap);
+}
+
+static inline u8i rank_bitvec(BitVec *bitv, u8i idx){
+	u8i p, s, sum;
+	p = (idx>>6)>>3;
+	s = (idx >> 6) & 0x07U;
+	sum = bitv->sums[p<<1];
+	if(s) sum += (bitv->sums[(p<<1)+1] >> (9 * (s - 1))) & 0x1FFU;
+	if(idx & 0x3FU) sum += count_ones_bit64(bitv->bits[idx>>6]<<(64-(idx&0x3FU)));
+	return sum;
+}
+
+static inline u1i select_8bytes(u8i word, u1i n_one){
+	u1i idx, n, m;
+	n = count_ones_bit32((u4i)word);
+	if(n >= n_one){
+		n = 0;
+		idx = 0;
+		word = word & 0xFFFFFFFFU;
+	} else {
+		idx = 32;
+		word = word >> 32;
+	}
+	while(1){
+		m = byte_ones_table[(u1i)word];
+		if(n + m >= n_one) break;
+		n += m;
+		idx += 8;
+		word >>= 8;
+	}
+	m = byte_ones_table[(u1i)(word & 0xF)];
+	if(n + m < n_one){
+		idx += 4;
+		word >>= 4;
+		n += m;
+	}
+	while(word){
+		idx ++;
+		if(word & 0x01){
+			n ++;
+			if(n == n_one) break;
+		}
+		word >>= 1;
+	}
+	return idx;
+}
+
+/*
+ * To select the 1'st one, use select_bitvec(bitv, 1) - 1
+ * */
+static inline u8i select_bitvec(BitVec *bitv, u8i idx){
+	u8i i, p, s, sum, t;
+	p = bitv->hash[idx / bitv->hash_mod];
+	while(p + 1 < bitv->sum_size && bitv->sums[(p + 1) << 1] < idx) p ++;
+	sum = bitv->sums[p << 1];
+	i = 0;
+	t = sum;
+	while(i < 7){
+		s = (bitv->sums[(p << 1) + 1] >> (9 * i)) & 0x1FFU;
+		if(sum + s >= idx) break;
+		t = sum + s;
+		i ++;
+	}
+	p = p * 8 + i;
+	s = idx - t;
+	return p * 64 + select_8bytes(bitv->bits[p], s);
+}
+
+static inline void begin_iter_bitvec(BitVec *bitv){ bitv->iter_idx = -1; }
+
+static inline u8i iter_bitvec(BitVec *bitv){
+	if((u8i)(bitv->iter_idx + 1) > bitv->n_cap) return 0xFFFFFFFFFFFFFFFFLLU;
+	bitv->iter_idx = next_one_bitvec(bitv, bitv->iter_idx + 1);
+	return (u8i)bitv->iter_idx;
+}
+
+static inline void free_bitvec(BitVec *bitv){
+	free(bitv->bits);
+	if(bitv->sums) free(bitv->sums);
+	if(bitv->hash) free(bitv->hash);
+	free(bitv);
+}
+
+#if 0
+
+static inline size_t mem_size_bitvec(BitVec *bitv){
+	size_t m;
+	m = (sizeof(BitVec) + 7) / 8 * 8 + ((bitv->n_cap / 64) * 8);
+	if(bitv->sums){
+		m += (bitv->sum_size * 2 + 1) * 8;
+	}
+	if(bitv->hash){
+		m += bitv->hash_size * 8;
+	}
+	return m;
+}
+
+static inline size_t mem_dump_bitvec(BitVec *bitv, void *mem){
+	BitVec *clone;
+	size_t off;
+	clone = mem;
+	memcpy(clone, bitv, sizeof(BitVec));
+	off = ((sizeof(BitVec) + 7) / 8) * 8;
+	clone->bits = mem + off;
+	memcpy(clone->bits, bitv->bits, (bitv->n_cap / 64) * 8);
+	off += (bitv->n_cap / 64) * 8;
+	if(bitv->sums){
+		clone->sums = mem + off;
+		memcpy(clone->sums, bitv->sums, (bitv->sum_size * 2 + 1) * 8);
+		off += (bitv->sum_size * 2 + 1) * 8;
+	}
+	if(bitv->hash){
+		clone->hash = mem + off;
+		memcpy(clone->hash, bitv->hash, bitv->hash_size * 8);
+		off += bitv->hash_size * 8;
+	}
+	return off;
+}
+#endif
+
+#endif

chararray.h

@@ -0,0 +1,354 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __STRING_RJ_H
+#define __STRING_RJ_H
+
+#include <stdio.h>
+#include <stdarg.h>
+#include <string.h>
+#include <stdlib.h>
+#include "list.h"
+#include "mem_share.h"
+
+/**
+ * String
+ */
+
+
+#ifdef HUGE_STRING
+typedef long int string_size_t;
+#else
+typedef int string_size_t;
+#endif
+
+typedef struct {
+	union { char *string; char *buffer; };
+	string_size_t size;
+	string_size_t capacity;
+} String;
+
+typedef struct {
+	char *string;
+	string_size_t  size;
+} VString;
+define_list(VStrv, VString);
+
+static inline String* init_string(string_size_t cap){
+	String *str;
+	str = (String*)malloc(sizeof(String));
+	str->size = 0;
+	str->capacity = (cap&0x1)? cap + 1 : cap + 2;
+	str->string = (char*)malloc(sizeof(char) * (str->capacity));
+	str->string[0] = 0;
+	return str;
+}
+
+static inline size_t string_obj_desc_cnt(void *obj, int idx){
+	return ((String*)obj)->size + 1;
+	idx = idx;
+}
+
+static const obj_desc_t string_obj_desc = {"string_obj_desc", sizeof(String), 1, {1}, {offsetof(String, string)}, {(obj_desc_t*)&OBJ_DESC_DATA}, string_obj_desc_cnt, NULL};
+
+static inline unsigned int _string_size_roundup_power2(unsigned int v){
+	v--;
+	v |= v >> 1;
+	v |= v >> 2;
+	v |= v >> 4;
+	v |= v >> 8;
+	v |= v >> 16;
+	return v + 1;
+}
+
+static inline void encap_string(String *str, string_size_t inc){
+	if(inc + str->size + 1 > str->capacity){
+		if(inc + str->size + 1 < 0xFFFFFFF){
+			str->capacity = _string_size_roundup_power2(inc + str->size + 1);
+		} else {
+			str->capacity += (((inc + str->size + 1) - str->capacity + 0xFFFFFFF - 1) / 0xFFFFFFF) * 0xFFFFFFF;
+		}
+		str->string = (char*)realloc(str->string, str->capacity);
+	}
+}
+
+static inline void recap_string(String *str, string_size_t cap){
+	if(cap <= str->size) return;
+	str->string = (char*)realloc(str->string, cap);
+	str->capacity = cap;
+}
+
+static inline void uc_string(String *str){
+	string_size_t i;
+	for(i=0;i<str->size;i++){
+		if(str->string[i] >= 'a' && str->string[i] <= 'z') str->string[i] = str->string[i] + 'A' - 'a';
+	}
+}
+
+static inline void lc_string(String *str){
+	string_size_t i;
+	for(i=0;i<str->size;i++){
+		if(str->string[i] >= 'A' && str->string[i] <= 'Z') str->string[i] = str->string[i] + 'a' - 'A';
+	}
+}
+
+static inline char* substr(char *string, string_size_t len, char *dst){
+	char *str;
+	if(len < 0) len = strlen(string);
+	if(dst != NULL) str = dst;
+	else str = (char*)malloc(sizeof(char) * (len + 1));
+	strncpy(str, string, len);
+	str[len] = '\0';
+	return str;
+}
+
+static inline char* catstr(string_size_t n_str, ...){
+	char *str, *s;
+	string_size_t i, len, inc;
+	va_list params;
+	
+	len = 0;
+	str = NULL;
+	va_start(params, n_str);
+	for(i=0;i<n_str;i++){
+		s = va_arg(params, char*);
+		if(s == NULL) continue;
+		inc = strlen(s);
+		str = realloc(str, len + inc + 1);
+		memcpy(str + len, s, inc + 1);
+		len += inc;
+	}
+	va_end(params);
+	return str;
+}
+
+static inline void chomp_string(String *str){
+	if(str->size && str->string[str->size - 1] == '\n'){
+		str->size --;
+		str->string[str->size] = 0;
+	}
+}
+
+static inline void chomp_vstring(VString *str){
+	if(str->size && str->string[str->size - 1] == '\n'){
+		str->size --;
+	}
+}
+
+static inline void trim_string(String *str){
+	string_size_t i, j;
+	i = str->size - 1;
+	while(i >= 0 && (str->string[i] == '\n' || str->string[i] == '\t' || str->string[i] == ' ')) i--; 
+	str->size = i + 1;
+	i = 0;
+	while(i < str->size && (str->string[i] == '\n' || str->string[i] == '\t' || str->string[i] == ' ')) i++;
+	if(i){
+		for(j=i;j<str->size;j++){ str->string[j-i] = str->string[j]; }
+		str->size -= i;
+	}
+	str->string[str->size] = 0;
+}
+
+static inline void trim_vstring(VString *str){
+	string_size_t i;
+	i = str->size - 1;
+	while(i >= 0 && (str->string[i] == '\n' || str->string[i] == '\t' || str->string[i] == ' ')) i--; 
+	str->size = i + 1;
+	i = 0;
+	while(i < str->size && (str->string[i] == '\n' || str->string[i] == '\t' || str->string[i] == ' ')) i++;
+	str->string += i;
+}
+
+static inline void append_string(String *str, char *src, string_size_t offlen){
+	encap_string(str, offlen);
+	memcpy(str->string + str->size, src, offlen);
+	str->size += offlen;
+	str->string[str->size] = 0;
+}
+
+static inline void append_char_string(String *str, char c, string_size_t num){
+	encap_string(str, num);
+	while(num-- > 0){ str->string[str->size ++] = c; }
+	str->string[str->size] = 0;
+}
+
+static inline String* as_string(char *chs, string_size_t len){
+	String *str;
+	str = init_string(len);
+	memcpy(str->string, chs, len);
+	str->size = len;
+	str->string[len] = 0;
+	return str;
+}
+
+static inline VString* as_vstring(char *chs){
+	string_size_t len;
+	VString *str;
+	len = strlen(chs);
+	str = malloc(sizeof(VString));
+	str->string = chs;
+	str->size = len;
+	return str;
+}
+
+static inline void add_char_string(String *str, char ch){
+	encap_string(str, 1);
+	str->string[str->size++] = ch;
+	str->string[str->size] = 0;
+}
+
+#define push_string(str, ch) add_char_string(str, ch)
+
+static inline void add_int_string(String *str, long long val){
+	string_size_t n;
+	long long v;
+	encap_string(str, 30);
+	if(val == 0){
+		str->string[str->size++] = '0';
+	} else {
+		if(val < 0){
+			val = - val;
+			str->string[str->size++] = '-';
+		}
+		v = val;
+		for(n=0;v;n++) v /= 10;
+		str->size += n;
+		v = val;
+		while(v){
+			str->string[--str->size] = '0' + (v % 10);
+			v /= 10;
+		}
+		str->size += n;
+	}
+	str->string[str->size] = 0;
+}
+
+static inline void clear_string(String *str){ str->size = 0; str->string[0] = 0; }
+
+static inline string_size_t split_string(String *str, char separator, VStrv *vstrs){
+	VString *vstr;
+	string_size_t n_tab, i, s;
+	for(i=s=n_tab=0;i<str->size;i++){
+		if(str->string[i] == separator){
+			if(i > s){
+				str->string[i] = '\0';
+				vstr = next_ref_VStrv(vstrs);
+				vstr->string = str->string + s;
+				n_tab ++;
+				vstr->size = i - s;
+			}
+			s = i + 1;
+		}
+	}
+	if(i > s){
+		str->string[i] = '\0';
+		vstr = next_ref_VStrv(vstrs);
+		vstr->string = str->string + s;
+		n_tab ++;
+		vstr->size = i - s;
+	}
+	return n_tab;
+}
+
+static inline string_size_t split_vstring(VString *str, char separator, VStrv *vstrs, string_size_t cut){
+	VString *vstr;
+	string_size_t n_tab, i, s;
+	for(i=s=n_tab=0;i<str->size;i++){
+		if(str->string[i] == separator){
+			if(i > s){
+				if(cut) str->string[i] = '\0';
+				vstr = next_ref_VStrv(vstrs);
+				vstr->string = str->string + s;
+				n_tab ++;
+				vstr->size = i - s;
+			}
+			s = i + 1;
+		}
+	}
+	if(i > s){
+		if(cut) str->string[i] = '\0';
+		vstr = next_ref_VStrv(vstrs);
+		vstr->string = str->string + s;
+		n_tab ++;
+		vstr->size = i - s;
+	}
+	return n_tab;
+}
+
+static inline void reverse_string(String *str){
+	string_size_t i, j;
+	char c;
+	i = 0;
+	j = str->size - 1;
+	while(i < j){
+		swap_tmp(str->string[i], str->string[j], c);
+		i ++;
+		j --;
+	}
+}
+
+static inline void reverse_str(char *str, string_size_t len){
+	string_size_t i, j;
+	char c;
+	i = 0;
+	j = len - 1;
+	while(i < j){
+		swap_tmp(str[i], str[j], c);
+		i ++;
+		j --;
+	}
+}
+
+static inline void tidy_string(String *src, String *dst, char ch){
+	string_size_t i;
+	encap_string(dst, src->size);
+	for(i=0;i<src->size;i++){
+		if(src->string[i] != ch){
+			dst->string[dst->size ++] = src->string[i];
+		}
+	}
+	dst->string[dst->size] = 0;
+}
+
+static inline string_size_t occ_str(char *str, string_size_t len, char c){
+	string_size_t i, ret;
+	for(i=ret=0;i<len;i++){
+		if(str[i] == c) ret ++;
+	}
+	return ret;
+}
+
+static inline void trunc_string(String *str, string_size_t size){
+	if(size >= str->size || size < 0) return;
+	str->size = size;
+	str->string[size] = 0;
+}
+
+static inline String* clone_string(String *str){
+	String *clone;
+	clone = init_string(str->size);
+	append_string(clone, str->string, str->size);
+	return clone;
+}
+
+static inline void free_string(String *str){ free(str->string); free(str); }
+
+static inline void free_vstring(VString *str){ free(str); }
+
+#endif

dagcns.h

@@ -0,0 +1,899 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "dna.h"
+#include "chararray.h"
+#include "list.h"
+#include "queue.h"
+#include "hashset.h"
+#include "general_graph.h"
+#include <math.h>
+#include <float.h>
+
+#ifndef PACBIO_PROBS_DAGCNS_RJ_H
+#define PACBIO_PROBS_DAGCNS_RJ_H
+
+#define DAGCNS_MAX_LEN	0x3FFF // 16k
+
+static int dagcns_debug = 0;
+
+typedef struct {
+	int x, y;
+	u4i dnidx;
+	f8i probs[2];
+} bdp_node_t;
+define_list(bdpnodev, bdp_node_t);
+
+typedef struct {
+	f8i prob;
+	u1i cigar, base;
+} bdp_edge_t;
+define_list(bdpedgev, bdp_edge_t);
+
+define_simple_geg_callback(bdp, bdpnodev, bdp_node_t, bdpedgev, bdp_edge_t);
+
+typedef struct {
+	u4i gnidx, dnidx;
+} bdp_link_t;
+define_list(bdplinkv, bdp_link_t);
+
+typedef struct {
+	uint32_t nodes[2];
+	uint32_t links[2];
+	uint32_t cov:28, visit:1, closed:1, cns:1, alt:1;
+	double   score;
+} dagedge_t;
+define_list(dagedgev, dagedge_t);
+
+#define NODE_MAX_FW_EDGE	0xFFFFFFFFU
+typedef struct dagnode_t {
+	uint32_t pos:28, base:2, cns:1, visit:1;
+	uint32_t fw_edge;
+	uint32_t edges[2];
+	f8i      aux;
+} dagnode_t;
+define_list(dagnodev, dagnode_t);
+
+typedef struct {
+	uint32_t pos;
+	uint32_t bases[4];
+} dagsnp_t;
+define_list(dagsnpv, dagsnp_t);
+
+typedef struct {
+	u8list   *cns;
+	u32list  *deps;
+	dagnodev *nodes;
+	dagedgev *edges;
+	u32list  *trash;
+	String   *alns[2];
+	int      W, M, X, I, D, E;
+	f4i      pM, pX, pI, pD; // log(prob_M)
+	double   ref_penalty, alt_penalty; // 0.5 and 0.2
+	double   cns_score;
+	uint32_t cns_head, backbone_size;
+} DAGCNS;
+
+static inline DAGCNS* init_dagcns(int W, int M, int X, int I, int D, int E, f4i pM, f4i pX, f4i pI, f4i pD){
+	DAGCNS *g;
+	g = malloc(sizeof(DAGCNS));
+	g->cns = init_u8list(1024);
+	g->deps = init_u32list(1024);
+	g->nodes = init_dagnodev(1024);
+	g->edges = init_dagedgev(1024);
+	g->trash = init_u32list(1024);
+	g->alns[0] = init_string(1024);
+	g->alns[1] = init_string(1024);
+	g->cns_score = 0;
+	g->cns_head = 0xFFFFFFFFU;
+	g->backbone_size = 0;
+	g->W = W;
+	g->M = M;
+	g->X = X;
+	g->I = I;
+	g->D = D;
+	g->E = E;
+	g->pM = pM;
+	g->pX = pX;
+	g->pI = pI;
+	g->pD = pD;
+	g->ref_penalty = 0.5;
+	g->alt_penalty = 0.2;
+	return g;
+}
+
+static inline void free_dagcns(DAGCNS *g){
+	free_dagnodev(g->nodes);
+	free_dagedgev(g->edges);
+	free_u32list(g->trash);
+	free_u8list(g->cns);
+	free_u32list(g->deps);
+	free_string(g->alns[0]);
+	free_string(g->alns[1]);
+	free(g);
+}
+
+static inline void reset_dagcns(DAGCNS *g){
+	clear_dagnodev(g->nodes);
+	clear_dagedgev(g->edges);
+	clear_u32list(g->trash);
+	clear_u8list(g->cns);
+	clear_u32list(g->deps);
+	g->cns_score = 0;
+	g->cns_head = 0xFFFFFFFFU;
+	g->backbone_size = 0;
+}
+
+static uint32_t prepare_node_dagcns(DAGCNS *g, uint32_t pos, uint8_t base){
+	dagnode_t *n;
+	n = next_ref_dagnodev(g->nodes);
+	n->pos = pos;
+	n->base = base;
+	n->cns = 0;
+	n->aux  = 0;
+	n->visit = 0;
+	n->edges[0] = 0xFFFFFFFFU;
+	n->edges[1] = 0xFFFFFFFFU;
+	n->fw_edge = NODE_MAX_FW_EDGE;
+	return g->nodes->size - 1;
+}
+
+static inline dagedge_t* find_edge_by_node_dagcns(DAGCNS *g, uint32_t nid1, uint32_t nid2, int dir){
+	dagnode_t *n;
+	dagedge_t *e;
+	n = ref_dagnodev(g->nodes, nid1);
+	if(n->edges[dir] != 0xFFFFFFFFU){
+		e = ref_dagedgev(g->edges, n->edges[dir]);
+		while(1){
+			if(e->nodes[dir] == nid2) return e;
+			if(e->links[dir] == 0xFFFFFFFFU) break;
+			e = ref_dagedgev(g->edges, e->links[dir]);
+		}
+	}
+	return NULL;
+}
+
+static inline dagedge_t* add_edge_dagcns(DAGCNS *g, uint32_t nid1, uint32_t nid2, int dir){
+	dagnode_t *n;
+	dagedge_t *e;
+	uint32_t eid;
+	n = ref_dagnodev(g->nodes, nid1);
+	if(pop_u32list(g->trash, &eid)){
+		e = ref_dagedgev(g->edges, eid);
+	} else {
+		eid = g->edges->size;
+		e = next_ref_dagedgev(g->edges);
+	}
+	e->nodes[!dir] = nid1;
+	e->nodes[dir] = nid2;
+	e->cov = 1;
+	e->score = 0;
+	e->visit = 0;
+	e->closed = 0;
+	e->cns = 0;
+	e->alt = 0;
+	e->links[dir] = n->edges[dir];
+	n->edges[dir] = eid;
+	n = ref_dagnodev(g->nodes, nid2);
+	e->links[!dir] = n->edges[!dir];
+	n->edges[!dir] = eid;
+	return e;
+}
+
+static inline dagedge_t* prepare_edge_dagcns(DAGCNS *g, uint32_t nid1, uint32_t nid2, int dir){
+	dagedge_t *e;
+	e = find_edge_by_node_dagcns(g, nid1, nid2, dir);
+	if(e){ e->cov ++; return e; }
+	return add_edge_dagcns(g, nid1, nid2, dir);
+}
+
+static inline void gen_pregraph_dagcns(DAGCNS *g){
+	dagedge_t *e;
+	uint32_t i;
+	clear_dagnodev(g->nodes);
+	clear_dagedgev(g->edges);
+	clear_u32list(g->trash);
+	clear_u32list(g->deps);
+	g->backbone_size = g->cns->size;
+	for(i=0;i<g->cns->size;i++){
+		push_u32list(g->deps, 0);
+		prepare_node_dagcns(g, i, g->cns->buffer[i]);
+		if(i){ // make sure the graph is conntective even the alignment is partial
+			e = add_edge_dagcns(g, i - 1, i, 0);
+			e->cov = 0;
+		}
+	}
+}
+
+static inline int remove_edge_dagcns(DAGCNS *g, uint32_t eid){
+	dagnode_t *n;
+	dagedge_t *e;
+	uint32_t i, lst;
+	for(i=0;i<2;i++){
+		e = ref_dagedgev(g->edges, eid);
+		lst = e->links[i];
+		n = ref_dagnodev(g->nodes, e->nodes[!i]);
+		if(n->edges[i] == eid){
+			n->edges[i] = lst;
+		} else if(n->edges[i] == 0xFFFFFFFFU){
+			//fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr); abort();
+			return 0;
+		} else {
+			e = ref_dagedgev(g->edges, n->edges[i]);
+			while(1){
+				if(e->links[i] == eid){
+					e->links[i] = lst; break;
+				} else if(e->links[i] == 0xFFFFFFFFU){
+					//fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr); abort();
+					return 0;
+				} else {
+					e = ref_dagedgev(g->edges, e->links[i]);
+				}
+			}
+		}
+	}
+	push_u32list(g->trash, eid);
+	return 1;
+}
+
+#define MIN_SCORE	-0x0FFFFFFF
+
+static inline f8i log_sum(f8i a, f8i b){
+	f8i c;
+	c = num_max(a, b);
+	if(c - a >= 10 || c - b >= 10) return c;
+	return logl(expl(a - c) + expl(b - c)) + c;
+}
+
+typedef struct {
+	int x, y, d;
+} bdp_beg_t;
+define_list(bdpbegv, bdp_beg_t);
+
+static inline void fprint_dot_bdpgraph(GEGraph *g, bdpnodev *bnodes, bdpedgev *bedges, char *prefix, char *suffix){
+	static const char *colors[2][2] = {{"blue", "green"}, {"red", "gray"}};
+	FILE *out;
+	ge_node_t *n;
+	bdp_node_t *bn;
+	ge_edge_t *e;
+	bdp_edge_t *be;
+	u8i i;
+	out = open_file_for_write(prefix, suffix, 1);
+	fprintf(out, "digraph {\nnode [shape=record]\nrankdir=LR\n");
+	for(i=0;i<g->nodes->size;i++){
+		n = ref_genodev(g->nodes, i);
+		if(n->closed) continue;
+		bn = ref_bdpnodev(bnodes, offset_genodev(g->nodes, n));
+		fprintf(out, " N%llu [label=\"{N%llu|%d|%d}|{%.4Lf|%.4Lf}\"]\n", i, i, bn->x, bn->y, bn->probs[0], bn->probs[1]);
+	}
+	for(i=1;i<g->edges->size;i++){
+		e = ref_geedgev(g->edges, i);
+		if(e->closed) continue;
+		be = ref_bdpedgev(bedges, offset_geedgev(g->edges, e));
+		fprintf(out, " N%llu -> N%llu [label=\"%c%c:%d:%c:%.4Lf\" color=%s]\n", (u8i)e->node1, (u8i)e->node2, "+-"[e->dir1], "+-"[e->dir2], e->cov, "MIDX"[be->cigar], be->prob, colors[e->dir1][e->dir2]);
+		fprintf(out, " N%llu -> N%llu [label=\"%c%c:%d:%c:%.4Lf\" color=%s]\n", (u8i)e->node2, (u8i)e->node1, "-+"[e->dir2], "-+"[e->dir1], e->cov, "MIDX"[be->cigar], be->prob, colors[!e->dir2][!e->dir1]);
+	}
+	fprintf(out, "}\n");
+	fclose(out);
+}
+
+static inline u4i dp_matrix2alignment_graph(DAGCNS *dag, u1i *query, u1i *z, int x, int y, GEGraph *g, bdpnodev *bnodes, bdpedgev *bedges){
+	ge_node_t *gn, *gn2;
+	ge_edge_ref_t *gf;
+	ge_edge_t *ge;
+	bdp_node_t *bn, *bn2;
+	bdp_edge_t *be;
+	bdp_beg_t BEG;
+	UUhash *bnhash;
+	UUhash_t *u;
+	bdpbegv *stack;
+	f8i cigar_probs[4], curator;
+	u8v *idxs;
+	u1i *target;
+	u4i i, k, nidx, nlst, nbegs[2], visit, ret;
+	int d, f, n_col, ops[2], base, exists;
+	cigar_probs[0] = dag->pM;
+	cigar_probs[1] = dag->pI;
+	cigar_probs[2] = dag->pD;
+	cigar_probs[3] = dag->pX;
+	n_col = dag->cns->size;
+	target = dag->cns->buffer;
+	bdp_set_callbacks_gegraph(g, bnodes, bedges);
+	reset_gegraph(g);
+	stack = init_bdpbegv(4);
+	push_bdpbegv(stack, (bdp_beg_t){x, y, 0});
+	idxs = init_u8v(4);
+	nbegs[0] = nbegs[1] = 0;
+	bnhash = init_UUhash(1023);
+	ret = 0;
+	while(stack->size){
+		BEG = stack->buffer[--stack->size];
+		u = prepare_UUhash(bnhash, (((b8i)BEG.x) << 32) | BEG.y, &exists);
+		if(exists){
+			nidx = u->val;
+		} else {
+			gn = add_node_gegraph(g);
+			nidx = offset_genodev(g->nodes, gn);
+			u->key = (((b8i)BEG.x) << 32) | BEG.y;
+			u->val = nidx;
+			bn = ref_bdpnodev(bnodes, nidx);
+			bn->x = BEG.x; bn->y = BEG.y;
+			bn->dnidx = MAX_VALUE_U4;
+		}
+		nlst = nidx;
+		x = BEG.x; y = BEG.y;
+		d = BEG.d;
+		f = 0;
+		clear_u8v(idxs);
+		while(x >= 0 && y >= 0){
+			d = (z[x * n_col + y] >> (d << 1)) & 0x03;
+			if(d == 0){
+				if(query[x] == target[y]){
+					if(z[x * n_col + y] & (1 << 6)){
+						f = 1;
+						break;
+					} else if(BEG.d == 0){
+						z[x * n_col + y] |= 1 << 6;
+						push_bdpbegv(stack, (bdp_beg_t){x, y, 1});
+						push_bdpbegv(stack, (bdp_beg_t){x, y, 2});
+					}
+					ops[0] = 0; ops[1] = 3;
+				} else {
+					ops[0] = 1; ops[1] = 2;
+				}
+				//x --; y --; 
+			} else if(d == 1){
+				ops[0] = 1; ops[1] = 3;
+				//x --;
+			} else {
+				ops[0] = 2; ops[1] = 3;
+				//y --;
+			}
+			for(i=0;i<2;i++){
+				if(ops[i] == 3) break;
+				base = query[x];
+				if(ops[i] == 0){ x --; y --; }
+				else if(ops[i] == 1){ x --; }
+				else { y --; }
+				u = prepare_UUhash(bnhash, (((b8i)x) << 32) | y, &exists);
+				if(exists){
+					nidx = u->val;
+				} else {
+					gn = add_node_gegraph(g);
+					nidx = offset_genodev(g->nodes, gn);
+					u->key = (((b8i)x) << 32) | y;
+					u->val = nidx;
+					bn = ref_bdpnodev(bnodes, nidx);
+					bn->x = x; bn->y = y;
+					bn->dnidx = MAX_VALUE_U4;
+				}
+				ge = prepare_edge_gegraph(g, nlst, 1, nidx, 1, &exists);
+				if(exists){
+					ge->cov ++;
+				} else {
+					ge->cov = 1;
+					be = ref_bdpedgev(bedges, offset_geedgev(g->edges, ge));
+					be->cigar = ops[i];
+					be->base  = base;
+				}
+				nlst = nidx;
+				if(BEG.d) push_u8v(idxs, offset_geedgev(g->edges, ge));
+			}
+		}
+		if(BEG.d){
+			if(f == 0){
+				// not a bubble
+				for(i=0;i<idxs->size;i++){
+					ge = ref_geedgev(g->edges, idxs->buffer[i]);
+					ge->cov --;
+					if(ge->cov == 0) cut_edge_gegraph(g, ge);
+				}
+			} else ret ++;
+		} else {
+			ret ++;
+			nbegs[0] = g->nodes->size > 1? g->nodes->size - 2 : 0;
+			nbegs[1] = 0;
+		}
+	}
+	free_bdpbegv(stack);
+	free_UUhash(bnhash);
+	u4i cnt = 0;
+	for(i=0;i<g->nodes->size;i++){
+		gn = ref_genodev(g->nodes, i);
+		if(gn->edges[0].cnt || gn->edges[1].cnt){ cnt++; continue; }
+		gn->closed = 1;
+	}
+	// calculate probabilities (forward + backward)
+	if(nbegs[0] == nbegs[1]){
+		free_u8v(idxs);
+		return ret;
+	}
+	for(k=0;k<2;k++){
+		bn = ref_bdpnodev(bnodes, nbegs[k]);
+		bn->probs[k] = 0;
+		clear_u8v(idxs);
+		push_u8v(idxs, nbegs[k]);
+		visit = k + 1;
+		while(idxs->size){
+			gn = ref_genodev(g->nodes, idxs->buffer[idxs->size - 1]);
+			bn = ref_bdpnodev(bnodes, idxs->buffer[idxs->size - 1]);
+			idxs->size --;
+			geg_beg_iter_edges(g, gn, k, gf, ge);
+			if(ge->closed) continue;
+			be = ref_bdpedgev(bedges, offset_geedgev(g->edges, ge));
+			gn2 = ref_genodev(g->nodes, gf->flg? ge->node1 : ge->node2);
+			bn2 = ref_bdpnodev(bnodes, gf->flg? ge->node1 : ge->node2);
+			if(gn2->bt_visit == visit){
+				bn2->probs[k] = log_sum(bn2->probs[k], bn->probs[k] + cigar_probs[be->cigar]); // p1 + p2
+			} else {
+				gn2->bt_visit = visit;
+				gn2->unvisit = gn2->edges[!k].cnt;
+				bn2->probs[k] = bn->probs[k] + cigar_probs[be->cigar]; // p1 * p2
+			}
+			if(gn2->unvisit) gn2->unvisit --;
+			if(gn2->unvisit == 0){
+				push_u8v(idxs, offset_genodev(g->nodes, gn2));
+			}
+			geg_end_iter_edges();
+		}
+	}
+	free_u8v(idxs);
+	// calculate edge prob
+	curator = -FLT_MAX;
+	for(i=1;i<g->edges->size;i++){
+		ge = ref_geedgev(g->edges, i);
+		if(ge->closed) continue;
+		be = ref_bdpedgev(bedges, i);
+		be->prob = cigar_probs[be->cigar] + ref_bdpnodev(bnodes, ge->node1)->probs[ge->dir1] + ref_bdpnodev(bnodes, ge->node2)->probs[!ge->dir2];
+		if(be->prob > curator) curator = be->prob;
+	}
+	for(i=1;i<g->edges->size;i++){
+		ge = ref_geedgev(g->edges, i);
+		if(ge->closed) continue;
+		be = ref_bdpedgev(bedges, i);
+		be->prob = expl(be->prob - curator);
+	}
+	//fprint_dot_bdpgraph(g, bnodes, bedges, "geg.dot", NULL);
+	return nbegs[0];
+}
+
+static inline u4i branched_dynamic_programming_alignment(DAGCNS *g, u1i *query, int ql, GEGraph *geg, bdpnodev *bnodes, bdpedgev *bedges, u1v *mem_buffer){
+	u4i nbeg;
+	int *rh, *re;
+	u1i *z, *zi, d, *target;
+	int tl, n_col;
+	int i, j, jc, jb, je, h1, h, m, e, f, t;
+	int mi, mj, max;
+	int W, M, X, I, D, E;
+	tl = g->cns->size;
+	target = g->cns->buffer;
+	if(ql <= 0 || tl <= 0){ return 0; }
+	n_col = tl;
+	encap_u1v(mem_buffer, kswx_roundup8x((tl + 2) * sizeof(int)) + kswx_roundup8x((tl + 2) * sizeof(int)) + kswx_roundup8x(((long long)ql) * n_col));
+	rh = (int*)(mem_buffer->buffer + mem_buffer->size);
+	re = (int*)(mem_buffer->buffer + mem_buffer->size + kswx_roundup8x((tl + 2) * sizeof(int)));
+	z  =       (mem_buffer->buffer + mem_buffer->size + kswx_roundup8x((tl + 2) * sizeof(int)) + kswx_roundup8x((tl + 2) * sizeof(int)));
+	W = g->W;
+	M = g->M;
+	X = g->X;
+	I = g->I;
+	D = g->D;
+	E = g->E;
+	// banded DP, global alignment
+	rh[0] = 0;
+	re[1] = 0 + D + E;
+	for(j=2;j<=tl&&j<=W;j++) rh[j] = rh[j-1] + E;
+	for(;j<tl;j++) rh[j] = MIN_SCORE;
+	for(j=0;j<=tl;j++) re[j] = MIN_SCORE;
+	mi = mj = 0;
+	max = MIN_SCORE;
+	for(i=0;i<ql;i++){
+		f = MIN_SCORE;
+		jc = (i * tl / ql); // in case of biased gaps
+		jb = jc > W? jc - W : 0;
+		je = jc + W + 1 < tl? jc + W + 1 : tl;
+		h1 = jb == 0? (I + E * (i + 1)) : MIN_SCORE;
+		zi = &z[i * n_col];
+		for(j=jb;j<je;j++){
+			m = rh[j] + ((query[i] == target[j])? M : X);
+			rh[j] = h1;
+			e = re[j];
+			d = m >= e? 0 : 1;
+			h = m >= e? m : e;
+			d = h >= f? d : 2;
+			h = h >= f? h : f;
+			h1 = h;
+			t = m + I + E;
+			e = e + E;
+			d |= e > t? 1<<2 : 0;
+			e = e > t? e : t;
+			re[j] = e;
+			t = m + D + E;
+			f = f + E;
+			d |= f > t? 2<<4 : 0;
+			f = f > t? f : t;
+			zi[j] = d;
+		}
+		rh[j] = h1; re[j] = MIN_SCORE;
+		if(i + 1 == ql){
+			for(j=jb;j<je;j++){
+				if(rh[j + 1] > max){
+					max = rh[j + 1];
+					mi = i; mj = j;
+				}
+			}
+		} else if(je == tl){
+			if(h1 > max){
+				max = h1;
+				mi = i; mj = tl - 1;
+			}
+		}
+	}
+	if(max == MIN_SCORE) return 0;
+	nbeg = dp_matrix2alignment_graph(g, query, z, mi, mj, geg, bnodes, bedges);
+	return nbeg;
+}
+
+static inline void bdpgraph2dagcns(DAGCNS *dg, GEGraph *gg, bdpnodev *bnodes, bdpedgev *bedges, u4i nbeg, bdplinkv *stack){
+	dagnode_t *dn, *dn2;
+	dagedge_t *de;
+	ge_node_t *gn, *gn2;
+	ge_edge_ref_t *gf;
+	ge_edge_t *ge;
+	bdp_node_t *bn, *bn2;
+	bdp_edge_t *be;
+	bdp_link_t T;
+	u4i i, j, beg, end;
+	int open;
+	for(i=0;i<gg->nodes->size;i++) gg->nodes->buffer[i].bt_visit = 0;
+	clear_bdplinkv(stack);
+	push_bdplinkv(stack, (bdp_link_t){nbeg, 0xFFFFFFFFU});
+	beg = bnodes->buffer[nbeg].y;
+	end = beg;
+	open = 0;
+	while(stack->size){
+		T = stack->buffer[--stack->size];
+		gn = ref_genodev(gg->nodes, T.gnidx);
+		bn = ref_bdpnodev(bnodes, T.gnidx);
+		if(bn->y > (int)end) end = bn->y;
+		dn = (T.dnidx == 0xFFFFFFFFU)? NULL : ref_dagnodev(dg->nodes, T.dnidx);
+		geg_beg_iter_edges(gg, gn, 0, gf, ge);
+		if(ge->closed) continue;
+		be = ref_bdpedgev(bedges, offset_geedgev(gg->edges, ge));
+		gn2 = ref_genodev(gg->nodes, gf->flg? ge->node1 : ge->node2);
+		bn2 = ref_bdpnodev(bnodes, gf->flg? ge->node1 : ge->node2);
+		if(gn2->bt_visit == 0){
+			gn2->bt_visit = 1;
+			open ++;
+			gn2->unvisit = gn2->edges[1].cnt;
+		}
+		if(gn2->unvisit) gn2->unvisit --;
+		if(gn2->unvisit == 0){
+			open --;
+		}
+		if(open < 0){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+		}
+		if(bn2->dnidx == MAX_VALUE_U4){
+			if(be->cigar == 2){
+				dn2 = dn;
+			} else if(dn && (open || be->cigar == 1)){
+				dn2 = ref_dagnodev(dg->nodes, prepare_node_dagcns(dg, bn2->y, be->base));
+				dn  = dn? ref_dagnodev(dg->nodes, T.dnidx) : NULL;
+			} else if((dn || open == 0) && be->cigar == 0){ // cigar == 0 && open == 0
+				dn2 = ref_dagnodev(dg->nodes, bn2->y);
+			} else dn2 = NULL;
+			bn2->dnidx = dn2? offset_dagnodev(dg->nodes, dn2) : MAX_VALUE_U4;
+		} else {
+			dn2 = ref_dagnodev(dg->nodes, bn2->dnidx);
+		}
+		if(dn && dn2 && dn != dn2){
+			de = prepare_edge_dagcns(dg, offset_dagnodev(dg->nodes, dn), offset_dagnodev(dg->nodes, dn2), 0);
+			de->score += be->prob;
+		}
+		if(gn2->unvisit == 0){
+			push_bdplinkv(stack, (bdp_link_t){offset_genodev(gg->nodes, gn2), dn2? offset_dagnodev(dg->nodes, dn2) : 0xFFFFFFFFU});
+		}
+		geg_end_iter_edges();
+	}
+	for(j=beg;j<end;j++) dg->deps->buffer[j] ++;
+	return;
+}
+
+static inline void merge_nodes_core_dagcns(DAGCNS *g, uint32_t nid, u32list *stack, u32list *cache[4], int dir){
+	dagnode_t *n0, *n2, *n;
+	dagedge_t *e, *e2, *e1;
+	uint32_t base, eid, nid1, i, ret;
+	clear_u32list(stack);
+	push_u32list(stack, nid);
+	ret = 0;
+	while(pop_u32list(stack, &nid)){
+		n0 = ref_dagnodev(g->nodes, nid);
+		if((eid = n0->edges[dir]) == 0xFFFFFFFFU) continue;
+		clear_u32list(cache[0]);
+		clear_u32list(cache[1]);
+		clear_u32list(cache[2]);
+		clear_u32list(cache[3]);
+		while(1){
+			e = ref_dagedgev(g->edges, eid);
+			n = ref_dagnodev(g->nodes, e->nodes[dir]);
+			e2 = ref_dagedgev(g->edges, n->edges[!dir]);
+			if(e2->links[!dir] == 0xFFFFFFFFU){ // check whether there is only one edge from n -(!dir)-> n0
+				push_u32list(cache[n->base], eid);
+			}
+			if((eid = e->links[dir]) == 0xFFFFFFFFU) break;
+		}
+		for(base=0;base<4;base++){
+			if(cache[base]->size < 2) continue;
+			for(i=0;i<cache[base]->size;i++) ref_dagedgev(g->edges, cache[base]->buffer[i])->visit = 1;
+			e1 = ref_dagedgev(g->edges, cache[base]->buffer[0]);
+			n = ref_dagnodev(g->nodes, e1->nodes[dir]);
+			eid = n->edges[dir];
+			nid1 = e1->nodes[dir];
+			while(eid != 0xFFFFFFFFU){
+				e = ref_dagedgev(g->edges, eid);
+				e->visit = 1;
+				eid = e->links[dir];
+			}
+			for(i=1;i<cache[base]->size;i++){
+				e2 = ref_dagedgev(g->edges, cache[base]->buffer[i]);
+				n2 = ref_dagnodev(g->nodes, e2->nodes[dir]);
+				e1->cov += e2->cov;
+				e1->score += e2->score;
+				remove_edge_dagcns(g, cache[base]->buffer[i]);
+				eid = n2->edges[dir];
+				while(eid != 0xFFFFFFFFU){
+					e2 = ref_dagedgev(g->edges, eid);
+					e  = prepare_edge_dagcns(g, nid1, e2->nodes[dir], dir);
+					{
+						e1 = ref_dagedgev(g->edges, cache[base]->buffer[0]); // memory referred by e1 may be freed in prepare_edge_dagcns
+					}
+					e->cov = e->cov - 1 + e2->cov;
+					e->score = e->score + e2->score;
+					e->visit = 1;
+					eid = e2->links[dir];
+				}
+				eid = n2->edges[dir];
+				while(eid != 0xFFFFFFFFU){
+					e2 = ref_dagedgev(g->edges, eid);
+					remove_edge_dagcns(g, eid); // e2->links retain the same values after removing
+					eid = e2->links[dir];
+				}
+			}
+			//n = ref_dagnodev(g->nodes, e1->nodes[dir]);
+			//eid = n->edges[dir];
+			//if(eid != 0xFFFFFFFFU && g->edges->buffer[eid].links[dir] == 0xFFFFFFFFU) continue; // we had merged a bubble branch1:A->C->T, branch2:A->C->T
+			push_u32list(stack, nid1);
+		}
+	}
+}
+
+static inline int has_non_visited_edge_dagcns(DAGCNS *g, uint32_t nid, int dir){
+	dagnode_t *n;
+	dagedge_t *e;
+	uint32_t eid;
+	n = ref_dagnodev(g->nodes, nid);
+	eid = n->edges[dir];
+	while(eid != 0xFFFFFFFFU){
+		e = ref_dagedgev(g->edges, eid);
+		if(e->visit == 0) return 1;
+		eid = e->links[dir];
+	}
+	return 0;
+}
+
+static inline void print_local_dot_dagcns(DAGCNS *g, uint32_t nid, int distance, FILE *out){
+	u32list *stack;
+	u32hash *hash;
+	dagnode_t *n, *n1, *n2;
+	dagedge_t *e;
+	uint32_t id1, id2, eid, *u;
+	int lo, hi, dir, exists;
+	n = ref_dagnodev(g->nodes, nid);
+	stack = init_u32list(32);
+	hash = init_u32hash(1023);
+	lo = n->pos - distance;
+	hi = n->pos + distance;
+	push_u32list(stack, nid);
+	put_u32hash(hash, nid);
+	fprintf(out, "digraph {\nrankdir=LR\n");
+	while(stack->size){
+		id1 = stack->buffer[--stack->size];
+		n1 = ref_dagnodev(g->nodes, id1);
+		for(dir=0;dir<1;dir++){
+			eid = n1->edges[dir];
+			while(eid != 0xFFFFFFFFU){
+				e = ref_dagedgev(g->edges, eid);
+				id2 = e->nodes[dir];
+				n2 = ref_dagnodev(g->nodes, id2);
+				fprintf(out, "N%d_%d_%c -> N%d_%d_%c [label=\"%d:%0.6f\" color=%s]\n", id1, n1->pos, "ACGT"[n1->base], id2, n2->pos, "ACGT"[n2->base], e->cov, e->score, e->visit? "blue" : "black");
+				if(n2->pos >= lo && n2->pos <= hi){
+					u = prepare_u32hash(hash, id2, &exists);
+					if(exists){
+					} else {
+						*u = id2;
+						push_u32list(stack, id2);
+					}
+				}
+				eid = e->links[dir];
+			}
+		}
+	}
+	fprintf(out, "}\n");
+	fflush(out);
+	free_u32list(stack);
+	free_u32hash(hash);
+}
+
+static inline void fprint_local_dot_dagcns(DAGCNS *g, u4i nid, int level, char *prefix, char *suffix){
+	FILE *out;
+	out = open_file_for_write(prefix, suffix, 1);
+	print_local_dot_dagcns(g, nid, level, out);
+	fclose(out);
+}
+
+static inline void merge_nodes_dagcns(DAGCNS *g){
+	dagnode_t *n;
+	dagedge_t *e;
+	u32list *stack, *cache[4];
+	u32fifo *queue;
+	uint32_t i, nid, eid;
+	stack = init_u32list(1024);
+	cache[0] = init_u32list(4);
+	cache[1] = init_u32list(4);
+	cache[2] = init_u32list(4);
+	cache[3] = init_u32list(4);
+	for(i=0;i<g->edges->size;i++) g->edges->buffer[i].visit = 0;
+	for(i=0;i<g->nodes->size;i++) g->nodes->buffer[i].visit = 0;
+	queue = init_u32fifo();
+	for(i=0;i<g->nodes->size;i++){
+		n = ref_dagnodev(g->nodes, i);
+		if(n->edges[1] != 0xFFFFFFFFU) continue;
+		push_u32fifo(queue, i);
+	}
+	//dagcns_debug = 2;
+	while(pop_u32fifo(queue, &nid)){
+		if(dagcns_debug > 1) fprintf(stdout, "\npop(%u) %u\n", (u4i)queue->size, nid);
+		n = ref_dagnodev(g->nodes, nid);
+		if(n->visit) continue;
+		n->visit = 1;
+		merge_nodes_core_dagcns(g, nid, stack, cache, 1);
+		merge_nodes_core_dagcns(g, nid, stack, cache, 0);
+		n = ref_dagnodev(g->nodes, nid);
+		eid = n->edges[0];
+		while(eid != 0xFFFFFFFFU){
+			e = ref_dagedgev(g->edges, eid);
+			e->visit = 1;
+			eid = e->links[0];
+		}
+		eid = n->edges[0];
+		while(eid != 0xFFFFFFFFU){
+			e = ref_dagedgev(g->edges, eid);
+			if(!has_non_visited_edge_dagcns(g, e->nodes[0], 1)){
+				if(dagcns_debug > 1) fprintf(stdout, "push %u\n", e->nodes[0]);
+				push_u32fifo(queue, e->nodes[0]);
+			}
+			eid = e->links[0];
+		}
+	}
+	free_u32fifo(queue);
+	free_u32list(stack);
+	free_u32list(cache[0]);
+	free_u32list(cache[1]);
+	free_u32list(cache[2]);
+	free_u32list(cache[3]);
+}
+
+static inline void print_seq_dagcns(DAGCNS *g, FILE *out){
+	char buffer[100];
+	uint32_t i, j;
+	for(i=j=0;i<g->cns->size;i++){
+		buffer[j++] = bit_base_table[g->cns->buffer[i]];
+		if(j == 99){
+			buffer[j] = '\0';
+			fprintf(out, "%s", buffer);
+			j = 0;
+		}
+	}
+	buffer[j] = '\0';
+	fprintf(out, "%s", buffer);
+}
+
+static inline void gen_consensus_dagcns(DAGCNS *g, u32list *map){
+	dagnode_t *n1, *n2;
+	dagedge_t *e;
+	u32fifo *queue;
+	uint32_t i, lst, nid, eid, best_e;
+	f8i best_s, score;
+	queue = init_u32fifo();
+	if(queue == NULL){ // un-reachable, but is used to call fprint_local_dot_dagcns in gdb Debug
+		fprint_local_dot_dagcns(g, 0, 10, "test.dot", NULL);
+	}
+	for(i=0;i<g->nodes->size;i++){
+		n1 = ref_dagnodev(g->nodes, i);
+		if(n1->edges[0] == 0xFFFFFFFFU && n1->edges[1] != 0xFFFFFFFFU){
+			push_u32fifo(queue, i);
+			n1->fw_edge = NODE_MAX_FW_EDGE;
+			n1->aux = 0;
+		}
+	}
+	for(i=0;i<g->edges->size;i++) g->edges->buffer[i].visit = 0;
+	while(pop_u32fifo(queue, &nid)){
+		best_s = - FLT_MAX;
+		best_e = 0xFFFFFFFFU;
+		n1 = ref_dagnodev(g->nodes, nid);
+		eid = n1->edges[0];
+		while(eid != 0xFFFFFFFFU){
+			e = ref_dagedgev(g->edges, eid);
+			n2 = ref_dagnodev(g->nodes, e->nodes[0]);
+			if(e->nodes[0] < g->backbone_size){
+				//score = n2->aux + e->cov - g->ref_penalty * g->deps->buffer[n1->pos];
+				score = n2->aux + e->score - g->ref_penalty * g->deps->buffer[n1->pos];
+			} else {
+				//score = n2->aux + e->cov - g->alt_penalty * g->deps->buffer[n1->pos];
+				score = n2->aux + e->score - g->alt_penalty * g->deps->buffer[n1->pos];
+			}
+			if(score > best_s){
+				best_s = score;
+				best_e = eid;
+			}
+			eid = e->links[0];
+		}
+		if(best_s > - FLT_MAX) n1->aux = best_s;
+		n1->fw_edge = best_e;
+		eid = n1->edges[1];
+		while(eid != 0xFFFFFFFFU){
+			e = ref_dagedgev(g->edges, eid);
+			e->visit = 1;
+			if(!has_non_visited_edge_dagcns(g, e->nodes[1], 0)){
+				push_u32fifo(queue, e->nodes[1]);
+			}
+			eid = e->links[1];
+		}
+	}
+	free_u32fifo(queue);
+	clear_u8list(g->cns);
+	clear_u32list(g->deps);
+	if(map) clear_u32list(map);
+	g->cns_head = 0;
+	if(g->nodes->size == 0) return;
+	n1 = ref_dagnodev(g->nodes, g->cns_head);
+	g->cns_score = n1->aux;
+	n1->cns = 1;
+	lst = 0;
+	if(map && g->cns_head < g->backbone_size){
+		while(lst < g->cns_head){ push_u32list(map, g->cns->size); lst ++; }
+	}
+	push_u8list(g->cns, n1->base);
+	push_u32list(g->deps, 0);
+	while(n1->fw_edge != NODE_MAX_FW_EDGE){
+		e = ref_dagedgev(g->edges, n1->fw_edge);
+		e->cns = 1;
+		if(map && e->nodes[0] < g->backbone_size){
+			while(lst < e->nodes[0]){ push_u32list(map, g->cns->size); lst ++; }
+		}
+		n1 = ref_dagnodev(g->nodes, e->nodes[0]);
+		n1->cns = 1;
+		push_u8list(g->cns, n1->base);
+		push_u32list(g->deps, 0);
+	}
+	if(map) while(lst <= g->backbone_size){ push_u32list(map, g->cns->size); lst ++; }
+}
+
+#endif

filereader.h

@@ -0,0 +1,673 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __FILEREADER_RJ_H
+#define __FILEREADER_RJ_H
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <errno.h>
+#include "chararray.h"
+#include "mem_share.h"
+#include "list.h"
+#include "thread.h"
+#include "pgzf.h"
+
+#define BIOSEQ_ATTR_NULL	0
+#define BIOSEQ_ATTR_TAG		1
+#define BIOSEQ_ATTR_SEQ		2
+#define BIOSEQ_ATTR_QLT		4
+#define BIOSEQ_ATTR_FULL	7
+
+typedef struct {
+	String *tag, *seq, *dsc, *qlt;
+	u4i attr;
+} BioSequence;
+
+#define FILEREADER_TYPE_NULL	0
+#define FILEREADER_TYPE_FASTA	1
+#define FILEREADER_TYPE_FASTQ	2
+#define FILEREADER_TYPE_TEXT	3
+
+#define FILEREADER_ATTR_NULL	0
+#define FILEREADER_ATTR_NORMAL	1
+#define FILEREADER_ATTR_STDIN	2
+#define FILEREADER_ATTR_PROC	3
+#define FILEREADER_ATTR_TEXT	4
+#define FILEREADER_ATTR_USER	5 // defined by user
+
+typedef size_t (*read_data_func)(void *obj, void *dat, size_t len);
+typedef void (*close_input_func)(void *obj);
+
+static inline size_t _read_data_file(void *obj, void *dat, size_t len){ return fread(dat, 1, len, (FILE*)obj); }
+static inline void _close_input_file(void *obj){ if(obj) fclose((FILE*)obj); }
+static inline void _close_input_proc(void *obj){ if(obj) pclose((FILE*)obj); }
+
+typedef struct {
+	int file_attr;
+	char *filename;
+	void *_file;
+	read_data_func _read;
+	close_input_func _close;
+} file_src_t;
+define_list_core(filesrcv, file_src_t, int, 0xFF);
+
+typedef struct {
+	filesrcv *files;
+	int fidx;
+	char *buffer[2];
+	int ridx, widx, flag;
+	u8i bufmax, bufoff, bufcnt[2];
+#ifdef FR_USE_SPINLOCK
+	pthread_spinlock_t lock;
+#else
+	pthread_mutex_t lock;
+#endif
+	char line_breaker;
+	char delimiter;
+	u8i  n_char, n_line;
+	String *line, *line2;
+	VStrv  *tabs;
+	int rollback; // line will be re-used in next readline
+	// thread
+	pthread_t pid;
+	int running;
+	int eof;
+} FileReader;
+
+static inline BioSequence* init_biosequence(){
+	BioSequence *seq;
+	seq = malloc(sizeof(BioSequence));
+	seq->tag = init_string(32);
+	seq->seq = init_string(32);
+	seq->dsc = init_string(32);
+	seq->qlt = init_string(32);
+	seq->attr = BIOSEQ_ATTR_FULL;
+	return seq;
+}
+
+static inline void reset_biosequence(BioSequence *seq){
+	clear_string(seq->tag);
+	clear_string(seq->seq);
+	clear_string(seq->dsc);
+	clear_string(seq->qlt);
+}
+
+static inline void free_biosequence(BioSequence *seq){
+	free_string(seq->tag);
+	free_string(seq->seq);
+	free_string(seq->dsc);
+	free_string(seq->qlt);
+	free(seq);
+}
+
+static inline void* file_src_thread_func(void *obj){
+	FileReader *fr;
+	file_src_t *fc;
+	void *_file;
+	read_data_func _read;
+	close_input_func _close;
+	size_t off, cnt, len;
+	fr = (FileReader*)obj;
+	while(fr->running){
+		if(fr->fidx >= fr->files->size){
+			fr->eof = 1;
+			microsleep(1);
+		} else {
+			fr->eof = 0;
+			fc = ref_filesrcv(fr->files, fr->fidx);
+			_file = NULL;
+			_read = NULL;
+			_close = NULL;
+			switch(fc->file_attr){
+				case FILEREADER_ATTR_TEXT:
+					len = strlen(fc->filename);
+					off = 0;
+					while(fr->running && len){
+						while(fr->flag == 1 && fr->running){ nano_sleep(1); }
+						cnt = num_min(len, fr->bufmax);
+						memcpy(fr->buffer[fr->widx], fc->filename + off, cnt);
+						fr->flag = 1;
+						off += cnt;
+						len -= cnt;
+						fr->widx = !fr->widx;
+					}
+					break;
+				case FILEREADER_ATTR_STDIN:
+					if(_file == NULL){
+						_file = fc->_file = stdin;
+						_read = fc->_read = _read_data_file;
+						_close = fc->_close = NULL;
+					}
+				case FILEREADER_ATTR_PROC:
+					if(_file == NULL){
+						_file = fc->_file = popen(fc->filename, "r");
+						_read = fc->_read = _read_data_file;
+						_close = fc->_close = _close_input_proc;
+					}
+				case FILEREADER_ATTR_USER:
+					if(_file == NULL){
+						_file = fc->_file;
+						_read = fc->_read;
+						_close = fc->_close;
+					}
+				default:
+					if(_file == NULL){
+						_file = fc->_file = open_file_for_read(fc->filename, NULL);
+						_read = fc->_read = _read_data_file;
+						_close = fc->_close = _close_input_file;
+					}
+					while(fr->running){
+						while(fr->flag == 1){
+							nano_sleep(1);
+							if(fr->running == 0){
+								break;
+							}
+						}
+						if(fr->flag == 1) break;
+						fr->bufcnt[fr->widx] = _read(_file, fr->buffer[fr->widx], fr->bufmax);
+						fr->widx = !fr->widx;
+						fr->flag = 1;
+						if(fr->bufcnt[!fr->widx] == 0) break;
+					}
+			}
+			if(_file && _close){
+				_close(_file);
+			}
+			fr->fidx ++;
+		}
+	}
+	return NULL;
+}
+
+static inline FileReader* init_filereader(){
+	FileReader *fr;
+	fr = malloc(sizeof(FileReader));
+	fr->files = init_filesrcv(4);
+	fr->fidx = 0;
+	fr->bufmax = 128 * 1024;
+	fr->bufoff = 0;
+	fr->bufcnt[0] = 0;
+	fr->bufcnt[1] = 0;
+	fr->ridx = 0;
+	fr->widx = 1;
+	fr->flag = 0;
+#ifdef FR_USE_SPINLOCK
+	pthread_spin_init(&fr->lock, 0);
+#else
+	pthread_mutex_init(&fr->lock, NULL);
+#endif
+	fr->buffer[0] = malloc(fr->bufmax);
+	fr->buffer[1] = malloc(fr->bufmax);
+	fr->line_breaker = '\n';
+	fr->delimiter = '\t';
+	fr->n_char = 0;
+	fr->n_line = 0;
+	fr->line = init_string(32);
+	fr->line2 = init_string(32);
+	fr->tabs = init_VStrv(16);
+	fr->rollback = 0;
+	fr->pid  = 0;
+	fr->running = 1;
+	fr->eof = 0;
+	return fr;
+}
+
+static inline void beg_asyn_filereader(FileReader *fr){
+	if(pthread_create(&fr->pid, NULL, file_src_thread_func, fr) != 0){
+		fprintf(stderr, " -- Failed to create thread [%s] in %s -- %s:%d --\n", "file_src_thread_func", __FUNCTION__, __FILE__, __LINE__);
+		fr->pid = 0; // switch to directed read
+	}
+}
+
+static inline void reset_filereader(FileReader *fr){
+	if(fr->pid){
+		fr->running = 0;
+		pthread_join(fr->pid, NULL);
+	}
+	fr->fidx = 0;
+	fr->bufoff = 0;
+	fr->bufcnt[0] = 0;
+	fr->bufcnt[1] = 0;
+	fr->ridx = 0;
+	fr->widx = 0;
+	fr->flag = 0;
+#ifdef FR_USE_SPINLOCK
+	pthread_spin_destroy(&fr->lock);
+	pthread_spin_init(&fr->lock, 0);
+#else
+	pthread_mutex_destroy(&fr->lock);
+	pthread_mutex_init(&fr->lock, NULL);
+#endif
+	clear_string(fr->line);
+	clear_VStrv(fr->tabs);
+	fr->rollback = 0;
+	fr->n_line = 0;
+	fr->n_char = 0;
+	fr->running = 1;
+	fr->eof = 0;
+	if(fr->pid){
+		fr->pid = 0;
+		beg_asyn_filereader(fr);
+	}
+}
+
+static inline void free_filereader(FileReader *fr){
+	file_src_t *f;
+	int i;
+	if(fr->pid){
+		fr->running = 0;
+		pthread_join(fr->pid, NULL);
+	}
+	for(i=0;i<fr->files->size;i++){
+		f = ref_filesrcv(fr->files, i);
+		if(f->filename) free(f->filename);
+	}
+#ifdef FR_USE_SPINLOCK
+	pthread_spin_destroy(&fr->lock);
+#else
+	pthread_mutex_destroy(&fr->lock);
+#endif
+	free(fr->buffer[0]);
+	free(fr->buffer[1]);
+	free_filesrcv(fr->files);
+	free_string(fr->line);
+	free_string(fr->line2);
+	free_VStrv(fr->tabs);
+	free(fr);
+}
+
+static inline int push_filereader(FileReader *fr, char *filename){
+	file_src_t *f;
+	int len;
+	f = next_ref_filesrcv(fr->files);
+	f->_file = NULL;
+	f->_read = NULL;
+	f->_close = NULL;
+	len = filename? strlen(filename) : 0;
+	while(len && filename[len-1] == ' ') len --;
+	if(len == 0 || strcmp(filename, "-") == 0){
+		f->filename = NULL;
+		f->file_attr = FILEREADER_ATTR_STDIN;
+	} else if(filename[len-1] == '|'){
+		f->filename = malloc(len);
+		strncpy(f->filename, filename, len - 1);
+		f->file_attr = FILEREADER_ATTR_PROC;
+	} else if(len > 3 && strcmp(filename + len - 3, ".gz") == 0){
+		//f->filename = malloc(len + 20);
+		//sprintf(f->filename, "gzip -dc %s", filename);
+		//f->file_attr = FILEREADER_ATTR_PROC;
+		f->filename = strdup(filename);
+		f->file_attr = FILEREADER_ATTR_USER;
+		f->_file = open_pgzf_reader(open_file_for_read(f->filename, NULL), 0, 4);
+		f->_read = read_pgzf4filereader;
+		f->_close = close_pgzf4filereader;
+	} else if(len > 5 && strcmp(filename + len - 5, ".pgzf") == 0){
+		f->filename = strdup(filename);
+		f->file_attr = FILEREADER_ATTR_USER;
+		f->_file = open_pgzf_reader(open_file_for_read(f->filename, NULL), 0, 4);
+		f->_read = read_pgzf4filereader;
+		f->_close = close_pgzf4filereader;
+	} else {
+		f->filename = strdup(filename);
+		f->file_attr = FILEREADER_ATTR_NORMAL;
+	}
+	return f->file_attr;
+}
+
+static inline int push_text_filereader(FileReader *fr, char *str, size_t len){
+	file_src_t *f;
+	UNUSED(len);
+	f = next_ref_filesrcv(fr->files);
+	f->_file = NULL;
+	f->_read = NULL;
+	f->_close = NULL;
+	f->filename = str;
+	f->file_attr = FILEREADER_ATTR_TEXT;
+	return f->file_attr;
+}
+
+static inline int push_user_filereader(FileReader *fr, void *_file, read_data_func _read, close_input_func _close){
+	file_src_t *f;
+	f = next_ref_filesrcv(fr->files);
+	f->_file = _file;
+	f->_read = _read;
+	f->_close = _close;
+	f->filename = NULL;
+	f->file_attr = FILEREADER_ATTR_USER;
+	return f->file_attr;
+}
+
+static inline void push_all_filereader(FileReader *fr, int nfile, char **filenames){
+	int i;
+	for(i=0;i<nfile;i++) push_filereader(fr, filenames[i]);
+}
+
+// asyn: asynchronous reading
+static inline FileReader* open_filereader(char *filename, int asyn){
+	FileReader *fr;
+	fr = init_filereader();
+	push_filereader(fr, filename);
+	if(asyn) beg_asyn_filereader(fr);
+	return fr;
+}
+
+static inline FileReader* string_filereader(char *str, int asyn){
+	int len;
+	FileReader *fr;
+	len = str? strlen(str) : 0;
+	fr = init_filereader();
+	push_text_filereader(fr, str, len);
+	if(asyn) beg_asyn_filereader(fr);
+	return fr;
+}
+
+static inline FileReader* open_all_filereader(int nfile, char **filenames, int asyn){
+	FileReader *fr;
+	fr = init_filereader();
+	push_all_filereader(fr, nfile, filenames);
+	if(asyn) beg_asyn_filereader(fr);
+	return fr;
+}
+
+#define close_filereader(fr) free_filereader(fr)
+
+static inline int asyn_readline_filereader(FileReader *fr, String *line){
+	char *buffer;
+	u8i i, nc;
+	int ret;
+	if(fr->rollback){
+		fr->rollback = 0;
+		return line->size + 1; // in case of end of file and not terminated by line_breaker, the return value is bigger by 1
+	} else if(fr->eof && fr->bufoff == fr->bufcnt[fr->ridx]){
+		return 0;
+	} else {
+		clear_string(line);
+		nc = fr->n_char;
+		while(1){
+			buffer = fr->buffer[fr->ridx];
+			ret = 0;
+			for(i=fr->bufoff;i<fr->bufcnt[fr->ridx];){
+				if(buffer[i++] == fr->line_breaker){
+					ret = 1;
+					break;
+				}
+			}
+			fr->n_char += i - fr->bufoff;
+			encap_string(line, i - fr->bufoff);
+			append_string(line, buffer + fr->bufoff, i - fr->bufoff - ret);
+			fr->bufoff = i;
+			if(ret){
+				return fr->n_char - nc;
+			} else if(fr->eof){
+				return fr->n_char - nc;
+			}
+			fr->bufoff = 0;
+			fr->bufcnt[fr->ridx] = 0;
+			while(fr->flag == 0){
+				nano_sleep(1);
+				if(fr->eof){
+					if(fr->flag) break;
+					else {
+						return fr->n_char - nc;
+					}
+				}
+			}
+			fr->flag = 0;
+			fr->ridx = !fr->ridx;
+		}
+		return 0;
+	}
+}
+
+static inline int directed_readline_filereader(FileReader *fr, String *line){
+	file_src_t *fc;
+	void *_file;
+	read_data_func _read;
+	close_input_func _close;
+	u8i i, nc;
+	int ch;
+	int ret;
+	if(fr->eof) return 0;
+	else if(fr->rollback){
+		fr->rollback = 0;
+		return line->size + 1; // in case of end of file and not terminated by line_breaker, the return value is bigger by 1
+	}
+	clear_string(line);
+	nc = fr->n_char;
+	while(fr->fidx < fr->files->size){
+		fc = ref_filesrcv(fr->files, fr->fidx);
+		_file = NULL;
+		_read = NULL;
+		_close = NULL;
+		if(fr->flag == 0){
+			switch(fc->file_attr){
+				case FILEREADER_ATTR_TEXT:
+					break;
+				case FILEREADER_ATTR_STDIN:
+					_file = fc->_file = stdin;
+					_read = fc->_read = _read_data_file;
+					_close = fc->_close = NULL;
+					break;
+				case FILEREADER_ATTR_PROC:
+					_file = fc->_file = popen(fc->filename, "r");
+					_read = fc->_read = _read_data_file;
+					_close = fc->_close = _close_input_proc;
+					break;
+				case FILEREADER_ATTR_USER:
+					_file = fc->_file;
+					_read = fc->_read;
+					_close = fc->_close;
+					break;
+				default:
+					_file = fc->_file = open_file_for_read(fc->filename, NULL);
+					_read = fc->_read = _read_data_file;
+					_close = fc->_close = _close_input_file;
+					break;
+			}
+			fr->flag = 1;
+			fr->bufoff = 0;
+			fr->bufcnt[0] = fr->bufcnt[1] = 0;
+		} else {
+			_file = fc->_file;
+			_read = fc->_read;
+			_close = fc->_close;
+		}
+		switch(fc->file_attr){
+			case FILEREADER_ATTR_TEXT:
+				ret = 0;
+				for(i=fr->bufoff;fc->filename[i];){
+					if(fc->filename[i++] == fr->line_breaker){
+						ret = 1;
+						break;
+					}
+				}
+				fr->n_char += i - fr->bufoff;
+				encap_string(line, i - fr->bufoff);
+				append_string(line, fc->filename + fr->bufoff, i - fr->bufoff - ret);
+				fr->bufoff = i;
+				if(ret){
+					break;
+				}
+				break;
+			case FILEREADER_ATTR_STDIN:
+				while((ch = fgetc(stdin)) != EOF){
+					fr->n_char ++;
+					if(ch == fr->line_breaker){
+						break;
+					}
+					add_char_string(line, ch);
+				}
+				break;
+			default:
+				while(1){
+					if(fr->bufoff >= fr->bufcnt[0]){
+						fr->bufoff = 0;
+						fr->bufcnt[0] = _read(_file, fr->buffer[0], fr->bufmax);
+						if(fr->bufcnt[0] == 0) break;
+					}
+					ret = 0;
+					for(i=fr->bufoff;i<fr->bufcnt[0];){
+						if(fr->buffer[0][i++] == fr->line_breaker){
+							ret = 1;
+							break;
+						}
+					}
+					fr->n_char += i - fr->bufoff;
+					encap_string(line, i - fr->bufoff);
+					append_string(line, fr->buffer[0] + fr->bufoff, i - fr->bufoff - ret);
+					fr->bufoff = i;
+					if(ret){
+						break;
+					}
+				}
+				break;
+		}
+		if(fr->n_char > nc){
+			return fr->n_char - nc;
+		} else {
+			if(_file && _close){
+				_close(_file);
+			}
+			fr->flag = 0;
+			fr->fidx ++;
+		}
+	}
+	fr->eof = 1;
+	return 0;
+}
+
+int readline_filereader(FileReader *fr){
+	int ret;
+	ret = ((fr)->pid? asyn_readline_filereader(fr, (fr)->line) : directed_readline_filereader(fr, (fr)->line));
+	if(ret > 0){
+		fr->n_line ++;
+	}
+	return ret;
+}
+
+static inline void rollback_filereader(FileReader *fr){
+	fr->rollback = 1;
+	fr->n_line --;
+}
+
+static inline int split_line_filereader(FileReader *fr, char delimiter){
+	VString *vs;
+	int i;
+	clear_VStrv(fr->tabs);
+	vs = next_ref_VStrv(fr->tabs);
+	vs->string = fr->line->string;
+	vs->size = 0;
+	for(i=0;i<fr->line->size;i++){
+		if(fr->line->string[i] == delimiter){
+			vs->size = fr->line->string + i - vs->string;
+			vs = next_ref_VStrv(fr->tabs);
+			vs->string = fr->line->string + i + 1;
+			vs->size = 0;
+		}
+	}
+	vs->size = fr->line->string + fr->line->size - vs->string;
+	return (int)fr->tabs->size;
+}
+
+static inline int readtable_filereader(FileReader *fr){
+	if(readline_filereader(fr) == 0) return -1;
+	return split_line_filereader(fr, fr->delimiter);
+}
+
+static inline int get_col_len(FileReader *fr, int col){
+	return fr->tabs->buffer[col].size;
+}
+
+static inline char* get_col_str(FileReader *fr, int col){
+	VString *vs;
+	vs = ref_VStrv(fr->tabs, col);
+	vs->string[vs->size] = '\0';
+	return vs->string;
+}
+
+static inline char* get_line_str(FileReader *fr){
+	int i;
+	for(i=0;i<fr->line->size;i++){
+		if(fr->line->string[i] == 0){
+			fr->line->string[i] = fr->delimiter;
+		}
+	}
+	return fr->line->string;
+}
+
+// @return FILEREADER_TYPE_NULL (end of files), _FASTA, _FASTQ, or _TEXT (cannot parse sequence type)
+static inline int readseq_filereader(FileReader *fr, BioSequence *seq){
+	int n, i;
+	do {
+		if((n = readline_filereader(fr)) == 0) return FILEREADER_TYPE_NULL;
+	} while(n == 0);
+	reset_biosequence(seq);
+	if(fr->line->string[0] == '>'){
+		if(seq->attr & BIOSEQ_ATTR_TAG){
+			for(i=1;i<fr->line->size;i++){
+				if(fr->line->string[i] == ' ' || fr->line->string[i] == '\t') break;
+			}
+			append_string(seq->tag, fr->line->string + 1, i - 1);
+			append_string(seq->dsc, fr->line->string + i, fr->line->size - i);
+		}
+		while((n = readline_filereader(fr))){
+			if(fr->line->string[0] == '>'){
+				rollback_filereader(fr);
+				break;
+			} else if(seq->attr & BIOSEQ_ATTR_SEQ){
+				append_string(seq->seq, fr->line->string, fr->line->size);
+			}
+		}
+		return FILEREADER_TYPE_FASTA;
+	} else if(fr->line->string[0] == '@'){
+		if(seq->attr & BIOSEQ_ATTR_TAG){
+			for(i=1;i<fr->line->size;i++){
+				if(fr->line->string[i] == ' ' || fr->line->string[i] == '\t') break;
+			}
+			append_string(seq->tag, fr->line->string + 1, i - 1);
+			append_string(seq->dsc, fr->line->string + i, fr->line->size - i);
+		}
+		if((n = readline_filereader(fr))){
+			if(seq->attr & BIOSEQ_ATTR_SEQ) append_string(seq->seq, fr->line->string, fr->line->size);
+		} else {
+			return FILEREADER_TYPE_FASTQ;
+		}
+		if((n = readline_filereader(fr))){
+			// expected '+'
+		} else {
+			return FILEREADER_TYPE_FASTQ;
+		}
+		if((n = readline_filereader(fr))){
+			if(seq->attr & BIOSEQ_ATTR_QLT) append_string(seq->qlt, fr->line->string, fr->line->size);
+		} else {
+			return FILEREADER_TYPE_FASTQ;
+		}
+		return FILEREADER_TYPE_FASTQ;
+	} else {
+		append_string(seq->dsc, fr->line->string, fr->line->size);
+		return FILEREADER_TYPE_TEXT;
+	}
+}
+
+#endif

filewriter.h

@@ -0,0 +1,205 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __FILEWRITER_RJ_H
+#define __FILEWRITER_RJ_H
+
+#include "mem_share.h"
+#include "thread.h"
+#include "pgzf.h"
+
+typedef size_t (*write_data_func)(void *obj, void *dat, size_t len);
+typedef void (*close_output_func)(void *obj);
+
+static inline size_t _write_data_file(void *obj, void *dat, size_t len){ return fwrite(dat, 1, len, (FILE*)obj); }
+static inline void _close_output_file(void *obj){ if(obj) fclose((FILE*)obj); }
+
+/**
+ * BufferedWriter
+ */
+typedef struct {
+	FILE *bios[2];
+	FILE *out;
+	void *_file;
+	write_data_func _write;
+	close_output_func _close;
+	int bidx;
+	size_t buf_size;
+	char *buffs[2];
+	size_t blens[2];
+	size_t nbytes;
+	pthread_mutex_t lock;
+	pthread_t pid;
+	int running, flush;
+} BufferedWriter;
+
+static inline void* _buffered_writer_thread_func(void *obj){
+	BufferedWriter *bw;
+	size_t bsize[2];
+	int bidx, lock;
+	bw = (BufferedWriter*)obj;
+	bw->running = 1;
+	bw->flush = 0;
+	bw->nbytes = 0;
+	while(bw->running){
+		bidx = bw->bidx;
+		bsize[0] = ftell(bw->bios[0]);
+		bsize[1] = ftell(bw->bios[1]);
+		if(bsize[bidx] >= bw->buf_size || (bsize[bidx] && bw->flush == 1)){
+			lock = 1;
+			pthread_mutex_lock(&bw->lock);
+		} else {
+			lock = 0;
+		}
+		if(bsize[!bidx]){
+			fflush(bw->bios[!bidx]);
+			bw->_write(bw->_file, bw->buffs[!bidx], bsize[!bidx]);
+			bw->nbytes += bsize[!bidx];
+			fseek(bw->bios[!bidx], 0, SEEK_SET);
+		}
+		if(lock){
+			bw->bidx = !bidx;
+			pthread_mutex_unlock(&bw->lock);
+		} else if(bsize[bidx]){
+			pthread_mutex_lock(&bw->lock);
+			bw->bidx = !bidx;
+			pthread_mutex_unlock(&bw->lock);
+		}
+		if(bw->flush && bsize[0] == 0 && bsize[1] == 0){
+			bw->flush = 2;
+			while(bw->flush == 2){
+				nano_sleep(1);
+			}
+			bw->flush = 0;
+		}
+		nano_sleep(10);
+	}
+	{
+		bsize[0] = ftell(bw->bios[0]);
+		bsize[1] = ftell(bw->bios[1]);
+		fflush(bw->bios[0]);
+		fflush(bw->bios[1]);
+		bidx = bw->bidx;
+		if(bsize[!bidx]){
+			bw->_write(bw->_file, bw->buffs[!bidx], bsize[!bidx]);
+			bw->nbytes += bsize[!bidx];
+		}
+		if(bsize[bidx]){
+			bw->_write(bw->_file, bw->buffs[bidx], bsize[bidx]);
+			bw->nbytes += bsize[bidx];
+		}
+	}
+	return NULL;
+}
+
+static inline BufferedWriter* open2_bufferedwriter(void *obj, write_data_func _write, close_output_func _close, size_t buf_size){
+	BufferedWriter *bw;
+	bw = malloc(sizeof(BufferedWriter));
+	bw->_file = obj;
+	bw->_write = _write;
+	bw->_close = _close;
+	bw->buffs[0] = NULL;
+	bw->buffs[1] = NULL;
+	bw->blens[0] = 0;
+	bw->blens[1] = 0;
+	bw->bios[0] = open_memstream(bw->buffs + 0, bw->blens + 0);
+	bw->bios[1] = open_memstream(bw->buffs + 1, bw->blens + 1);
+	bw->out = NULL;
+	bw->bidx = 0;
+	bw->buf_size = buf_size? buf_size : 4 * 1024;
+	bw->nbytes = 0;
+	bw->lock = (pthread_mutex_t)PTHREAD_MUTEX_INITIALIZER;
+	bw->running = 0;
+	bw->flush = 0;
+	if(pthread_create(&bw->pid, NULL, _buffered_writer_thread_func, bw) != 0){
+		fprintf(stderr, " -- Failed to create thread [%s] in %s -- %s:%d --\n", "_buffered_writer_thread_func", __FUNCTION__, __FILE__, __LINE__);
+		bw->pid = 0;
+	}
+	while(bw->running != 1){ nano_sleep(1); }
+	return bw;
+}
+
+static inline BufferedWriter* open_bufferedwriter(FILE *out, size_t buf_size){
+	return open2_bufferedwriter(out, _write_data_file, NULL, buf_size);
+}
+
+static inline BufferedWriter* zopen_bufferedwriter(FILE *out, size_t buf_size, int ncpu, int level){
+	PGZF *pz;
+	pz = open_pgzf_writer(out, buf_size, ncpu, level);
+	return open2_bufferedwriter(pz, write_pgzf4filewriter, close_pgzf4filewriter, pz->bufsize);
+}
+
+static inline int beg_bufferedwriter(BufferedWriter *bw){
+	if(bw->pid){
+		while(bw->flush){ nano_sleep(1); }
+		pthread_mutex_lock(&bw->lock);
+		bw->out = bw->bios[bw->bidx];
+		return 0;
+	} else {
+		bw->out = NULL;
+		return 1; // error
+	}
+}
+
+static inline int end_bufferedwriter(BufferedWriter *bw){
+	if(bw->pid){
+		pthread_mutex_unlock(&bw->lock);
+	}
+	bw->out = NULL;
+	return 0;
+}
+
+static inline size_t flush_bufferedwriter(BufferedWriter *bw){
+	size_t ret;
+	if(bw->pid){
+		pthread_mutex_unlock(&bw->lock);
+		while(bw->flush == 1){ nano_sleep(1); }
+		bw->flush = 1;
+		while(bw->flush == 1){
+			nano_sleep(1);
+		}
+		pthread_mutex_lock(&bw->lock);
+		bw->flush = 0;
+		bw->out = bw->bios[bw->bidx];
+		ret = bw->nbytes;
+	} else {
+		ret = 0;
+	}
+	return ret;
+}
+
+static inline size_t close_bufferedwriter(BufferedWriter *bw){
+	size_t ret;
+	if(bw->pid){
+		bw->running = 0;
+		pthread_join(bw->pid, NULL);
+	}
+	fclose(bw->bios[0]);
+	fclose(bw->bios[1]);
+	if(bw->buffs[0]) free(bw->buffs[0]);
+	if(bw->buffs[1]) free(bw->buffs[1]);
+	if(bw->_close){
+		bw->_close(bw->_file);
+	}
+	ret = bw->nbytes;
+	free(bw);
+	return ret;
+}
+
+#endif

general_graph.h

@@ -0,0 +1,387 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __GENERAL_GRAPH_RJ_H
+#define __GENERAL_GRAPH_RJ_H
+
+#include "list.h"
+#include "hashset.h"
+
+#define GEG_MAX_NODE	0xFFFFFFFFFFLLU
+#define GEG_MAX_EDGE_CNT	0x3FFFF
+#define GEG_MAX_EDGE_COV	0x7FFFF
+#define GEG_MAX_EDGE_OFF	0x7FFFFF
+#define GEG_MIN_EDGE_OFF	-0x7FFFFF
+
+typedef struct {
+	u8i node1:40, dir1:1, dir2:1, closed:2, cov:19, visit:1;
+	u8i node2:40; b8i off:24;
+} ge_edge_t;
+define_list(geedgev, ge_edge_t);
+
+static inline uint64_t _ge_edge_hashcode(ge_edge_t e){
+	const uint64_t m = 0xc6a4a7935bd1e995LLU;
+	const int r = 47;
+	uint64_t h = 1023 ^ (16 * m);
+	uint64_t k = (e.node1 << 1) | e.dir1;
+	k *= m;
+	k ^= k >> r;
+	k *= m;
+	h ^= k;
+	h *= m;
+	k = (e.node2 << 1) | e.dir2;
+	k *= m;
+	k ^= k >> r;
+	k *= m;
+	h ^= k;
+	h *= m;
+	h ^= h >> r;
+	h *= m;
+	h ^= h >> r;
+	return h;
+}
+#define GEEDGEHASH(idx) ((geedgev *)set->userdata)->buffer[idx]
+#define ge_edge_hashcode(E) _ge_edge_hashcode(GEEDGEHASH(E))
+#define ge_edge_hashequals(E1, E2) (GEEDGEHASH(E1).node1 == GEEDGEHASH(E2).node1 && GEEDGEHASH(E1).node2 == GEEDGEHASH(E2).node2 \
+			&& GEEDGEHASH(E1).dir1 == GEEDGEHASH(E2).dir1 && GEEDGEHASH(E1).dir2 == GEEDGEHASH(E2).dir2)
+define_hashset(geedgehash, u8i, ge_edge_hashcode, ge_edge_hashequals);
+
+typedef struct {
+	u8i idx:63, flg:1;
+	u8i next;
+} ge_edge_ref_t;
+define_list(geedgerefv, ge_edge_ref_t);
+
+typedef struct { uint64_t idx:46, cnt:18; } ge_ptr_ref_t;
+static const ge_ptr_ref_t GE_PTR_REF_NULL = (ge_ptr_ref_t){0, 0};
+define_list(geptrrefv, ge_ptr_ref_t);
+typedef struct { uint64_t idx:46, cnt:18; } ge_vec_ref_t;
+static const ge_vec_ref_t GE_VEC_REF_NULL = (ge_vec_ref_t){0, 0};
+define_list(gevecrefv, ge_vec_ref_t);
+
+typedef struct {
+	u8i closed:1, bt_visit:40, bt_dir:1, bt_idx:18, status:4;
+	u8i unvisit:18, aux:46;
+	ge_ptr_ref_t edges[2];
+} ge_node_t;
+define_list(genodev, ge_node_t);
+
+#define GEG_TRACE_MSG_ZERO	0
+#define GEG_TRACE_MSG_ONE	1
+#define GEG_TRACE_MSG_MORE	2
+#define GEG_TRACE_MSG_VISITED	3
+#define GEG_TRACE_MSG_UNDEF	4
+
+typedef void (*geg_clr_node_callback)(void *userdata);
+typedef void (*geg_add_node_callback)(void *userdata, u8i nidx);
+typedef void (*geg_del_node_callback)(void *userdata, u8i nidx);
+typedef void (*geg_clr_edge_callback)(void *userdata);
+typedef void (*geg_add_edge_callback)(void *userdata, u8i eidx);
+typedef void (*geg_del_edge_callback)(void *userdata, u8i eidx);
+
+#define define_simple_geg_callback(tag, node_tag, node_tag_t, edge_tag, edge_tag_t)	\
+void tag##nodeclr(void *aux){ clear_##node_tag((node_tag*)aux); }	\
+void tag##nodeadd(void *aux, u8i idx){	\
+	node_tag *nodes = (node_tag*)aux;	\
+	if(idx < nodes->size){	\
+		memset(ref_##node_tag(nodes, idx), 0, sizeof(node_tag_t));	\
+	} else {	\
+		memset(next_ref_##node_tag(nodes), 0, sizeof(node_tag_t));	\
+	}	\
+}	\
+void tag##nodedel(void *aux, u8i idx){ UNUSED(aux); UNUSED(idx); }	\
+void tag##edgeclr(void *aux){ clear_##edge_tag((edge_tag*)aux); }	\
+void tag##edgeadd(void *aux, u8i idx){	\
+	edge_tag *edges = (edge_tag*)aux;	\
+	if(idx < edges->size){	\
+		memset(ref_##edge_tag(edges, idx), 0, sizeof(edge_tag_t));	\
+	} else {	\
+		memset(next_ref_##edge_tag(edges), 0, sizeof(edge_tag_t));	\
+	}	\
+}	\
+void tag##edgedel(void *aux, u8i idx){ UNUSED(aux); UNUSED(idx); }	\
+static inline void tag##_set_callbacks_gegraph(GEGraph *g, node_tag *nodeaux, edge_tag *edgeaux){	\
+	set_callbacks_gegraph(g, (void*)nodeaux, (void*)edgeaux, tag##nodeclr, tag##nodeadd, tag##nodedel, tag##edgeclr, tag##edgeadd, tag##edgedel);	\
+}
+
+typedef struct {
+	genodev    *nodes;
+	geedgev    *edges;
+	geedgehash *ehash;
+	geedgerefv *erefs;
+	void       *nodeaux;
+	void       *edgeaux;
+	geg_clr_node_callback nodeclr;
+	geg_add_node_callback nodeadd;
+	geg_del_node_callback nodedel;
+	geg_clr_edge_callback edgeclr;
+	geg_add_edge_callback edgeadd;
+	geg_del_edge_callback edgedel;
+} GEGraph;
+
+static inline GEGraph* init_gegraph(){
+	GEGraph *g;
+	g = malloc(sizeof(GEGraph));
+	g->nodes = init_genodev(32);
+	g->edges = init_geedgev(32);
+	g->ehash = init_geedgehash(1023);
+	set_userdata_geedgehash(g->ehash, g->edges);
+	g->erefs = init_geedgerefv(32);
+	g->nodeaux  = NULL;
+	g->edgeaux  = NULL;
+	g->nodeclr  = NULL;
+	g->nodeadd  = NULL;
+	g->nodedel  = NULL;
+	g->edgeclr  = NULL;
+	g->edgeadd  = NULL;
+	g->edgedel  = NULL;
+	return g;
+}
+
+static inline void set_callbacks_gegraph(GEGraph *g, void *nodeaux, void *edgeaux, geg_clr_node_callback nodeclr, geg_add_node_callback nodeadd, geg_del_node_callback nodedel, geg_clr_edge_callback edgeclr, geg_add_edge_callback edgeadd, geg_del_edge_callback edgedel){
+	g->nodeaux  = nodeaux;
+	g->edgeaux  = edgeaux;
+	g->nodeclr  = nodeclr;
+	g->nodeadd  = nodeadd;
+	g->nodedel  = nodedel;
+	g->edgeclr  = edgeclr;
+	g->edgeadd  = edgeadd;
+	g->edgedel  = edgedel;
+}
+
+static inline void free_gegraph(GEGraph *g){
+	free_genodev(g->nodes);
+	free_geedgev(g->edges);
+	free_geedgehash(g->ehash);
+	free_geedgerefv(g->erefs);
+	free(g);
+}
+
+static inline void reset_gegraph(GEGraph *g){
+	clear_genodev(g->nodes);
+	if(g->nodeclr) g->nodeclr(g->nodeaux);
+	clear_geedgev(g->edges);
+	memset(next_ref_geedgev(g->edges), 0, sizeof(ge_edge_t));
+	if(g->edgeclr){
+		g->edgeclr(g->edgeaux);
+		g->edgeadd(g->edgeaux, 0);
+	}
+	clear_geedgehash(g->ehash);
+	clear_geedgerefv(g->erefs);
+	memset(next_ref_geedgerefv(g->erefs), 0, sizeof(ge_edge_ref_t));
+}
+
+static inline ge_node_t* add_node_gegraph(GEGraph *g){
+	ge_node_t *n;
+	n = next_ref_genodev(g->nodes);
+	memset(n, 0, sizeof(ge_node_t));
+	if(g->nodeadd) g->nodeadd(g->nodeaux, offset_genodev(g->nodes, n));
+	return n;
+}
+
+static inline ge_edge_t* prepare_edge_gegraph(GEGraph *g, u8i node1, int dir1, u8i node2, int dir2, int *exists){
+	ge_node_t *n;
+	ge_edge_t *e;
+	ge_edge_ref_t *f;
+	u8i *u;
+	e = ref_geedgev(g->edges, 0);
+	if(node1 <= node2){
+		e->node1  = node1;
+		e->dir1   = dir1;
+		e->node2  = node2;
+		e->dir2   = dir2;
+	} else {
+		e->node1  = node2;
+		e->dir1   = !dir2;
+		e->node2  = node1;
+		e->dir2   = !dir1;
+	}
+	e->cov    = 0;
+	e->off    = 0;
+	e->visit = 0;
+	e->closed = 0;
+	u = prepare_geedgehash(g->ehash, 0, exists);
+	if(*exists){
+		return g->edges->buffer + *u;
+	} else {
+		*u = g->edges->size;
+		e = next_ref_geedgev(g->edges);
+		*e = g->edges->buffer[0];
+		n = g->nodes->buffer + e->node1;
+		f = next_ref_geedgerefv(g->erefs);
+		f->idx = *u;
+		f->flg = 0;
+		f->next = n->edges[e->dir1].idx;
+		n->edges[e->dir1].idx = g->erefs->size - 1;
+		n->edges[e->dir1].cnt ++;
+		n = g->nodes->buffer + e->node2;
+		f = next_ref_geedgerefv(g->erefs);
+		f->idx = *u;
+		f->flg = 1;
+		f->next = n->edges[!e->dir2].idx;
+		n->edges[!e->dir2].idx = g->erefs->size - 1;
+		n->edges[!e->dir2].cnt ++;
+		if(g->edgeadd) g->edgeadd(g->edgeaux, offset_geedgev(g->edges, e));
+		return e;
+	}
+}
+
+static inline void cut_edge_core_gegraph(GEGraph *g, ge_edge_t *e, int closed_val){
+	if(e->closed) return;
+	e->closed = closed_val;
+	ref_genodev(g->nodes, e->node1)->edges[e->dir1].cnt --;
+	ref_genodev(g->nodes, e->node2)->edges[!e->dir2].cnt --;
+	if(g->edgedel) g->edgedel(g->edgeaux, offset_geedgev(g->edges, e));
+}
+
+#define cut_edge_gegraph(g, e) cut_edge_core_gegraph(g, e, 1)
+
+static inline void revive_edge_gegraph(GEGraph *g, ge_edge_t *e){
+	if(e->closed == 0) return;
+	e->closed = 0;
+	ref_genodev(g->nodes, e->node1)->edges[e->dir1].cnt ++;
+	ref_genodev(g->nodes, e->node2)->edges[!e->dir2].cnt ++;
+	if(g->edgedel) g->edgeadd(g->edgeaux, offset_geedgev(g->edges, e));
+}
+
+static inline ge_edge_ref_t* single_edge_gegraph(GEGraph *g, ge_node_t *n, int dir, int *info){
+	ge_edge_ref_t *f, *ret;
+	uint64_t idx;
+	ret = NULL;
+	if(info){
+		*info = GEG_TRACE_MSG_ZERO;
+		if(n->edges[dir].cnt == 0) return NULL;
+		idx = n->edges[dir].idx;
+		while(idx){
+			f = ref_geedgerefv(g->erefs, idx);
+			idx = f->next;
+			if(g->edges->buffer[f->idx].closed) continue;
+			if(ret){ *info = GEG_TRACE_MSG_MORE; return NULL; }
+			else { *info = GEG_TRACE_MSG_ONE; ret = f; }
+		}
+	} else {
+		if(n->edges[dir].cnt == 0) return NULL;
+		idx = n->edges[dir].idx;
+		while(idx){
+			f = ref_geedgerefv(g->erefs, idx);
+			idx = f->next;
+			if(g->edges->buffer[f->idx].closed) continue;
+			if(ret){ return NULL; }
+			else { ret = f; }
+		}
+	}
+	return ret;
+}
+
+#define count_edges_gegraph(g, n, dir) (n)->edges[dir].cnt
+
+// dir = 2 means either strand
+static inline ge_edge_ref_t* edge_node2node_gegraph(GEGraph *g, u8i node1, int dir1, u8i node2, int dir2){
+	ge_node_t *n;
+	ge_edge_ref_t *f;
+	ge_edge_t *e;
+	uint64_t idx;
+	int dire;
+	n = ref_genodev(g->nodes, node1);
+	if(dir1 > 1){
+		dir1 = 0; dire = 2;
+	} else {
+		dire = dir1 + 1;
+	}
+	while(dir1 < dire){
+		idx = n->edges[dir1].idx;
+		while(idx){
+			f = ref_geedgerefv(g->erefs, idx);
+			idx = f->next;
+			e = ref_geedgev(g->edges, f->idx);
+			if(f->flg){
+				if(e->node1 == node2 && (dir2 > 1? 1 : (dir2 == (!e->dir1)))) return f;
+			} else {
+				if(e->node2 == node2 && (dir2 > 1? 1 : (dir2 == e->dir2))) return f;
+			}
+		}
+		dir1 ++;
+	}
+	return NULL;
+}
+
+static inline void del_node_edges_gegraph(GEGraph *g, ge_node_t *n, int closed_val){
+	ge_edge_ref_t *f;
+	ge_edge_t *e;
+	uint64_t idx;
+	uint32_t k;
+	for(k=0;k<2;k++){
+		idx = n->edges[k].idx;
+		while(idx){
+			f = ref_geedgerefv(g->erefs, idx);
+			idx = f->next;
+			e = g->edges->buffer + f->idx;
+			cut_edge_core_gegraph(g, e, closed_val);
+		}
+	}
+}
+
+static inline void del_node_gegraph(GEGraph *g, ge_node_t *n){
+	del_node_edges_gegraph(g, n, 1);
+	n->closed = 1;
+	if(g->nodedel) g->nodeadd(g->nodeaux, offset_genodev(g->nodes, n));
+}
+
+#define geg_beg_iter_edges(g, n, dir, f, e)	\
+{	\
+	u8i _geg_iter_idx;	\
+	_geg_iter_idx = (n)->edges[dir].idx;	\
+	while(_geg_iter_idx){	\
+		(f) = ref_geedgerefv((g)->erefs, _geg_iter_idx);	\
+		_geg_iter_idx = (f)->next;	\
+		(e) = (g)->edges->buffer + (f)->idx
+
+#define geg_end_iter_edges()	\
+	}	\
+}
+
+static inline void print_dot_gegraph(GEGraph *g, FILE *out){
+	static const char *colors[2][2] = {{"blue", "green"}, {"red", "gray"}};
+	ge_node_t *n;
+	ge_edge_t *e;
+	u8i i;
+	fprintf(out, "digraph {\n");
+	for(i=0;i<g->nodes->size;i++){
+		n = ref_genodev(g->nodes, i);
+		if(n->closed) continue;
+		fprintf(out, " N%llu\n", i);
+	}
+	for(i=1;i<g->edges->size;i++){
+		e = ref_geedgev(g->edges, i);
+		if(e->closed) continue;
+		fprintf(out, " N%llu -> N%llu [label=\"%c%c:%d:%d\" color=%s]\n", (u8i)e->node1, (u8i)e->node2, "+-"[e->dir1], "+-"[e->dir2], e->cov, e->off, colors[e->dir1][e->dir2]);
+		fprintf(out, " N%llu -> N%llu [label=\"%c%c:%d:%d\" color=%s]\n", (u8i)e->node2, (u8i)e->node1, "-+"[e->dir2], "-+"[e->dir1], e->cov, e->off, colors[!e->dir2][!e->dir1]);
+	}
+	fprintf(out, "}\n");
+	fflush(out);
+}
+
+static inline void fprint_dot_gegraph(GEGraph *g, char *prefix, char *suffix){
+	FILE *out;
+	out = open_file_for_write(prefix, suffix, 1);
+	print_dot_gegraph(g, out);
+	fclose(out);
+}
+
+#endif

hashset.h

@@ -0,0 +1,614 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __HASH_SET_RJ
+#define __HASH_SET_RJ
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <math.h>
+#include "mem_share.h"
+#include "bitvec.h"
+
+static const uint64_t sys_prime_list[61] = {
+	0x7LLU, 0xfLLU, 0x1fLLU, 0x43LLU, 0x89LLU,
+	0x115LLU, 0x22dLLU, 0x45dLLU, 0x8bdLLU, 0x1181LLU,
+	0x2303LLU, 0x4609LLU, 0x8c17LLU, 0x1183dLLU, 0x2307bLLU,
+	0x460fdLLU, 0x8c201LLU, 0x118411LLU, 0x230833LLU, 0x461069LLU,
+	0x8c20e1LLU, 0x11841cbLLU, 0x2308397LLU, 0x461075bLLU, 0x8c20ecbLLU,
+	0x11841da5LLU, 0x23083b61LLU, 0x461076c7LLU, 0x8c20ed91LLU, 0x11841db31LLU,
+	0x23083b673LLU, 0x461076d1bLLU, 0x8c20eda41LLU, 0x11841db48dLLU, 0x23083b6937LLU,
+	0x461076d27fLLU, 0x8c20eda50dLLU, 0x11841db4a59LLU, 0x23083b694ebLLU, 0x461076d29f1LLU,
+	0x8c20eda5441LLU, 0x11841db4a887LLU, 0x23083b69511fLLU, 0x461076d2a2c1LLU, 0x8c20eda54591LLU,
+	0x11841db4a8b55LLU, 0x23083b69516c1LLU, 0x461076d2a2da5LLU, 0x8c20eda545b55LLU, 0x11841db4a8b6b5LLU,
+	0x23083b69516d91LLU, 0x461076d2a2db3bLLU, 0x8c20eda545b69dLLU, 0x11841db4a8b6d5dLLU, 0x23083b69516daf5LLU,
+	0x461076d2a2db5edLLU, 0x8c20eda545b6c5fLLU, 0x11841db4a8b6d8ebLLU, 0x23083b69516db1ffLLU, 0x461076d2a2db643fLLU,
+	0x8c20eda545b6c8f3LLU
+};
+
+static inline uint64_t _rj_hashset_find_prime(uint64_t n){
+	uint32_t i;
+	i = 0;
+	while(i < 60 && n > sys_prime_list[i]) i ++;
+	return sys_prime_list[i];
+}
+
+#define init_hashset_macro(hash_type, hash_ele_type) \
+typedef struct { hash_ele_type *array;  BitVec *ones, *dels; size_t e_size; size_t ocp; size_t size; size_t count; size_t max; float load_factor; size_t iter_ptr; void *userdata; } hash_type; \
+static inline size_t hash_type##_obj_desc_cnt(void *obj, int idx){	\
+	hash_type *set;	\
+	set = (hash_type*)obj;	\
+	if(set->dels){	\
+		switch(idx){	\
+			case 0: return ((hash_type*)obj)->size * sizeof(hash_ele_type);	\
+			default: return 1;	\
+		}	\
+	} else {	\
+		switch(idx){	\
+			case 0: return ((hash_type*)obj)->count * sizeof(hash_ele_type);	\
+			case 1: return 1;	\
+			default: return 0;	\
+		}	\
+	}	\
+}	\
+static const obj_desc_t hash_type##_obj_desc = {TOSTR(_hashset_##hash_type), sizeof(hash_type), 3, {1, 1, 1}, {offsetof(hash_type, array), offsetof(hash_type, ones), offsetof(hash_type, dels)}, {(obj_desc_t*)&OBJ_DESC_DATA, (obj_desc_t*)&bitvec_obj_desc, (obj_desc_t*)&bitvec_obj_desc}, hash_type##_obj_desc_cnt, NULL};	\
+static inline int hash_type##_is_prime(uint64_t num){                          \
+	uint64_t i, max;                                                           \
+	if(num < 4) return 1;                                                      \
+	if(num % 2 == 0) return 0;                                                 \
+	max = (uint64_t)sqrt((double)num);                                         \
+	for(i=3;i<max;i+=2){ if(num % i == 0) return 0; }                          \
+	return 1;                                                                  \
+}                                                                              \
+static inline uint64_t hash_type##_find_next_prime(uint64_t num){              \
+	if(num % 2 == 0) num ++;                                                   \
+	while(1){ if(hash_type##_is_prime(num)) return num; num += 2; }            \
+}                                                                              \
+static inline hash_type* init2_##hash_type(uint32_t size, float factor){       \
+	hash_type *set;                                                            \
+	set = (hash_type*)calloc(1, sizeof(hash_type));                            \
+	set->e_size = sizeof(hash_ele_type);                                       \
+	set->size   = _rj_hashset_find_prime(size);                                \
+	set->count  = 0;                                                           \
+	set->ocp    = 0;                                                           \
+	set->load_factor = factor;                                                 \
+	set->max    = set->size * set->load_factor;                                \
+	set->iter_ptr    = 0;                                                      \
+	set->array       = calloc(set->size, set->e_size);                         \
+	set->ones = init_bitvec(set->size);                                        \
+	set->dels = init_bitvec(set->size);                                        \
+	set->userdata = NULL;                                                      \
+	return set;                                                                \
+}                                                                              \
+static inline void set_userdata_##hash_type(hash_type *set, void *userdata){ set->userdata = userdata; } \
+static inline hash_type* init_##hash_type(uint32_t size){ return init2_##hash_type(size, 0.67f); }
+
+#define get_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal, hash_val_type, hash_ele2val) \
+static inline hash_ele_type* get_##hash_type(hash_type *set, hash_key_type key){\
+	hash_ele_type *e;                                                          \
+	size_t hc, hi;                                                             \
+	hc = hash_key_code(key) % set->size;                                       \
+	if(set->dels){	\
+		while(1){                                                                  \
+			if(get_bitvec(set->ones, hc) == 0){	\
+				return NULL;	\
+			} else if(get_bitvec(set->dels, hc)){	\
+			} else {	\
+				e = ((hash_ele_type*)set->array) + hc;                             \
+				if(hash_key_equal((key), (*e))) return e;                          \
+			}	\
+			hc = (hc + 1) % set->size;                                             \
+		}                                                                          \
+	} else {	\
+		hi = MAX_U8;	\
+		while(1){	\
+			if(get_bitvec(set->ones, hc)){	\
+				if(hi == MAX_U8){	\
+					hi = rank_bitvec(set->ones, hc);	\
+				}	\
+				e = ((hash_ele_type*)set->array) + hi;	\
+				if(hash_key_equal((key), (*e))) return e;	\
+			} else {	\
+				return NULL;	\
+			}	\
+			hc ++; \
+			hi ++;	\
+		}	\
+	}	\
+	return NULL;                                                               \
+}                                                                              \
+static inline size_t offset_##hash_type(hash_type *set, hash_ele_type *ptr){   \
+	return ptr - set->array;                                                   \
+}	\
+static inline hash_ele_type* ref_##hash_type(hash_type *set, size_t off){ return set->array + off; }	\
+static inline hash_val_type getval_##hash_type(hash_type *set, hash_key_type key){	\
+	hash_ele_type *e;	\
+	e = get_##hash_type(set, key);	\
+	return hash_ele2val(e);	\
+}
+
+#define prepare_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal) \
+static inline void encap_##hash_type(hash_type *set, size_t num);              \
+static inline hash_ele_type* prepare_##hash_type(hash_type *set, hash_key_type key, int *exists){\
+	hash_ele_type *e;                                                          \
+	size_t hc, d;                                                              \
+	if(set->dels == NULL){ *exists = 0;  return NULL; }	\
+	encap_##hash_type(set, 1);                                                 \
+	hc = hash_key_code((key)) % set->size;                                     \
+	d = set->size;                                                             \
+	while(1){	\
+		if(get_bitvec(set->ones, hc) == 0){	\
+			if(d == set->size){	\
+				one_bitvec(set->ones, hc);	\
+				set->ocp ++;	\
+			} else {	\
+				hc = d;	\
+				zero_bitvec(set->dels, hc);	\
+			}	\
+			if(exists) *exists = 0;	\
+			set->count ++;	\
+			e = ((hash_ele_type*)set->array) + hc;	\
+			return e;	\
+		} else if(get_bitvec(set->dels, hc)){	\
+			if(d == set->size) d = hc;	\
+		} else {	\
+			e = ((hash_ele_type*)set->array) + hc;	\
+			if(hash_key_equal((key), (*e))){	\
+				if(exists) *exists = 1;	\
+				return e;	\
+			}	\
+		}	\
+		hc = (hc + 1) % set->size;	\
+	}	\
+	return NULL;                                                               \
+}
+
+#define exists_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal) \
+static inline int exists_##hash_type(hash_type *set, hash_key_type key){       \
+	return get_##hash_type(set, key) != NULL;	\
+}
+
+#define add_hashset_macro(hash_type, hash_ele_type, hash_code_macro, hash_equal_macro) \
+static inline hash_ele_type* add_##hash_type(hash_type *set, hash_ele_type ele){       \
+	hash_ele_type *e;                                                          \
+	size_t d, hc;                                                              \
+	if(set->dels == NULL) return NULL;	\
+	hc = hash_code_macro(ele) % set->size;                                     \
+	d  = set->size;                                                            \
+	do {	\
+		if(get_bitvec(set->ones, hc) == 0){	\
+			if(d == set->size){	\
+				one_bitvec(set->ones, hc);	\
+				set->ocp ++;	\
+			} else {	\
+				hc = d;	\
+				zero_bitvec(set->dels, hc);	\
+			}	\
+			set->count ++;	\
+			e = ((hash_ele_type*)set->array) + hc;	\
+			*e = ele;	\
+			return e;	\
+		} else if(get_bitvec(set->dels, hc)){	\
+			if(d == set->size) d = hc;	\
+		} else {	\
+			e = ((hash_ele_type*)set->array) + hc;	\
+			if(hash_equal_macro((ele), (*e))){	\
+				*e = ele;	\
+				return e;	\
+			}	\
+		}	\
+		hc = (hc + 1) % set->size;	\
+	} while(1);	\
+	return NULL;                                                                  \
+}
+
+#define put_hashset_macro(hash_type, hash_ele_type) \
+static inline hash_ele_type* put_##hash_type(hash_type *set, hash_ele_type ele){         \
+	encap_##hash_type(set, 1);                                                 \
+	return add_##hash_type(set, ele);                                          \
+}
+
+#define remove_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal) \
+static inline int delete_##hash_type(hash_type *set, hash_ele_type *ele){	\
+	size_t hc;	\
+	if(set->dels == NULL) return 0;	\
+	hc = offset_##hash_type(set, ele);	\
+	if(get_bitvec(set->ones, (hc + 1) % set->size) == 0){	\
+		zero_bitvec(set->ones, hc);	\
+		set->ocp --;	\
+	} else {	\
+		one_bitvec(set->dels, hc);	\
+	}	\
+	set->count --;	\
+	return 1;	\
+}	\
+	\
+static inline int remove_##hash_type(hash_type *set, hash_key_type key){       \
+	hash_ele_type *e;                                                          \
+	size_t hc;                                                                 \
+	if(set->dels == NULL) return 0;                                           \
+	hc = hash_key_code(key) % set->size;                                       \
+	while(1){	\
+		if(get_bitvec(set->ones, hc) == 0){	\
+			return 0;	\
+		} else if(get_bitvec(set->dels, hc)){	\
+		} else {	\
+			e = ((hash_ele_type*)set->array) + hc;	\
+			if(hash_key_equal((key), (*e))){	\
+				if(get_bitvec(set->ones, (hc + 1) % set->size) == 0){	\
+					zero_bitvec(set->ones, hc);	\
+					set->ocp --;	\
+				} else {	\
+					one_bitvec(set->dels, hc);	\
+				}	\
+				set->count --;	\
+				return 1;	\
+			}	\
+		}	\
+		hc = (hc + 1) % set->size;	\
+	}	\
+	return 0;                                                                  \
+}
+
+#define reset_iter_hashset_macro(hash_type) static inline void reset_iter_##hash_type(hash_type *set){ set->iter_ptr = 0; }
+
+#define ref_iter_hashset_macro(hash_type, hash_ele_type) \
+static inline hash_ele_type* ref_iter2_##hash_type(hash_type *set, size_t *iter_ptr){             \
+	if(set->dels){	\
+		while(((*iter_ptr) = next_one_bitvec(set->ones, (*iter_ptr))) < set->size){	\
+			if(get_bitvec(set->dels, (*iter_ptr))){	\
+				(*iter_ptr) ++;	\
+			} else {	\
+				return (((hash_ele_type*)set->array) + (*iter_ptr)++);	\
+			}	\
+		}	\
+	} else {	\
+		while((*iter_ptr) < set->count){                                          \
+				return (((hash_ele_type*)set->array) + (*iter_ptr)++);           \
+		}                                                                          \
+	}	\
+	return NULL;                                                               \
+}	\
+static inline hash_ele_type* ref_iter_##hash_type(hash_type *set){             \
+	return ref_iter2_##hash_type(set, &(set->iter_ptr));	\
+}
+
+#define count_hashset_macro(hash_type) static inline int64_t count_##hash_type(hash_type *set){ return set->count; }
+
+#define freeze_hashset_macro(hash_type, hash_ele_type, hash_code_macro)	\
+static inline int freeze_##hash_type(hash_type *set, float load_factor){	\
+	size_t *hvs, i, j, sz;	\
+	if(set->dels == NULL) return 0;	\
+	if(load_factor == 0) load_factor = set->load_factor;	\
+	sz = set->count / load_factor;	\
+	sz = _rj_hashset_find_prime(sz);	\
+	for(i=j=0;(i=next_one_bitvec(set->ones, i))<set->size;i++){	\
+		if(get_bitvec(set->dels, i)) continue;	\
+		if(j < i){	\
+			set->array[j] = set->array[i];	\
+		}	\
+		j ++;	\
+	}	\
+	free_bitvec(set->ones);	\
+	set->ones = NULL;	\
+	free_bitvec(set->dels);	\
+	set->dels = NULL;	\
+	set->size = sz;	\
+	set->load_factor = load_factor;	\
+	set->ocp = set->count;	\
+	set->array = realloc(set->array, (set->count + 1) * sizeof(hash_ele_type));	\
+	memset(set->array + set->count, 0, sizeof(hash_ele_type));	\
+	hvs = malloc(set->count * sizeof(size_t));	\
+	for(i=0;i<set->count;i++){	\
+		hvs[i] = hash_code_macro(set->array[i]) % sz;	\
+	}	\
+	sort_array_adv(set->count, hvs[a] > hvs[b], swap_var(hvs[a], hvs[b]); swap_var(set->array[a], set->array[b]));	\
+	for(i=j=0;i<set->count;i++){	\
+		if(j < hvs[i]) j = hvs[i];	\
+		j ++;	\
+	}	\
+	if(j < sz) j = sz;	\
+	set->ones = init_bitvec(j + 1);	\
+	for(i=j=0;i<set->count;i++){	\
+		if(j < hvs[i]) j = hvs[i];	\
+		one_bitvec(set->ones, j);	\
+		j ++;	\
+	}	\
+	free(hvs);	\
+	index_bitvec(set->ones);	\
+	return 1;	\
+}
+
+#define clear_hashset_macro(hash_type) \
+static inline void clear_##hash_type(hash_type *set){                          \
+	if(set->dels == NULL){	\
+		return;	\
+	}	\
+	zeros_bitvec(set->ones);	\
+	zeros_bitvec(set->dels);	\
+	set->count = 0;                                                            \
+	set->ocp   = 0;                                                            \
+	set->iter_ptr = 0;                                                         \
+}
+
+#define free_hashset_macro(hash_type) \
+static inline void free_##hash_type(hash_type *set){                           \
+	free(set->array);                                                          \
+	if(set->ones) free_bitvec(set->ones);                                      \
+	if(set->dels) free_bitvec(set->dels);                                      \
+	free(set);                                                                 \
+}
+
+#define encap_hashset_macro(hash_type, hash_ele_type, hash_code_macro) \
+static inline void encap_##hash_type(hash_type *set, size_t num){             \
+	BitVec *ones, *dels;	\
+	size_t i, n, hc;                                                  \
+	hash_ele_type key;                                                        \
+	if(set->dels == NULL) return;	\
+	if(set->ocp + num <= set->max) return;                                    \
+	n = set->size;                                                            \
+	do{ n = _rj_hashset_find_prime(n * 2); } while(n * set->load_factor < set->count + num);    \
+	set->array = realloc(set->array, n * set->e_size);                        \
+	if(set->array == NULL){                                                   \
+		fprintf(stderr, "-- Out of memory --\n");                             \
+		print_backtrace(stderr, 20);                                          \
+		exit(1);                                                              \
+	}                                                                         \
+	ones = init_bitvec(n);	\
+	dels = init_bitvec(n);	\
+	set->ocp  = set->count;                                                   \
+	set->max = n * set->load_factor;                                          \
+	for(i=0;(i=next_one_bitvec(set->ones, i))<set->size;i++){	\
+		if(get_bitvec(set->dels, i)) continue;	\
+		key = ((hash_ele_type*)set->array)[i];	\
+		one_bitvec(set->dels, i);	\
+		while(1){	\
+			hc = hash_code_macro(key) % n;	\
+			while(get_bitvec(ones, hc)){	\
+				hc = (hc + 1) % n;	\
+			}	\
+			one_bitvec(ones, hc);	\
+			if(hc < set->size && get_bitvec(set->ones, hc) && get_bitvec(set->dels, hc) == 0){	\
+				swap_var(key, ((hash_ele_type*)set->array)[hc]);	\
+				one_bitvec(set->dels, hc);	\
+			} else {	\
+				((hash_ele_type*)set->array)[hc] = key;	\
+				break;	\
+			}	\
+		}	\
+	}	\
+	swap_var(ones, set->ones);	\
+	swap_var(dels, set->dels);	\
+	set->size = n;	\
+	free_bitvec(ones);	\
+	free_bitvec(dels);	\
+}                                                                             \
+static inline size_t offsetof_##hash_type(hash_type *set, hash_ele_type *ptr){ return ptr - set->array; }	\
+
+
+#define ITSELF(E) (E)
+#define NUM_EQUALS(E1, E2) ((E1) == (E2))
+
+#define define_hashtable(hash_type, hash_ele_type, hash_code_macro, hash_equal_macro, hash_key_type, hash_key_code, hash_key_equal, hash_val_type, hash_ele2val)    \
+	init_hashset_macro(hash_type, hash_ele_type);                              \
+	get_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal, hash_val_type, hash_ele2val);    \
+	prepare_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal);    \
+	exists_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal);    \
+	add_hashset_macro(hash_type, hash_ele_type, hash_code_macro, hash_equal_macro);    \
+	put_hashset_macro(hash_type, hash_ele_type);                               \
+	remove_hashset_macro(hash_type, hash_ele_type, hash_key_type, hash_key_code, hash_key_equal);    \
+	ref_iter_hashset_macro(hash_type, hash_ele_type);                          \
+	reset_iter_hashset_macro(hash_type);                                       \
+	count_hashset_macro(hash_type);                                            \
+	clear_hashset_macro(hash_type);                                            \
+	freeze_hashset_macro(hash_type, hash_ele_type, hash_code_macro);           \
+	free_hashset_macro(hash_type);                                             \
+	encap_hashset_macro(hash_type, hash_ele_type, hash_code_macro);
+
+#define define_hashset(hash_type, hash_ele_type, hash_code_macro, hash_equal_macro)  define_hashtable(hash_type, hash_ele_type, hash_code_macro, hash_equal_macro, hash_ele_type, hash_code_macro, hash_equal_macro, hash_ele_type*, ITSELF)
+
+/* ------------------ Useful functions ------------------------------------- */
+
+static inline uint32_t __lh3_Jenkins_hash_int(uint32_t key){
+	key += (key << 12);
+	key ^= (key >> 22);
+	key += (key << 4);
+	key ^= (key >> 9);
+	key += (key << 10);
+	key ^= (key >> 2);
+	key += (key << 7);
+	key ^= (key >> 12);
+	return key;
+}
+
+static inline uint64_t __lh3_Jenkins_hash_64(uint64_t key){
+	key += ~(key << 32);
+	key ^= (key >> 22);
+	key += ~(key << 13);
+	key ^= (key >> 8);
+	key += (key << 3);
+	key ^= (key >> 15);
+	key += ~(key << 27);
+	key ^= (key >> 31);
+	return key;
+}
+
+static inline uint32_t jenkins_one_at_a_time_hash(char *key, size_t len){
+	uint32_t hash, i;
+	for(hash = i = 0; i < len; ++i){
+		hash += key[i];
+		hash += (hash << 10);
+		hash ^= (hash >> 6);
+	}
+	hash += (hash << 3);
+	hash ^= (hash >> 11);
+	hash += (hash << 15);
+	return hash;
+}
+
+static inline u8i invertible_hashcode(u8i x, int p){
+	u8i m;
+	m = 0xFFFFFFFFFFFFFFFFLLU >> (64 - p);
+	x = ((~x) + (x << 21)) & m;
+	x = x ^ (x >> 24);
+	x = (x + (x << 3) + (x << 8)) & m;
+	x = x ^ (x >> 14);
+	x = (x + (x << 2) + (x << 4)) & m;
+	x = x ^ (x >> 28);
+	x = (x + (x << 31)) & m;
+	return x;
+}
+
+static inline uint64_t hash64shift(uint64_t key){
+	key = (~key) + (key << 21); // key = (key << 21) - key - 1;
+	key = key ^ (key >> 24);
+	key = (key + (key << 3)) + (key << 8); // key * 265
+	key = key ^ (key >> 14);
+	key = (key + (key << 2)) + (key << 4); // key * 21
+	key = key ^ (key >> 28);
+	key = key + (key << 31);
+	return key;
+}
+
+
+static inline uint64_t MurmurHash64A(const void * key, int len, uint32_t seed){
+	const uint64_t m = 0xc6a4a7935bd1e995LLU;
+	const int r = 47;
+
+	uint64_t h = seed ^ (len * m);
+
+	const uint64_t * data = (const uint64_t *)key;
+	const uint64_t * end = data + (len/8);
+
+	while(data != end){
+		uint64_t k = *data++;
+
+		k *= m;
+		k ^= k >> r;
+		k *= m;
+
+		h ^= k;
+		h *= m;
+	}
+
+	const unsigned char * data2 = (const unsigned char*)data;
+
+	switch(len & 7){
+	case 7: h ^= ((uint64_t)data2[6]) << 48;
+	case 6: h ^= ((uint64_t)data2[5]) << 40;
+	case 5: h ^= ((uint64_t)data2[4]) << 32;
+	case 4: h ^= ((uint64_t)data2[3]) << 24;
+	case 3: h ^= ((uint64_t)data2[2]) << 16;
+	case 2: h ^= ((uint64_t)data2[1]) << 8;
+	case 1: h ^= ((uint64_t)data2[0]);
+	        h *= m;
+	};
+
+	h ^= h >> r;
+	h *= m;
+	h ^= h >> r;
+
+	return h;
+}
+
+#define u32hashcode(key) __lh3_Jenkins_hash_int(key)
+#define u64hashcode(key) __lh3_Jenkins_hash_64(key)
+
+static inline uint32_t __string_hashcode(const char *s){
+	uint32_t h = *s;
+	if (h) for (++s ; *s; ++s) h = (h << 5) - h + *s;
+	return h;
+}
+
+#define u32hash_code(e) u32hashcode(e)
+#define u64hash_code(e) u64hashcode(e)
+#define uxxhash_equals(e1, e2) ((e1) == (e2))
+define_hashset(u32hash, uint32_t, u32hash_code, uxxhash_equals);
+define_hashset(u64hash, uint64_t, u64hash_code, uxxhash_equals);
+
+#define i32hash_code(e) u32hashcode((uint32_t)(e))
+#define i32hash_equals(e1, e2) ((e1) == (e2))
+define_hashset(i32hash, int, i32hash_code, i32hash_equals);
+
+#define chash_code(e) __string_hashcode(e)
+#define chash_equals(e1, e2) (strcmp(e1, e2) == 0)
+define_hashset(chash, char*, chash_code, chash_equals);
+
+#define KV_HASH_GET_VAL(e) (e)? (e)->val : ((typeof(e->val))MAX_U8)
+
+typedef struct { u4i key, val; } uuhash_t;
+#define uuhash_code(e) u32hashcode((e).key)
+#define uuhash_equals(e1, e2) ((e1).key == (e2).key)
+#define uuhash_key_equals(e1, e2) ((e1) == (e2).key)
+define_hashtable(uuhash, uuhash_t, uuhash_code, uuhash_equals, u4i, u32hashcode, uuhash_key_equals, u4i, KV_HASH_GET_VAL);
+
+typedef struct { u4i key; int val; } uihash_t;
+#define uihashcode(E) u32hashcode((E).key)
+#define uihashequals(E1, E2) (E1).key == (E2).key
+#define uihashkeyequals(E1, E2) (E1) == (E2).key
+define_hashtable(uihash, uihash_t, uihashcode, uihashequals, u4i, u32hashcode, uihashkeyequals, b4i, KV_HASH_GET_VAL);
+
+typedef struct { u8i key, val; } UUhash_t;
+#define UUhashcode(E) u64hashcode((E).key)
+#define UUhashequals(E1, E2) (E1).key == (E2).key
+#define UUhashkeyequals(E1, E2) (E1) == (E2).key
+define_hashtable(UUhash, UUhash_t, UUhashcode, UUhashequals, u8i, u64hashcode, UUhashkeyequals, u8i, KV_HASH_GET_VAL);
+
+typedef struct { char *key; u4i val; } cuhash_t;
+#define cuhash_code(e) __string_hashcode((e).key)
+#define cuhash_equals(e1, e2) (strcmp((e1).key, (e2).key) == 0)
+#define cuhash_key_equals(e1, e2) (strcmp((char*)(e1), (e2).key) == 0)
+define_hashtable(cuhash, cuhash_t, cuhash_code, cuhash_equals, char*, __string_hashcode, cuhash_key_equals, u4i, KV_HASH_GET_VAL);
+static const obj_desc_t cuhash_struct_deep_obj_desc = {"cuhash_struct_deep_obj_desc", sizeof(cuhash_t), 1, {1}, {offsetof(cuhash_t, key)}, {(obj_desc_t*)&OBJ_DESC_CHAR_ARRAY}, NULL, NULL};
+static const obj_desc_t cuhash_deep_obj_desc = {"cuhash_deep_obj_desc", sizeof(cuhash), 3, {1, 1, 1}, {offsetof(cuhash, array), offsetof(cuhash, ones), offsetof(cuhash, dels)}, {(obj_desc_t*)&cuhash_struct_deep_obj_desc, (obj_desc_t*)&bitvec_obj_desc, &bitvec_obj_desc}, cuhash_obj_desc_cnt, NULL};
+
+typedef struct { char *key; int val; } cihash_t;
+#define cihash_code(e) __string_hashcode((e).key)
+#define cihash_equals(e1, e2) (strcmp((e1).key, (e2).key) == 0)
+#define cihash_key_equals(e1, e2) (strcmp((char*)(e1), (e2).key) == 0)
+define_hashtable(cihash, cihash_t, cihash_code, cihash_equals, char*, __string_hashcode, cihash_key_equals, b4i, KV_HASH_GET_VAL);
+
+typedef struct { char *key; unsigned long long val; } clhash_t;
+#define clhash_code(e) __string_hashcode((e).key)
+#define clhash_equals(e1, e2) (strcmp((e1).key, (e2).key) == 0)
+#define clhash_key_equals(e1, e2) (strcmp((char*)(e1), (e2).key) == 0)
+define_hashtable(clhash, clhash_t, clhash_code, clhash_equals, char*, __string_hashcode, clhash_key_equals, u8i, KV_HASH_GET_VAL);
+
+typedef struct { char *key; char *val; } cchash_t;
+#define cchash_code(e) __string_hashcode((e).key)
+#define cchash_equals(e1, e2) (strcmp((e1).key, (e2).key) == 0)
+#define cchash_key_equals(e1, e2) (strcmp((char*)(e1), (e2).key) == 0)
+#define KV_CCHASH_GET_VAL(e) ((e)? (e)->val : NULL)
+define_hashtable(cchash, cchash_t, cchash_code, cchash_equals, char*, __string_hashcode, cchash_key_equals, char*, KV_CCHASH_GET_VAL);
+
+/**
+* Example of using userdata in thread-safe mode
+* char **strs;
+* ... codes init strs
+* #define test_hc(E) __string_hashcode(((char**)set->userdata)[E])
+* #define test_he(E1, E2) (strcmp(((char**)set->userdata)[E1], ((char**)set->userdata)[E2]) == 0)
+* define_hashset(testhash, uint32_t, test_hc, test_he);
+* testhash *hash = init_testhash(13);
+* set_userdata_testhash(hash, strs);
+* ... now, the key of testhash is uint32_t, but refer to strs
+*/
+
+#endif

kbm.c

@@ -0,0 +1,722 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "kbm.h"
+#include "kbmpoa.h"
+#include <regex.h>
+
+#ifndef VERSION
+#define VERSION 0.0
+#endif
+#ifndef RELEASE
+#define RELEASE 19830203
+#endif
+
+int kbm_usage(){
+	fprintf(stdout, "Program: kbm is a simple instance which implemented kmer-binmap\n");
+	fprintf(stdout, "         it maps query sequence against reference by kmer matching\n");
+	fprintf(stdout, "         matched kmer-pairs are bined (256bp) and counted in a matrix\n");
+	fprintf(stdout, "         dynamic programming is used to search the best path\n");
+	fprintf(stdout, "Version: %s (%s)\n", TOSTR(VERSION), TOSTR(RELEASE));
+	fprintf(stdout, "Author: Jue Ruan <ruanjue@gmail.com>\n");
+	fprintf(stdout, "Usage: kbm <options> [start|list|stop]\n");
+	fprintf(stdout, "Options:\n");
+	fprintf(stdout, " -i <string> File(s) of query sequences, +, [STDIN]\n");
+	fprintf(stdout, " -d <string> File(s) of reference sequences, +, [<-i>]\n");
+	fprintf(stdout, " -L <int>    Choose the longest subread and drop reads shorter than <int> (5000 recommended for PacBio) [0]\n");
+	fprintf(stdout, "             Negative integer indicate keeping read names, e.g. -5000.\n");
+	fprintf(stdout, " -o <string> Output file, [STDOUT]\n");
+	fprintf(stdout, " -I          Interactive mode\n");
+	fprintf(stdout, "             e.g. `mkfifo pipe` then `while true; do cat pipe && sleep 1; done | kbm -t 8 -I -d ref.fa -i - -Hk 21 -S 4`\n");
+	fprintf(stdout, "             then `cat 1.fq >pipe; cat 2.fq >pipe`, fastq format is better in interaction\n");
+	fprintf(stdout, " -f          Force overwrite\n");
+	fprintf(stdout, " -t <int>    Number of threads, 0: all cores, [1]\n");
+	fprintf(stdout, " -k <int>    Kmer-f size, <= %d, [0]\n", KBM_MAX_KSIZE);
+	fprintf(stdout, " -p <int>    Kmer-p size, <= %d, [21]\n", KBM_MAX_KSIZE);
+	fprintf(stdout, " -K <float>  Filter high frequency kmers, maybe repetitive, [1000]\n");
+	fprintf(stdout, "             if K >= 1, take the integer value as cutoff, MUST <= 65535\n");
+	fprintf(stdout, "             else, mask the top fraction part high frequency kmers\n");
+	fprintf(stdout, " -E <int>    Min kmer frequency, [1]\n");
+	fprintf(stdout, " -O <int>    Filter low complexity bins (#indexed_kmer less than <-O>), [2]\n");
+	fprintf(stdout, " -S <float>  Subsampling kmers, 1/(<-S>) kmers are indexed, [4.00]\n");
+	fprintf(stdout, "             -S is very useful in saving memeory and speeding up\n");
+	fprintf(stdout, "             please note that subsampling kmers will have less matched length\n");
+	fprintf(stdout, " -B <int>    Select no more than n seeds in a query bin, [256]\n");
+	// Obsolete
+	//fprintf(stdout, " -G <int>    Recognize error kmers in a bin when be aligned >= <-G> times, [0]\n");
+	fprintf(stdout, "             If you are using shared kbmidx by other process using -D too, it will bring wrong behavior\n");
+	fprintf(stdout, " -D <int>    Strand of alignment, 1: forward, 2: reverse, 3: both, [3]\n");
+	fprintf(stdout, " -X <int>    Max number of bin(256bp) in one gap, [4]\n");
+	fprintf(stdout, " -Y <int>    Max number of bin(256bp) in one deviation, [4]\n");
+	fprintf(stdout, " -Z <float>  Max fraction of gapped BINs / aligned BINs, [0.6]\n");
+	fprintf(stdout, " -x <int>    penalty for BIN gap, [-7]\n");
+	fprintf(stdout, " -y <int>    penalty for BIN deviation, [-21]\n");
+	fprintf(stdout, " -z <int>    Enable refine alignment with -p <-z> [0]\n");
+	fprintf(stdout, " -l <int>    Min alignment length, [2048]\n");
+	fprintf(stdout, " -m <int>    Min matched length, [200]\n");
+	fprintf(stdout, " -s <float>  Min similarity, calculated by kmer matched length / aligned length, [0.05]\n");
+	fprintf(stdout, " -r <float>  Max length variation of two aligned fragments, [0.25]\n");
+	fprintf(stdout, " -c          Insist to query contained reads against all\n");
+	fprintf(stdout, " -C          Chainning alignments\n");
+	fprintf(stdout, " -n <int>    Max hits per query, [1000]\n");
+#ifdef TEST_MODE
+	fprintf(stdout, " -T <int>    For debug, [0]\n");
+#endif
+	fprintf(stdout, " -W <string> Dump kbm index to file, [NULL]\n");
+	fprintf(stdout, " -R <string> Load kbm index from file, [NULL]\n");
+	fprintf(stdout, " -q          Quiet\n");
+	fprintf(stdout, " -V          Print version information and then exit\n");
+#if __DEBUG__
+	fprintf(stdout, " -v          Verbose, +\n");
+#endif
+	fprintf(stdout, "Server start: {kbm -R <wt.fa.kbmidx> start}, will mmap wt.fa.kbmidx into mmeory\n");
+	fprintf(stdout, "Server  list: {kbm -R <wt.fa.kbmidx> list [10]}, will list the object tree in file\n");
+	fprintf(stdout, "Server  stop: {kbm -R <wt.fa.kbmidx> stop},  will remove the mmap object\n");
+	return 1;
+}
+
+thread_beg_def(maln);
+CTGCNS *cc;
+KBMAux *aux;
+String *rdtag;
+BaseBank *rdseqs;
+u4i qidx;
+u8i rdoff;
+u4i rdlen;
+int corr_mode;
+float corr_cov;
+u4i corr_min, corr_max;
+FILE *out, *lay;
+int chainning;
+int interactive;
+int refine;
+thread_end_def(maln);
+
+thread_beg_func(maln);
+KBMPar *rpar;
+KBM *rkbm;
+KBMAux *raux;
+kbm_map_t HIT;
+u4v *tidxs;
+{
+	rpar = init_kbmpar();
+	rpar->ksize = 0;
+	rpar->psize = maln->refine;
+	rpar->min_bin_degree = 0;
+	rpar->kmin = 1;
+	rpar->kmax = 1000;
+	rpar->kmer_mod = KBM_N_HASH;
+	rkbm = init_kbm(rpar);
+	raux = init_kbmaux(rkbm);
+}
+tidxs = init_u4v(16);
+thread_beg_loop(maln);
+if(maln->rdlen == 0) break;
+if(maln->corr_mode){
+	if(map_kbmpoa(maln->cc, maln->aux, maln->rdtag->size? maln->rdtag->string : NULL, maln->qidx, maln->rdseqs, maln->rdoff, maln->rdlen, maln->corr_min, maln->corr_max, maln->corr_cov, maln->lay) == 0){
+		clear_kbmmapv(maln->aux->hits);
+		break;
+	}
+} else {
+	query_index_kbm(maln->aux, maln->rdtag->size? maln->rdtag->string : NULL, maln->qidx, maln->rdseqs, maln->rdoff, maln->rdlen);
+	map_kbm(maln->aux);
+	if(maln->refine && maln->aux->hits->size){
+		kbm_read_t *rd;
+		kbm_map_t *hit;
+		u4i i, j, tidx;
+		clear_kbm(rkbm);
+		bitpush_kbm(rkbm, maln->rdtag->size? maln->rdtag->string : NULL, maln->rdtag->size, maln->rdseqs->bits, 0, maln->rdlen);
+		ready_kbm(rkbm);
+		simple_index_kbm(rkbm, 0, rkbm->bins->size);
+		clear_u4v(tidxs);
+		for(i=0;i<maln->aux->hits->size;i++){
+			hit = ref_kbmmapv(maln->aux->hits, i);
+			if(tidxs->size == 0 || hit->tidx != tidxs->buffer[tidxs->size - 1]){
+				push_u4v(tidxs, hit->tidx);
+			}
+			if(KBM_LOG){
+				fprintf(maln->out, "#");
+				fprint_hit_kbm(maln->aux, i, maln->out);
+			}
+		}
+		clear_kbmmapv(maln->aux->hits);
+		clear_bitsvec(maln->aux->cigars);
+		for(i=0;i<tidxs->size;i++){
+			tidx = get_u4v(tidxs, i);
+			rd = ref_kbmreadv(maln->aux->kbm->reads, tidx);
+			query_index_kbm(raux, rd->tag, tidx, maln->aux->kbm->rdseqs, rd->seqoff * KBM_BIN_SIZE, rd->bincnt * KBM_BIN_SIZE);
+			map_kbm(raux);
+			for(j=0;j<raux->hits->size;j++){
+				flip_hit_kbmaux(maln->aux, raux, j);
+			}
+		}
+	}
+}
+if(maln->chainning){
+	u4i idx, lst;
+	for(idx=lst=0;idx<=maln->aux->hits->size;idx++){
+		if(idx == maln->aux->hits->size || maln->aux->hits->buffer[lst].tidx != maln->aux->hits->buffer[idx].tidx || maln->aux->hits->buffer[idx].qdir != maln->aux->hits->buffer[lst].qdir){
+			if(idx > lst + 1){
+				if(simple_chain_all_maps_kbm(maln->aux->hits->buffer + lst, idx - lst, maln->aux->cigars, &HIT, maln->aux->cigars, maln->aux->par->aln_var)){
+					maln->aux->hits->buffer[lst++] = HIT;
+					while(lst < idx){
+						maln->aux->hits->buffer[lst++].mat = 0;
+					}
+				}
+			}
+			lst = idx;
+		}
+	}
+}
+if(maln->aux->par->max_hit){
+	sort_array(maln->aux->hits->buffer, maln->aux->hits->size, kbm_map_t, num_cmpgt(b.mat, a.mat));
+	if(maln->aux->hits->size > maln->aux->par->max_hit) maln->aux->hits->size = maln->aux->par->max_hit;
+}
+if(maln->interactive){
+	u4i i;
+	thread_beg_syn(maln);
+	for(i=0;i<maln->aux->hits->size;i++){
+		fprint_hit_kbm(maln->aux, i, maln->out);
+	}
+	fflush(maln->out);
+	thread_end_syn(maln);
+}
+thread_end_loop(maln);
+{
+	free_kbmaux(raux);
+	free_kbm(rkbm);
+	free_kbmpar(rpar);
+}
+free_u4v(tidxs);
+thread_end_func(maln);
+
+int kbm_main(int argc, char **argv){
+	cplist *qrys, *refs;
+	char *outf, *loadf, *dumpf;
+	FILE *out, *dump;
+	KBM *kbm;
+	KBMPar *par;
+	KBMAux *aux;
+	BitVec *solids, *rdflags;
+	FileReader *fr;
+	BioSequence *seqs[2], *seq;
+	char regtag[14];
+	u8i tot_bp, max_bp, opt_flags, nhit;
+	u4i qidx, i;
+	int c, ncpu, buffered_read, overwrite, quiet, tidy_reads, tidy_rdtag, skip_ctn, chainning, interactive, server, tree_maxcnt;
+	int solid_kmer, refine;
+	float fval;
+	thread_preprocess(maln);
+	par = init_kbmpar();
+	par->rd_len_order = 1;
+	chainning = 0;
+	KBM_LOG = 0;
+	buffered_read = 1;
+	skip_ctn = 1;
+	interactive = 0;
+	solid_kmer = 0;
+	refine = 0;
+	solids = NULL;
+	rdflags = NULL;
+	ncpu = 1;
+	quiet = 0;
+	tidy_reads = 0;
+	tidy_rdtag = -1;
+	max_bp = MAX_U8;
+	qrys = init_cplist(4);
+	refs = init_cplist(4);
+	outf = NULL;
+	loadf = NULL;
+	dumpf = NULL;
+	overwrite = 0;
+	server = 0;
+	tree_maxcnt = 10;
+	opt_flags = 0;
+	while((c = getopt(argc, argv, "hi:d:o:fIt:k:p:K:E:O:S:B:G:D:X:Y:Z:x:y:z:l:m:n:s:cr:CT:W:R:qvV")) != -1){
+		switch(c){
+			case 'h': return kbm_usage();
+			case 'i': push_cplist(qrys, optarg); break;
+			case 'd': push_cplist(refs, optarg); break;
+			case 'L': tidy_reads = atoi(optarg); break;
+			case 'o': outf = optarg; break;
+			case 'f': overwrite = 1; break;
+			case 'I': interactive = 1; break;
+			case 't': ncpu = atoi(optarg); break;
+			case 'k': par->ksize = atoi(optarg); opt_flags |= (1 << 1); break;
+			case 'p': par->psize = atoi(optarg); opt_flags |= (1 << 0); break;
+			case 'K': fval = atof(optarg); par->kmax = fval; par->ktop = fval - par->kmax; break;
+			case 'E': par->kmin = atoi(optarg); break;
+			case 'O': par->min_bin_degree = atoi(optarg); break;
+			case 'S': par->kmer_mod = UInt(atof(optarg) * KBM_N_HASH); opt_flags |= (1 << 2); break;
+			case 'B': par->ksampling = atoi(optarg); break;
+			case 'G': solid_kmer = atoi(optarg); break;
+			case 'D': par->strand_mask = atoi(optarg); break;
+			case 'X': par->max_bgap = atoi(optarg); break;
+			case 'Y': par->max_bvar = atoi(optarg); break;
+			case 'Z': par->max_gap  = atof(optarg); break;
+			case 'x': par->pgap = atoi(optarg); break;
+			case 'y': par->pvar = atoi(optarg); break;
+			case 'z': refine = atoi(optarg); break;
+			case 'l': par->min_aln = atoi(optarg) / KBM_BIN_SIZE; break;
+			case 'm': par->min_mat = atoi(optarg); break;
+			case 'n': par->max_hit = atoi(optarg); break;
+			case 's': par->min_sim = atof(optarg); break;
+			case 'r': par->aln_var = atof(optarg); break;
+			case 'c': skip_ctn = 0; break;
+			case 'C': chainning = 1; break;
+#ifdef TEST_MODE
+			case 'T': par->test_mode = atoi(optarg); break;
+#endif
+			case 'W': dumpf = optarg; break;
+			case 'R': loadf = optarg; break;
+			case 'q': quiet = 1; break;
+			case 'v': KBM_LOG ++; break;
+			case 'V': fprintf(stdout, "kbm2 %s\n", TOSTR(VERSION)); return 0;
+			default: return kbm_usage();
+		}
+	}
+	if(quiet){
+		int devnull;
+		devnull = open("/dev/null", O_WRONLY);
+		dup2(devnull, KBM_LOGFNO);
+	}
+	if(tidy_rdtag == -1){
+		tidy_rdtag = (tidy_reads >= 0);
+	}
+	if(tidy_reads < 0) tidy_reads = - tidy_reads;
+	BEG_STAT_PROC_INFO(KBM_LOGF, argc, argv);
+	if(par->ksize + par->psize > KBM_MAX_KSIZE){
+		fprintf(stderr, " -- Invalid kmer size %d+%d=%d > %d in %s -- %s:%d --\n", par->ksize, par->psize, par->ksize + par->psize,  KBM_MAX_KSIZE, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+		return 1;
+	}
+	out = open_file_for_write(outf, NULL, overwrite);
+	if(refs->size == 0){
+		if(loadf){
+		} else if(qrys->size){
+			append_cplist(refs, qrys);
+			clear_cplist(qrys);
+		} else {
+			fprintf(KBM_LOGF, " ** -h to show document\n"); fflush(KBM_LOGF);
+			push_cplist(refs, "-");
+		}
+	}
+	if(qrys->size == 0){
+		par->self_aln = 1;
+		interactive = 0;
+		if(par->kmin < 2) par->kmin = 2;
+	}
+	if(interactive) buffered_read = 0;
+	if(ncpu <= 0){
+		ncpu = _proc_deamon->ncpu;
+		if(ncpu == 0) ncpu = 1; // failed to get number of cores
+	}
+	if(KBM_LOG > 0){
+		fprintf(KBM_LOGF, "KBM_LOG_LEVEL = %d\n", KBM_LOG);
+	}
+	if(optind < argc){
+		server = 0;
+		if(strcasecmp("start", argv[optind]) == 0) server = 1;
+		else if(strcasecmp("stop", argv[optind]) == 0) server = 2;
+		else if(strcasecmp("list", argv[optind]) == 0){
+			server = 3;
+			if(optind + 1 < argc) tree_maxcnt = atoi(argv[optind + 1]);
+		}
+		if(loadf == NULL) server = 0;
+	}
+	if(loadf){
+		if(server == 1){
+			fprintf(KBM_LOGF, "[%s] loading kbm index from %s\n", date(), loadf);
+			kbm = mem_load_obj_file(&kbm_obj_desc, loadf, NULL, NULL, NULL, NULL);
+			fprintf(KBM_LOGF, "[%s] Done. %u sequences, %llu bp, parameter('-k %d -p %d -S %d')\n", date(), (u4i)kbm->reads->size, (u8i)kbm->rdseqs->size, kbm->par->ksize, kbm->par->psize, kbm->par->kmer_mod / KBM_N_HASH);
+			fprintf(KBM_LOGF, "[%s] kbm-index server start\n", date());
+			return 0;
+		} else if(server == 2){
+			if(mem_stop_obj_file(loadf)){
+				fprintf(KBM_LOGF, "[%s] kbm-index server for '%s' stop\n", date(), loadf);
+			} else {
+				fprintf(KBM_LOGF, "[%s] unable to find kbm-index server for '%s'\n", date(), loadf);
+			}
+			return 0;
+		} else if(server == 3){
+			print_tree_obj_file(stdout, &kbm_obj_desc, loadf, tree_maxcnt, 0);
+			return 0;
+		} else {
+			fprintf(KBM_LOGF, "[%s] loading kbm index from %s\n", date(), loadf);
+			if((kbm = mem_find_obj_file(&kbm_obj_desc, loadf, NULL, NULL, NULL, NULL, 1)) == NULL){
+				fprintf(KBM_LOGF, " -- cannot find mmap object %s --\n", loadf);
+				fprintf(KBM_LOGF, " -- try read from file --\n");
+				kbm = mem_read_obj_file(&kbm_obj_desc, loadf, NULL, NULL, NULL, NULL);
+			}
+		}
+		fprintf(KBM_LOGF, "[%s] Done. %u sequences, %llu bp, parameter('-k %d -p %d -S %d')\n", date(), (u4i)kbm->reads->size, (u8i)kbm->rdseqs->size, kbm->par->ksize, kbm->par->psize, kbm->par->kmer_mod / KBM_N_HASH);
+		// Please note that, kbm->tag2idx is not functional after mem_load
+		// check KBMPar
+		if((opt_flags >> 0) & 0x01){
+			if(kbm->par->psize != par->psize){
+				fprintf(KBM_LOGF, " ** -p is different, %d != %d\n", kbm->par->psize, par->psize); exit(1);
+			}
+		} else {
+			par->psize = kbm->par->psize;
+		}
+		if((opt_flags >> 1) & 0x01){
+			if(kbm->par->ksize != par->ksize){
+				fprintf(KBM_LOGF, " ** -k is different, %d != %d\n", kbm->par->ksize, par->ksize); exit(1);
+			}
+		} else {
+			par->ksize = kbm->par->ksize;
+		}
+		if((opt_flags >> 2) & 0x01){
+			if(kbm->par->kmer_mod != par->kmer_mod){
+				fprintf(KBM_LOGF, " ** -S is different, %d != %d\n", kbm->par->kmer_mod / KBM_N_HASH, par->kmer_mod / KBM_N_HASH); exit(1);
+			}
+		} else {
+			par->kmer_mod = kbm->par->kmer_mod;
+		}
+		if((opt_flags >> 3) & 0x01){
+			if(kbm->par->rd_len_order != par->rd_len_order){
+				fprintf(KBM_LOGF, " ** par->rd_len_order is different, %d != %d\n", kbm->par->rd_len_order, par->rd_len_order); exit(1);
+			}
+		} else {
+			par->rd_len_order = kbm->par->rd_len_order;
+		}
+	} else {
+		kbm = init_kbm(par);
+		fprintf(KBM_LOGF, "[%s] loading sequences\n", date()); fflush(KBM_LOGF);
+		fr = open_all_filereader(refs->size, refs->buffer, buffered_read);
+		tot_bp = 0;
+		seqs[0] = init_biosequence();
+		seqs[1] = init_biosequence();
+		regex_t reg;
+		regmatch_t mats[3];
+		int z, tag_size, len;
+		z = regcomp(&reg, "^(.+?)/[0-9]+_[0-9]+$", REG_EXTENDED);
+		if(z){
+			regerror(z, &reg, regtag, 13);
+			fprintf(stderr, " -- REGCOMP: %s --\n", regtag); fflush(stderr);
+			return 1;
+		}
+		{
+			int k = 0;
+			reset_biosequence(seqs[0]);
+			reset_biosequence(seqs[1]);
+			while(1){
+				int has = readseq_filereader(fr, seqs[k]);
+				if(tidy_reads){
+					if(has){
+						if((z = regexec(&reg, seqs[k]->tag->string, 3, mats, 0)) == 0){
+							trunc_string(seqs[k]->tag, mats[1].rm_eo);
+						} else if(z != REG_NOMATCH){
+							regerror(z, &reg, regtag, 13);
+							fprintf(stderr, " -- REGEXEC: %s --\n", regtag); fflush(stderr);
+						}
+						//fprintf(stderr, "1: %s len=%d\n", seqs[k]->tag->string, seqs[k]->seq->size); fflush(stderr);
+						//fprintf(stderr, "2: %s len=%d\n", seqs[!k]->tag->string, seqs[!k]->seq->size); fflush(stderr);
+						if(seqs[k]->tag->size == seqs[!k]->tag->size && strcmp(seqs[k]->tag->string, seqs[!k]->tag->string) == 0){
+							if(seqs[k]->seq->size > seqs[!k]->seq->size){
+								k = !k;
+							}
+							continue;
+						} else {
+							seq = seqs[!k];
+							k = !k;
+						}
+					} else {
+						seq = seqs[!k];
+					}
+					if(seq->seq->size < tidy_reads){
+						if(has) continue;
+						else break;
+					}
+					if(tidy_rdtag){
+						sprintf(regtag, "S%010llu", (u8i)kbm->reads->size);
+						clear_string(seq->tag);
+						append_string(seq->tag, regtag, 11);
+					}
+				} else {
+					if(has == 0) break;
+					seq = seqs[k];
+				}
+				tag_size = seq->tag->size;
+				for(i=0;i<UInt(seq->seq->size);i+=KBM_MAX_RDLEN){
+					len = num_min(seq->seq->size - i, KBM_MAX_RDLEN);
+					if(i){
+						append_string(seq->tag, "_V", 2);
+						add_int_string(seq->tag, i / KBM_MAX_RDLEN);
+					}
+					if(!KBM_LOG && (kbm->reads->size % 10000) == 0){ fprintf(KBM_LOGF, "\r%u", (u4i)kbm->reads->size); fflush(KBM_LOGF); }
+					//fprintf(stderr, " -- %s len=%d in %s -- %s:%d --\n", seq->tag->string, seq->seq->size, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+					if(kbm->reads->size >= KBM_MAX_RDCNT){
+						fprintf(stderr, " -- Read Number Out of Range: %u --\n", (u4i)kbm->reads->size); fflush(stderr);
+						break;
+					}
+					push_kbm(kbm, seq->tag->string, seq->tag->size, seq->seq->string + i, len);
+					if(i){ seq->tag->size = tag_size; seq->tag->string[tag_size] = '\0'; }
+				}
+				tot_bp += seq->seq->size;
+				if(max_bp && tot_bp >= max_bp){ break; }
+				if(has == 0) break;
+				if(kbm->reads->size >= KBM_MAX_RDCNT){
+					fprintf(stderr, " -- Read Number Out of Range: %u --\n", (u4i)kbm->reads->size); fflush(stderr);
+					break;
+				}
+			}
+		}
+		close_filereader(fr);
+		regfree(&reg);
+		free_biosequence(seqs[0]);
+		free_biosequence(seqs[1]);
+		if(!KBM_LOG){ fprintf(KBM_LOGF, "\r%u reads", (unsigned)kbm->reads->size); fflush(KBM_LOGF); }
+		ready_kbm(kbm);
+		fprintf(KBM_LOGF, "\n[%s] Done, %u reads, %llu bp, %u bins\n", date(), (u4i)kbm->reads->size, tot_bp, (u4i)kbm->bins->size); fflush(KBM_LOGF);
+		fprintf(KBM_LOGF, "[%s] indexing, %d threads\n", date(), ncpu);
+		index_kbm(kbm, 0, kbm->bins->size, ncpu, NULL);
+		if(dumpf){
+			u8i size;
+			dump = open_file_for_write(dumpf, NULL, 1);
+			size = mem_dump_obj_file(kbm, 1, &kbm_obj_desc, 1, 0, dump);
+			fclose(dump);
+			fprintf(KBM_LOGF, "[%s] kbm index dumped to %s\n", date(), dumpf);
+		}
+	}
+	fprintf(KBM_LOGF, "[%s] mapping\n", date());
+	if(solid_kmer && par->self_aln){
+		solids = init_bitvec((kbm->rdseqs->size >> 1) + 1);
+	}
+	thread_beg_init(maln, ncpu);
+	maln->aux = init_kbmaux(kbm);
+	maln->aux->par = par; // par might be different from kbm->par
+	maln->aux->solids = solids;
+	//maln->aux->bmin = 611077674;
+	//maln->aux->bmax = 611077674 + 172;
+	maln->rdtag = init_string(64);
+	maln->rdseqs = qrys->size? init_basebank() : kbm->rdseqs;
+	maln->rdoff = 0;
+	maln->rdlen = 0;
+	maln->cc = NULL;
+	maln->corr_mode = 0;
+	maln->corr_cov = 0.75;
+	maln->corr_min = 5;
+	maln->corr_max = 10;
+	maln->out = out;
+	maln->lay = NULL;
+	maln->chainning = chainning;
+	maln->interactive = interactive;
+	maln->refine = refine;
+	thread_end_init(maln);
+	if(qrys->size){
+		int run_mode;
+		fr = open_all_filereader(qrys->size, qrys->buffer, buffered_read);
+		run_mode = 0;
+		if(readline_filereader(fr) > 0){
+			if(strcasecmp(fr->line->string, "#pairwise_test") == 0){
+				run_mode = 1;
+			} else if(strcasecmp(fr->line->string, "#print_exists") == 0){
+				run_mode = 2;
+			} else if(strcasecmp(fr->line->string, "#correct_align") == 0){
+				run_mode = 3;
+				cns_debug = KBM_LOG;
+				thread_beg_iter(maln);
+				maln->corr_mode = 1;
+				KBMBlock *kb;
+				POGPar par;
+				kb = init_kbmblock(2048, 2048 - 512);
+				par = DEFAULT_POG_PAR;
+				//maln->cc = init_ctgcns(kb, iter_kbmblock, info_kbmblock, 1, 1, 1, maln->corr_max, 200, 100, 1, 96, 2, -5, -2, -4, -1, 16, 3, 0.5, 512);
+				par.refmode = 1;
+				maln->cc = init_ctgcns(kb, iter_kbmblock, info_kbmblock, 1, 1, maln->corr_max, 200, 100, 1, 512, &par);
+				maln->lay = maln->out;
+				thread_end_iter(maln);
+			} else if(strcasecmp(fr->line->string, "#print_read") == 0){
+				run_mode = 4;
+			} else {
+				rollback_filereader(fr);
+			}
+		}
+		seq = init_biosequence();
+		qidx = 0;
+		nhit = 0;
+		if(run_mode == 0 || run_mode == 3){
+			while(readseq_filereader(fr, seq)){
+				if((qidx % 100) == 0){
+					fprintf(KBM_LOGF, "\r%u\t%llu", qidx, nhit); fflush(KBM_LOGF);
+				}
+				thread_wait_one(maln);
+				if(maln->rdlen && !maln->interactive){
+					{
+						aux = maln->aux;
+						for(i=0;i<aux->hits->size;i++){
+							fprint_hit_kbm(aux, i, out);
+						}
+						if(run_mode == 3) fflush(out);
+						nhit += aux->hits->size;
+					}
+				}
+				trunc_string(seq->seq, kbm_cvt_length(seq->seq->size));
+				clear_basebank(maln->rdseqs);
+				seq2basebank(maln->rdseqs, seq->seq->string, seq->seq->size);
+				clear_string(maln->rdtag);
+				append_string(maln->rdtag, seq->tag->string, seq->tag->size);
+				maln->qidx = qidx ++;
+				maln->rdoff = 0;
+				maln->rdlen = seq->seq->size;
+				thread_wake(maln);
+			}
+		} else if(run_mode == 1){
+			int nc;
+			u4i tidx;
+			thread_beg_operate(maln, 0);
+			aux = maln->aux;
+			free_basebank(maln->rdseqs);
+			maln->rdseqs = kbm->rdseqs;
+			while((nc = readtable_filereader(fr)) >= 0){
+				if(nc < 2) continue;
+				qidx = getval_cuhash(kbm->tag2idx, get_col_str(fr, 0));
+				tidx = getval_cuhash(kbm->tag2idx, get_col_str(fr, 1));
+				if(qidx == MAX_U4 || tidx == MAX_U4) continue;
+				if(qidx > tidx){ swap_var(qidx, tidx); }
+				maln->qidx = qidx;
+				maln->rdoff = kbm->reads->buffer[qidx].seqoff * KBM_BIN_SIZE;
+				maln->rdlen = kbm->reads->buffer[qidx].bincnt * KBM_BIN_SIZE;
+				aux->bmin = kbm->reads->buffer[tidx].binoff;
+				aux->bmax = kbm->reads->buffer[tidx].binoff + kbm->reads->buffer[tidx].bincnt;
+				fprintf(out, "%s <-> %s\n", kbm->reads->buffer[qidx].tag, kbm->reads->buffer[tidx].tag);
+				thread_wake(maln);
+				thread_wait(maln);
+				if(maln->rdlen && !maln->interactive){
+					for(i=0;i<aux->hits->size;i++){
+						fprint_hit_kbm(aux, i, out);
+					}
+					nhit += aux->hits->size;
+				}
+				fprintf(out, "//\n");
+				maln->rdlen = 0;
+			}
+		} else if(run_mode == 2){
+			kmeroffv *kmers[2];
+			int nc;
+			kmers[0] = adv_init_kmeroffv(32, 0, 1);
+			kmers[1] = adv_init_kmeroffv(32, 0, 1);
+			while((nc = readtable_filereader(fr)) >= 0){
+				//fprintf(stdout, "%s\n", fr->line->string);
+				if(nc < 1) continue;
+				qidx = getval_cuhash(kbm->tag2idx, get_col_str(fr, 0));
+				if(qidx == MAX_U4) continue;
+				print_exists_index_kbm(kbm, kbm->reads->buffer[qidx].tag, kbm->rdseqs, kbm->reads->buffer[qidx].seqoff * KBM_BIN_SIZE,
+					kbm->reads->buffer[qidx].bincnt * KBM_BIN_SIZE, kmers, stdout);
+			}
+			free_kmeroffv(kmers[0]);
+			free_kmeroffv(kmers[1]);
+		} else if(run_mode == 4){
+			int nc;
+			while((nc = readtable_filereader(fr)) >= 0){
+				if(nc < 1) continue;
+				if(fr->line->string[0] == '#') continue;
+				qidx = getval_cuhash(kbm->tag2idx, get_col_str(fr, 0));
+				if(qidx == MAX_U4){
+					fprintf(out, "#%s NOT FOUND\n", get_col_str(fr, 0));
+					continue;
+				}
+				fprintf(out, ">%s\n", kbm->reads->buffer[qidx].tag);
+				println_fwdseq_basebank(kbm->rdseqs, kbm->reads->buffer[qidx].seqoff * KBM_BIN_SIZE, kbm->reads->buffer[qidx].bincnt * KBM_BIN_SIZE, out);
+				fflush(out);
+			}
+		}
+		free_filereader(fr);
+		free_biosequence(seq);
+		thread_beg_iter(maln);
+		thread_wait(maln);
+		if(maln->rdlen && !maln->interactive){
+			aux = maln->aux;
+			for(i=0;i<aux->hits->size;i++){
+				fprint_hit_kbm(aux, i, out);
+			}
+			if(run_mode == 3) fflush(out);
+			nhit += aux->hits->size;
+			maln->rdlen = 0;
+		}
+		thread_end_iter(maln);
+		fprintf(KBM_LOGF, "\r%u\t%llu\n", qidx, (u8i)nhit); fflush(KBM_LOGF);
+	} else {
+		nhit = 0;
+		rdflags = init_bitvec(kbm->reads->size);
+		for(qidx=0;qidx<kbm->reads->size+ncpu;qidx++){
+			if(qidx < kbm->reads->size){
+				thread_wait_one(maln);
+			} else {
+				thread_wait_next(maln);
+			}
+			if(maln->rdlen){
+				aux = maln->aux;
+				for(i=0;i<aux->hits->size;i++){
+					if(par->test_mode == 0 && skip_ctn){
+						// whether reads[tidx] is contained by reads[qidx]
+						kbm_map_t *hit;
+						int margin = KBM_BIN_SIZE;
+						hit = ref_kbmmapv(aux->hits, i);
+						if((hit->tb <= margin && hit->te + margin >= (int)kbm->reads->buffer[hit->tidx].bincnt * KBM_BIN_SIZE)
+							&& (hit->qb > margin || hit->qe + margin < (int)kbm->reads->buffer[hit->qidx].bincnt * KBM_BIN_SIZE)){
+							one_bitvec(rdflags, hit->tidx);
+						}
+					}
+					fprint_hit_kbm(aux, i, out);
+				}
+				nhit += aux->hits->size;
+			}
+			if(qidx >= kbm->reads->size || get_bitvec(rdflags, qidx)){
+				maln->rdlen = 0;
+				continue;
+			} else if((qidx % 100) == 0){
+				fprintf(KBM_LOGF, "\r%u\t%llu", qidx, nhit); fflush(KBM_LOGF);
+			}
+			maln->qidx = qidx;
+			maln->rdoff = kbm->reads->buffer[qidx].seqoff * KBM_BIN_SIZE;
+			maln->rdlen = kbm->reads->buffer[qidx].bincnt * KBM_BIN_SIZE;
+			thread_wake(maln);
+		}
+		fprintf(KBM_LOGF, "\r%u\t%llu\n", qidx, nhit); fflush(KBM_LOGF);
+		free_bitvec(rdflags);
+	}
+	thread_beg_close(maln);
+	free_kbmaux(maln->aux);
+	free_string(maln->rdtag);
+	if(maln->rdseqs && maln->rdseqs != kbm->rdseqs) free_basebank(maln->rdseqs);
+	if(maln->cc){
+		free_kbmblock((KBMBlock*)maln->cc->obj);
+		free_ctgcns(maln->cc);
+	}
+	thread_end_close(maln);
+	if(solids) free_bitvec(solids);
+	fprintf(KBM_LOGF, "[%s] Done\n", date());
+	if(outf) fclose(out);
+	free_cplist(qrys);
+	free_cplist(refs);
+	free_kbmpar(par);
+	if(loadf){
+		// DON't free kbm, for it is mem_load object
+	} else {
+		free_kbm(kbm);
+	}
+	END_STAT_PROC_INFO(KBM_LOGF);
+	return 0;
+}
+
+int main(int argc, char **argv){
+	return kbm_main(argc, argv);
+}
+

kbmpoa.h

@@ -0,0 +1,412 @@
+/*
+ * 
+ * Copyright (c) 2018, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef KMER_BIN_MAP_PO_MSA_CNS_RJ_H
+#define KMER_BIN_MAP_PO_MSA_CNS_RJ_H
+
+#include "kbm.h"
+#include "wtpoa.h"
+
+typedef struct {
+	u4i chridx, cidx, bidx;
+	String   *rdtag;
+	BaseBank *rdseq;
+	u8i       rdoff;
+	u4i       rdlen;
+	KBMAux   *aux;
+	layseqr  *seqs;
+	u4v      *heap;
+	u2i bsize, bstep; // block size, and slide steps
+	u4i sidx, lidx, hidx;
+} KBMBlock;
+
+static inline KBMBlock* init_kbmblock(u2i bsize, u2i bstep){
+	KBMBlock *kb;
+	kb = malloc(sizeof(KBMBlock));
+	kb->rdtag = init_string(1024);
+	kb->rdseq = NULL;
+	kb->rdoff = 0;
+	kb->rdlen = 0;
+	kb->aux   = NULL;
+	kb->seqs  = init_layseqr(32);
+	kb->heap  = init_u4v(32);
+	bsize = ((bsize + KBM_BIN_SIZE - 1) / KBM_BIN_SIZE) * KBM_BIN_SIZE;
+	if(bsize == 0) bsize = KBM_BIN_SIZE;
+	bstep = ((bstep + KBM_BIN_SIZE - 1) / KBM_BIN_SIZE) * KBM_BIN_SIZE;
+	if(bstep == 0) bstep = KBM_BIN_SIZE;
+	kb->bsize = bsize;
+	kb->bstep = bstep;
+	kb->chridx = 0;
+	kb->cidx   = 0;
+	kb->bidx   = 0;
+	kb->sidx  = MAX_U4;
+	kb->lidx  = MAX_U4;
+	kb->hidx  = 0;
+	return kb;
+}
+
+static inline void free_kbmblock(KBMBlock *kb){
+	free_string(kb->rdtag);
+	free_layseqr(kb->seqs);
+	free(kb);
+}
+
+static inline void reset_kbmblock(KBMBlock *kb, char *rdtag, u4i chridx, BaseBank *rdseq, u8i rdoff, u4i rdlen, KBMAux *aux){
+	lay_seq_t *stop;
+	kb->chridx = chridx;
+	kb->cidx   = chridx;
+	kb->bidx = 0;
+	clear_string(kb->rdtag);
+	if(rdtag){
+		append_string(kb->rdtag, rdtag, strlen(rdtag));
+	} else {
+		append_string(kb->rdtag, "anonymous", strlen("anonymous"));
+	}
+	kb->rdseq = rdseq;
+	kb->rdoff = rdoff;
+	kb->rdlen = rdlen;
+	kb->aux   = aux;
+	recyc_all_layseqr(kb->seqs);
+	clear_u4v(kb->heap);
+	kb->sidx = MAX_U4;
+	kb->lidx  = MAX_U4;
+	sort_array(aux->hits->buffer, aux->hits->size, kbm_map_t, num_cmpgt(a.qb, b.qb));
+	kb->hidx  = 0;
+	stop = pop_layseqr(kb->seqs);
+	stop->chridx = chridx + 1;
+	stop->bidx   = 0;
+}
+
+static inline lay_seq_t* _push_padding_ref_kbmblock(KBMBlock *kb, lay_seq_t *sq){
+	lay_seq_t *st;
+	u8i off;
+	u4i sqidx, len;
+	sqidx = offset_layseqr(kb->seqs, sq);
+	while(kb->chridx == sq->chridx && kb->bidx <= sq->bidx){
+		st = pop_layseqr(kb->seqs);
+		st->chridx = kb->cidx;
+		st->bidx  = kb->bidx;
+		st->rdidx = 0;
+		st->rddir = 0;
+		st->rdoff = st->bidx * kb->bstep;
+		st->rbeg  = st->rdoff;
+		off = kb->rdoff;
+		len = kb->rdlen;
+		st->rend = num_min(st->rbeg + kb->bsize, len);
+		st->rdtag = kb->rdtag->string;
+		clear_basebank(st->seq);
+		fast_fwdbits2basebank(st->seq, kb->rdseq->bits, off + st->rdoff, st->rend - st->rbeg);
+		array_heap_push(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, offset_layseqr(kb->seqs, st),
+			num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+		if(st->rend >= len){
+			kb->chridx = MAX_U4;
+			kb->bidx = 0;
+			break;
+		} else {
+			kb->bidx ++;
+		}
+		sq = ref_layseqr(kb->seqs, sqidx);
+	}
+	sq = ref_layseqr(kb->seqs, sqidx);
+	return sq;
+}
+
+static inline lay_seq_t* iter_kbmblock(void *obj){
+	KBMBlock *kb;
+	kbm_map_t *hit;
+	lay_seq_t *sc, *sl;
+	u8i coff, tsoff;
+	u4i off, rdoff, rdlen, nxt, val, len, clen, bt;
+	kb = (KBMBlock*)obj;
+	if(kb->sidx != MAX_U4){
+		recyc_layseqr(kb->seqs, kb->sidx);
+		kb->sidx = MAX_U4;
+	}
+	sc = NULL;
+	sl = NULL;
+	do {
+		if(kb->heap->size == 0){
+			kb->lidx = MAX_U4;
+		}
+		while(kb->lidx == MAX_U4){
+			while(kb->hidx < kb->aux->hits->size){
+				if(kb->aux->hits->buffer[kb->hidx].mat == 0) kb->hidx ++;
+				else break;
+			}
+			if(kb->hidx >= kb->aux->hits->size){
+				sc = ref_layseqr(kb->seqs, 0);
+				sc = _push_padding_ref_kbmblock(kb, sc);
+				break;
+			}
+			hit = ref_kbmmapv(kb->aux->hits, kb->hidx ++);
+			tsoff = kb->aux->kbm->reads->buffer[hit->tidx].seqoff * KBM_BIN_SIZE;
+			rdlen = kb->aux->kbm->reads->buffer[hit->tidx].bincnt * KBM_BIN_SIZE;
+			off   = hit->qb;
+			rdoff = hit->qdir? Int(rdlen - hit->te) : hit->tb;
+			{
+				encap_layseqr(kb->seqs, 2);
+				sc = pop_layseqr(kb->seqs);
+				sc->chridx = kb->cidx;
+				sc->bidx = (off / kb->bstep);
+				sc->rdidx = hit->tidx;
+				sc->rddir = hit->qdir;
+				sc->rdoff = rdoff;
+				sc->rbeg = off;
+				sc->rend = 0;
+				sc->rdtag  = kb->aux->kbm->reads->buffer[hit->tidx].tag;
+				clear_basebank(sc->seq);
+			}
+			sl = NULL;
+			{
+				nxt = (off / kb->bstep) * kb->bstep;
+				if(nxt && off - nxt < UInt(kb->bsize - kb->bstep)){
+					if(sc->bidx){
+						sl = pop_layseqr(kb->seqs);
+						sl->chridx = kb->cidx;
+						sl->bidx   = sc->bidx - 1;
+						sl->rdidx  = hit->tidx;
+						sl->rddir  = hit->qdir;
+						sl->rdoff  = rdoff;
+						sl->rbeg   = off;
+						sl->rend   = 0;
+						sl->rdtag  = kb->aux->kbm->reads->buffer[hit->tidx].tag;
+						clear_basebank(sl->seq);
+					}
+					nxt += kb->bsize - kb->bstep;
+				} else {
+					nxt += kb->bstep;
+				}
+			}
+			coff = hit->cgoff;
+			clen = hit->cglen;
+			while(clen){
+				bt  = get_bitsvec(kb->aux->cigars, coff + clen - 1);
+				clen --;
+				val = KBM_BIN_SIZE;
+				bt  = (bt & 0x03)? (bt & 0x03) : 3;
+				while(val){
+					if(bt & 0b001){
+						len = num_min(val, nxt - off);
+						off += len;
+					} else {
+						len = val;
+					}
+					val -= len;
+					if(bt & 0b010){
+						{
+							if(sc){
+								if(sc->seq->size == 0){
+									sc->rbeg = (bt & 0b001)? off - len : off;
+								}
+								if(hit->qdir){
+									revbits2basebank(sc->seq, kb->aux->kbm->rdseqs->bits, tsoff + rdlen - (rdoff + len), len);
+								} else {
+									fwdbits2basebank(sc->seq, kb->aux->kbm->rdseqs->bits, tsoff + rdoff, len);
+								}
+								sc->rend = off;
+							}
+							if(sl){
+								if(sl->seq->size == 0){
+									sl->rbeg = (bt & 0b001)? off - len : off;
+								}
+								if(hit->qdir){
+									revbits2basebank(sl->seq, kb->aux->kbm->rdseqs->bits, tsoff + rdlen - (rdoff + len), len);
+								} else {
+									fwdbits2basebank(sl->seq, kb->aux->kbm->rdseqs->bits, tsoff + rdoff, len);
+								}
+								sl->rend = off;
+							}
+						}
+						rdoff += len;
+					}
+					if(off == nxt){
+						if(sl){
+							if(sl->rend > sl->rbeg){
+								u4i scidx;
+								scidx = offset_layseqr(kb->seqs, sc);
+								sl = _push_padding_ref_kbmblock(kb, sl);
+								sc = ref_layseqr(kb->seqs, scidx);
+								if(kb->lidx == MAX_U4){
+									kb->lidx = offset_layseqr(kb->seqs, sl);
+								}
+								array_heap_push(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, offset_layseqr(kb->seqs, sl),
+									num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+							}
+							sl = NULL;
+							nxt += 2 * kb->bstep - kb->bsize;
+						} else {
+							u4i scidx;
+							scidx = offset_layseqr(kb->seqs, sc);
+							encap_layseqr(kb->seqs, 1);
+							sl = ref_layseqr(kb->seqs, scidx);
+							sc = pop_layseqr(kb->seqs);
+							sc->chridx = kb->cidx;
+							sc->bidx  = sl->bidx + 1;
+							sc->rdidx = hit->tidx;
+							sc->rddir = hit->qdir;
+							sc->rdoff = rdoff;
+							sc->rbeg = off;
+							sc->rend = 0;
+							sc->rdtag  = kb->aux->kbm->reads->buffer[hit->tidx].tag;
+							clear_basebank(sc->seq);
+							nxt += kb->bsize - kb->bstep;
+						}
+					}
+				}
+			}
+			if(sl && sl->rend > sl->rbeg){
+				u4i scidx;
+				scidx = offset_layseqr(kb->seqs, sc);
+				sl = _push_padding_ref_kbmblock(kb, sl);
+				sc = ref_layseqr(kb->seqs, scidx);
+				if(kb->lidx == MAX_U4){
+					kb->lidx = offset_layseqr(kb->seqs, sl);
+				}
+				array_heap_push(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, offset_layseqr(kb->seqs, sl),
+					num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+			}
+			if(sc && sc->rend > sc->rbeg){
+				u4i scidx;
+				scidx = offset_layseqr(kb->seqs, sl);
+				sc = _push_padding_ref_kbmblock(kb, sc);
+				sl = ref_layseqr(kb->seqs, scidx);
+				if(kb->lidx == MAX_U4){
+					kb->lidx = offset_layseqr(kb->seqs, sc);
+				}
+				array_heap_push(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, offset_layseqr(kb->seqs, sc),
+					num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+			}
+		}
+		if(kb->heap->size == 0) break;
+		if(kb->lidx != MAX_U4){
+			sl = ref_layseqr(kb->seqs, kb->lidx);
+			kb->sidx = kb->heap->buffer[0];
+			sc = ref_layseqr(kb->seqs, kb->sidx);
+			if(sc->chridx == sl->chridx && sc->bidx + 1 >= sl->bidx){
+				kb->lidx = MAX_U4;
+				continue;
+			}
+			array_heap_remove(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, 0,
+				num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+			sc->rbeg -= sc->bidx * kb->bstep;
+			sc->rend -= sc->bidx * kb->bstep;
+		} else {
+			kb->sidx = kb->heap->buffer[0];
+			array_heap_remove(kb->heap->buffer, kb->heap->size, kb->heap->cap, u4i, 0,
+				num_cmpx(ref_layseqr(kb->seqs, a)->bidx, ref_layseqr(kb->seqs, b)->bidx, ref_layseqr(kb->seqs, a)->rdidx, ref_layseqr(kb->seqs, b)->rdidx));
+			sc = ref_layseqr(kb->seqs, kb->sidx);
+			sc->rbeg -= sc->bidx * kb->bstep;
+			sc->rend -= sc->bidx * kb->bstep;
+		}
+		return sc;
+	} while(1);
+	kb->sidx = MAX_U4;
+	return NULL;
+}
+
+static inline void info_kbmblock(void *obj, lay_seq_t *sq, lay_blk_t *bk){
+	KBMBlock *kb;
+	kb = (KBMBlock*)obj;
+	if(sq == NULL) return;
+	bk->node1 = sq->bidx;
+	bk->node2 = sq->bidx + 1;
+	bk->reftag = kb->rdtag->string;
+	bk->reflen = kb->rdlen;
+	bk->refoff = sq->bidx * kb->bstep;
+}
+
+static inline int map_kbmpoa(CTGCNS *cc, KBMAux *aux, char *rdtag, u4i qidx, BaseBank *rdseq, u8i seqoff, u4i seqlen, u4i corr_min, u4i corr_max, float corr_cov, FILE *layf){
+	ctg_cns_t *ctg;
+	KBMBlock *kb;
+	u4i i;
+	int self_aln, max_hit, min_aln, min_mat;
+	kb = (KBMBlock*)cc->obj;
+	reset_ctgcns(cc, kb, iter_kbmblock, info_kbmblock);
+	seqlen = kbm_cvt_length(seqlen);
+	if(seqlen < 4 * KBM_BIN_SIZE + UInt(aux->par->min_aln)) return 0;
+	if(rdseq && rdseq != aux->kbm->rdseqs){
+		self_aln = 0;
+	} else {
+		self_aln = 1;
+		rdseq = aux->kbm->rdseqs;
+	}
+	max_hit  = aux->par->max_hit;
+	min_aln  = aux->par->min_aln;
+	min_mat  = aux->par->min_mat;
+	aux->par->self_aln = 0;
+	aux->par->max_hit = corr_max;
+	aux->par->min_aln = num_max(seqlen * corr_cov, min_aln);
+	aux->par->min_mat = num_max(aux->par->min_aln * aux->par->min_sim, min_mat);
+	query_index_kbm(aux, rdtag, qidx, rdseq, seqoff, seqlen);
+	map_kbm(aux);
+	aux->par->self_aln = 0;
+	aux->par->max_hit  = max_hit;
+	aux->par->min_aln  = min_aln;
+	aux->par->min_mat  = min_mat;
+	if(KBM_LOG){
+		fprintf(KBM_LOGF, ">>> Contained alignments\n"); fflush(KBM_LOGF);
+		for(i=0;i<aux->hits->size;i++){
+			fprint_hit_kbm(aux, i, KBM_LOGF);
+		}
+	}
+	if(aux->hits->size < corr_min){
+		return 0;
+	}
+	if(self_aln){
+		for(i=0;i<aux->hits->size;i++){
+			if(aux->hits->buffer[i].tidx == qidx){
+				aux->hits->buffer[i].mat = 0;
+			}
+		}
+	}
+	reset_kbmblock(kb, rdtag, qidx, rdseq, seqoff, seqlen, aux);
+	if(layf){
+		print_lays_ctgcns(cc, layf);
+		fflush(layf);
+		reset_kbmblock(kb, rdtag, qidx, rdseq, seqoff, seqlen, aux);
+		reset_ctgcns(cc, kb, iter_kbmblock, info_kbmblock);
+	}
+	if((ctg = iter_ctgcns(cc))){
+		if(KBM_LOG){
+			fprintf(KBM_LOGF, ">%s corrected\n", ctg->tag->string);
+			print_lines_basebank(ctg->cns, 0, ctg->cns->size, KBM_LOGF, 100);
+			fflush(KBM_LOGF);
+		}
+	} else {
+		return 0;
+	}
+	clear_kbmmapv(aux->hits);
+	if(ctg->cns->size > seqlen){
+		ctg->cns->size = seqlen;
+		normalize_basebank(ctg->cns);
+	} else if(ctg->cns->size < seqlen){
+		ctg->cns->size = kbm_cvt_length(ctg->cns->size);
+		normalize_basebank(ctg->cns);
+	}
+	if(ctg->cns->size == 0){
+		repay_ctgcns(cc, ctg);
+		return 0;
+	}
+	query_index_kbm(aux, rdtag, qidx, ctg->cns, 0, ctg->cns->size);
+	map_kbm(aux);
+	repay_ctgcns(cc, ctg); // Please make sure ctg is not used unless this function return
+	return 1;
+}
+
+#endif

ksw.c

@@ -0,0 +1,693 @@
+/* The MIT License
+
+   Copyright (c) 2011 by Attractive Chaos <attractor@live.co.uk>
+
+   Permission is hereby granted, free of charge, to any person obtaining
+   a copy of this software and associated documentation files (the
+   "Software"), to deal in the Software without restriction, including
+   without limitation the rights to use, copy, modify, merge, publish,
+   distribute, sublicense, and/or sell copies of the Software, and to
+   permit persons to whom the Software is furnished to do so, subject to
+   the following conditions:
+
+   The above copyright notice and this permission notice shall be
+   included in all copies or substantial portions of the Software.
+
+   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+   EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+   MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+   NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+   BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+   ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+   CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+   SOFTWARE.
+*/
+
+#include <stdlib.h>
+#include <stdint.h>
+#include <emmintrin.h>
+#include "ksw.h"
+
+#ifdef USE_MALLOC_WRAPPERS
+#  include "malloc_wrap.h"
+#endif
+
+#ifdef __GNUC__
+#define LIKELY(x) __builtin_expect((x),1)
+#define UNLIKELY(x) __builtin_expect((x),0)
+#else
+#define LIKELY(x) (x)
+#define UNLIKELY(x) (x)
+#endif
+
+const kswr_t g_defr = { 0, -1, -1, -1, -1, -1, -1 };
+
+struct _kswq_t {
+	int qlen, slen;
+	uint8_t shift, mdiff, max, size;
+	__m128i *qp, *H0, *H1, *E, *Hmax;
+};
+
+/**
+ * Initialize the query data structure
+ *
+ * @param size   Number of bytes used to store a score; valid valures are 1 or 2
+ * @param qlen   Length of the query sequence
+ * @param query  Query sequence
+ * @param m      Size of the alphabet
+ * @param mat    Scoring matrix in a one-dimension array
+ *
+ * @return       Query data structure
+ */
+kswq_t *ksw_qinit(int size, int qlen, const uint8_t *query, int m, const int8_t *mat)
+{
+	kswq_t *q;
+	int slen, a, tmp, p;
+
+	size = size > 1? 2 : 1;
+	p = 8 * (3 - size); // # values per __m128i
+	slen = (qlen + p - 1) / p; // segmented length
+	q = (kswq_t*)malloc(sizeof(kswq_t) + 256 + 16 * slen * (m + 4)); // a single block of memory
+	q->qp = (__m128i*)(((size_t)q + sizeof(kswq_t) + 15) >> 4 << 4); // align memory
+	q->H0 = q->qp + slen * m;
+	q->H1 = q->H0 + slen;
+	q->E  = q->H1 + slen;
+	q->Hmax = q->E + slen;
+	q->slen = slen; q->qlen = qlen; q->size = size;
+	// compute shift
+	tmp = m * m;
+	for (a = 0, q->shift = 127, q->mdiff = 0; a < tmp; ++a) { // find the minimum and maximum score
+		if (mat[a] < (int8_t)q->shift) q->shift = mat[a];
+		if (mat[a] > (int8_t)q->mdiff) q->mdiff = mat[a];
+	}
+	q->max = q->mdiff;
+	q->shift = 256 - q->shift; // NB: q->shift is uint8_t
+	q->mdiff += q->shift; // this is the difference between the min and max scores
+	// An example: p=8, qlen=19, slen=3 and segmentation:
+	//  {{0,3,6,9,12,15,18,-1},{1,4,7,10,13,16,-1,-1},{2,5,8,11,14,17,-1,-1}}
+	if (size == 1) {
+		int8_t *t = (int8_t*)q->qp;
+		for (a = 0; a < m; ++a) {
+			int i, k, nlen = slen * p;
+			const int8_t *ma = mat + a * m;
+			for (i = 0; i < slen; ++i)
+				for (k = i; k < nlen; k += slen) // p iterations
+					*t++ = (k >= qlen? 0 : ma[query[k]]) + q->shift;
+		}
+	} else {
+		int16_t *t = (int16_t*)q->qp;
+		for (a = 0; a < m; ++a) {
+			int i, k, nlen = slen * p;
+			const int8_t *ma = mat + a * m;
+			for (i = 0; i < slen; ++i)
+				for (k = i; k < nlen; k += slen) // p iterations
+					*t++ = (k >= qlen? 0 : ma[query[k]]);
+		}
+	}
+	return q;
+}
+
+kswr_t ksw_u8(kswq_t *q, int tlen, const uint8_t *target, int _o_del, int _e_del, int _o_ins, int _e_ins, int xtra) // the first gap costs -(_o+_e)
+{
+	int slen, i, m_b, n_b, te = -1, gmax = 0, minsc, endsc;
+	uint64_t *b;
+	__m128i zero, oe_del, e_del, oe_ins, e_ins, shift, *H0, *H1, *E, *Hmax;
+	kswr_t r;
+
+#define __max_16(ret, xx) do { \
+		(xx) = _mm_max_epu8((xx), _mm_srli_si128((xx), 8)); \
+		(xx) = _mm_max_epu8((xx), _mm_srli_si128((xx), 4)); \
+		(xx) = _mm_max_epu8((xx), _mm_srli_si128((xx), 2)); \
+		(xx) = _mm_max_epu8((xx), _mm_srli_si128((xx), 1)); \
+    	(ret) = _mm_extract_epi16((xx), 0) & 0x00ff; \
+	} while (0)
+
+	// initialization
+	r = g_defr;
+	minsc = (xtra&KSW_XSUBO)? xtra&0xffff : 0x10000;
+	endsc = (xtra&KSW_XSTOP)? xtra&0xffff : 0x10000;
+	m_b = n_b = 0; b = 0;
+	zero = _mm_set1_epi32(0);
+	oe_del = _mm_set1_epi8(_o_del + _e_del);
+	e_del = _mm_set1_epi8(_e_del);
+	oe_ins = _mm_set1_epi8(_o_ins + _e_ins);
+	e_ins = _mm_set1_epi8(_e_ins);
+	shift = _mm_set1_epi8(q->shift);
+	H0 = q->H0; H1 = q->H1; E = q->E; Hmax = q->Hmax;
+	slen = q->slen;
+	for (i = 0; i < slen; ++i) {
+		_mm_store_si128(E + i, zero);
+		_mm_store_si128(H0 + i, zero);
+		_mm_store_si128(Hmax + i, zero);
+	}
+	// the core loop
+	for (i = 0; i < tlen; ++i) {
+		int j, k, cmp, imax;
+		__m128i e, h, t, f = zero, max = zero, *S = q->qp + target[i] * slen; // s is the 1st score vector
+		h = _mm_load_si128(H0 + slen - 1); // h={2,5,8,11,14,17,-1,-1} in the above example
+		h = _mm_slli_si128(h, 1); // h=H(i-1,-1); << instead of >> because x64 is little-endian
+		for (j = 0; LIKELY(j < slen); ++j) {
+			/* SW cells are computed in the following order:
+			 *   H(i,j)   = max{H(i-1,j-1)+S(i,j), E(i,j), F(i,j)}
+			 *   E(i+1,j) = max{H(i,j)-q, E(i,j)-r}
+			 *   F(i,j+1) = max{H(i,j)-q, F(i,j)-r}
+			 */
+			// compute H'(i,j); note that at the beginning, h=H'(i-1,j-1)
+			h = _mm_adds_epu8(h, _mm_load_si128(S + j));
+			h = _mm_subs_epu8(h, shift); // h=H'(i-1,j-1)+S(i,j)
+			e = _mm_load_si128(E + j); // e=E'(i,j)
+			h = _mm_max_epu8(h, e);
+			h = _mm_max_epu8(h, f); // h=H'(i,j)
+			max = _mm_max_epu8(max, h); // set max
+			_mm_store_si128(H1 + j, h); // save to H'(i,j)
+			// now compute E'(i+1,j)
+			e = _mm_subs_epu8(e, e_del); // e=E'(i,j) - e_del
+			t = _mm_subs_epu8(h, oe_del); // h=H'(i,j) - o_del - e_del
+			e = _mm_max_epu8(e, t); // e=E'(i+1,j)
+			_mm_store_si128(E + j, e); // save to E'(i+1,j)
+			// now compute F'(i,j+1)
+			f = _mm_subs_epu8(f, e_ins);
+			t = _mm_subs_epu8(h, oe_ins); // h=H'(i,j) - o_ins - e_ins
+			f = _mm_max_epu8(f, t);
+			// get H'(i-1,j) and prepare for the next j
+			h = _mm_load_si128(H0 + j); // h=H'(i-1,j)
+		}
+		// NB: we do not need to set E(i,j) as we disallow adjecent insertion and then deletion
+		for (k = 0; LIKELY(k < 16); ++k) { // this block mimics SWPS3; NB: H(i,j) updated in the lazy-F loop cannot exceed max
+			f = _mm_slli_si128(f, 1);
+			for (j = 0; LIKELY(j < slen); ++j) {
+				h = _mm_load_si128(H1 + j);
+				h = _mm_max_epu8(h, f); // h=H'(i,j)
+				_mm_store_si128(H1 + j, h);
+				h = _mm_subs_epu8(h, oe_ins);
+				f = _mm_subs_epu8(f, e_ins);
+				cmp = _mm_movemask_epi8(_mm_cmpeq_epi8(_mm_subs_epu8(f, h), zero));
+				if (UNLIKELY(cmp == 0xffff)) goto end_loop16;
+			}
+		}
+end_loop16:
+		//int k;for (k=0;k<16;++k)printf("%d ", ((uint8_t*)&max)[k]);printf("\n");
+		__max_16(imax, max); // imax is the maximum number in max
+		if (imax >= minsc) { // write the b array; this condition adds branching unfornately
+			if (n_b == 0 || (int32_t)b[n_b-1] + 1 != i) { // then append
+				if (n_b == m_b) {
+					m_b = m_b? m_b<<1 : 8;
+					b = (uint64_t*)realloc(b, 8 * m_b);
+				}
+				b[n_b++] = (uint64_t)imax<<32 | i;
+			} else if ((int)(b[n_b-1]>>32) < imax) b[n_b-1] = (uint64_t)imax<<32 | i; // modify the last
+		}
+		if (imax > gmax) {
+			gmax = imax; te = i; // te is the end position on the target
+			for (j = 0; LIKELY(j < slen); ++j) // keep the H1 vector
+				_mm_store_si128(Hmax + j, _mm_load_si128(H1 + j));
+			if (gmax + q->shift >= 255 || gmax >= endsc) break;
+		}
+		S = H1; H1 = H0; H0 = S; // swap H0 and H1
+	}
+	r.score = gmax + q->shift < 255? gmax : 255;
+	r.te = te;
+	// modified by Jue Ruan
+	//if (r.score != 255) { // get a->qe, the end of query match; find the 2nd best score
+	{ // get a->qe, the end of query match; find the 2nd best score
+		int max = -1, tmp, low, high, qlen = slen * 16;
+		uint8_t *t = (uint8_t*)Hmax;
+		for (i = 0; i < qlen; ++i, ++t)
+			if ((int)*t > max) max = *t, r.qe = i / 16 + i % 16 * slen;
+			else if ((int)*t == max && (tmp = i / 16 + i % 16 * slen) < r.qe) r.qe = tmp; 
+		//printf("%d,%d\n", max, gmax);
+		if (b) {
+			i = (r.score + q->max - 1) / q->max;
+			low = te - i; high = te + i;
+			for (i = 0; i < n_b; ++i) {
+				int e = (int32_t)b[i];
+				if ((e < low || e > high) && (int)(b[i]>>32) > r.score2)
+					r.score2 = b[i]>>32, r.te2 = e;
+			}
+		}
+	}
+	free(b);
+	return r;
+}
+
+kswr_t ksw_i16(kswq_t *q, int tlen, const uint8_t *target, int _o_del, int _e_del, int _o_ins, int _e_ins, int xtra) // the first gap costs -(_o+_e)
+{
+	int slen, i, m_b, n_b, te = -1, gmax = 0, minsc, endsc;
+	uint64_t *b;
+	__m128i zero, oe_del, e_del, oe_ins, e_ins, *H0, *H1, *E, *Hmax;
+	kswr_t r;
+
+#define __max_8(ret, xx) do { \
+		(xx) = _mm_max_epi16((xx), _mm_srli_si128((xx), 8)); \
+		(xx) = _mm_max_epi16((xx), _mm_srli_si128((xx), 4)); \
+		(xx) = _mm_max_epi16((xx), _mm_srli_si128((xx), 2)); \
+    	(ret) = _mm_extract_epi16((xx), 0); \
+	} while (0)
+
+	// initialization
+	r = g_defr;
+	minsc = (xtra&KSW_XSUBO)? xtra&0xffff : 0x10000;
+	endsc = (xtra&KSW_XSTOP)? xtra&0xffff : 0x10000;
+	m_b = n_b = 0; b = 0;
+	zero = _mm_set1_epi32(0);
+	oe_del = _mm_set1_epi16(_o_del + _e_del);
+	e_del = _mm_set1_epi16(_e_del);
+	oe_ins = _mm_set1_epi16(_o_ins + _e_ins);
+	e_ins = _mm_set1_epi16(_e_ins);
+	H0 = q->H0; H1 = q->H1; E = q->E; Hmax = q->Hmax;
+	slen = q->slen;
+	for (i = 0; i < slen; ++i) {
+		_mm_store_si128(E + i, zero);
+		_mm_store_si128(H0 + i, zero);
+		_mm_store_si128(Hmax + i, zero);
+	}
+	// the core loop
+	for (i = 0; i < tlen; ++i) {
+		int j, k, imax;
+		__m128i e, t, h, f = zero, max = zero, *S = q->qp + target[i] * slen; // s is the 1st score vector
+		h = _mm_load_si128(H0 + slen - 1); // h={2,5,8,11,14,17,-1,-1} in the above example
+		h = _mm_slli_si128(h, 2);
+		for (j = 0; LIKELY(j < slen); ++j) {
+			h = _mm_adds_epi16(h, *S++);
+			e = _mm_load_si128(E + j);
+			h = _mm_max_epi16(h, e);
+			h = _mm_max_epi16(h, f);
+			max = _mm_max_epi16(max, h);
+			_mm_store_si128(H1 + j, h);
+			e = _mm_subs_epu16(e, e_del);
+			t = _mm_subs_epu16(h, oe_del);
+			e = _mm_max_epi16(e, t);
+			_mm_store_si128(E + j, e);
+			f = _mm_subs_epu16(f, e_ins);
+			t = _mm_subs_epu16(h, oe_ins);
+			f = _mm_max_epi16(f, t);
+			h = _mm_load_si128(H0 + j);
+		}
+		for (k = 0; LIKELY(k < 16); ++k) {
+			f = _mm_slli_si128(f, 2);
+			for (j = 0; LIKELY(j < slen); ++j) {
+				h = _mm_load_si128(H1 + j);
+				h = _mm_max_epi16(h, f);
+				_mm_store_si128(H1 + j, h);
+				h = _mm_subs_epu16(h, oe_ins);
+				f = _mm_subs_epu16(f, e_ins);
+				if(UNLIKELY(!_mm_movemask_epi8(_mm_cmpgt_epi16(f, h)))) goto end_loop8;
+			}
+		}
+end_loop8:
+		__max_8(imax, max);
+		if (imax >= minsc) {
+			if (n_b == 0 || (int32_t)b[n_b-1] + 1 != i) {
+				if (n_b == m_b) {
+					m_b = m_b? m_b<<1 : 8;
+					b = (uint64_t*)realloc(b, 8 * m_b);
+				}
+				b[n_b++] = (uint64_t)imax<<32 | i;
+			} else if ((int)(b[n_b-1]>>32) < imax) b[n_b-1] = (uint64_t)imax<<32 | i; // modify the last
+		}
+		if (imax > gmax) {
+			gmax = imax; te = i;
+			for (j = 0; LIKELY(j < slen); ++j)
+				_mm_store_si128(Hmax + j, _mm_load_si128(H1 + j));
+			if (gmax >= endsc) break;
+		}
+		S = H1; H1 = H0; H0 = S;
+	}
+	r.score = gmax; r.te = te;
+	{
+		int max = -1, tmp, low, high, qlen = slen * 8;
+		uint16_t *t = (uint16_t*)Hmax;
+		for (i = 0, r.qe = -1; i < qlen; ++i, ++t)
+			if ((int)*t > max) max = *t, r.qe = i / 8 + i % 8 * slen;
+			else if ((int)*t == max && (tmp = i / 8 + i % 8 * slen) < r.qe) r.qe = tmp; 
+		if (b) {
+			i = (r.score + q->max - 1) / q->max;
+			low = te - i; high = te + i;
+			for (i = 0; i < n_b; ++i) {
+				int e = (int32_t)b[i];
+				if ((e < low || e > high) && (int)(b[i]>>32) > r.score2)
+					r.score2 = b[i]>>32, r.te2 = e;
+			}
+		}
+	}
+	free(b);
+	return r;
+}
+
+static inline void revseq(int l, uint8_t *s)
+{
+	int i, t;
+	for (i = 0; i < l>>1; ++i)
+		t = s[i], s[i] = s[l - 1 - i], s[l - 1 - i] = t;
+}
+
+kswr_t ksw_align2(int qlen, uint8_t *query, int tlen, uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int xtra, kswq_t **qry)
+{
+	int size;
+	kswq_t *q;
+	kswr_t r, rr;
+	kswr_t (*func)(kswq_t*, int, const uint8_t*, int, int, int, int, int);
+
+	q = (qry && *qry)? *qry : ksw_qinit((xtra&KSW_XBYTE)? 1 : 2, qlen, query, m, mat);
+	if (qry && *qry == 0) *qry = q;
+	func = q->size == 2? ksw_i16 : ksw_u8;
+	size = q->size;
+	r = func(q, tlen, target, o_del, e_del, o_ins, e_ins, xtra);
+	if (qry == 0) free(q);
+	if ((xtra&KSW_XSTART) == 0 || ((xtra&KSW_XSUBO) && r.score < (xtra&0xffff))) return r;
+	revseq(r.qe + 1, query); revseq(r.te + 1, target); // +1 because qe/te points to the exact end, not the position after the end
+	q = ksw_qinit(size, r.qe + 1, query, m, mat);
+	rr = func(q, tlen, target, o_del, e_del, o_ins, e_ins, KSW_XSTOP | r.score);
+	revseq(r.qe + 1, query); revseq(r.te + 1, target);
+	free(q);
+	if (r.score == rr.score)
+		r.tb = r.te - rr.te, r.qb = r.qe - rr.qe;
+	return r;
+}
+
+kswr_t ksw_align(int qlen, uint8_t *query, int tlen, uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int xtra, kswq_t **qry)
+{
+	return ksw_align2(qlen, query, tlen, target, m, mat, gapo, gape, gapo, gape, xtra, qry);
+}
+
+/********************
+ *** SW extension ***
+ ********************/
+
+typedef struct {
+	int32_t h, e;
+} eh_t;
+
+int ksw_extend2(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int w, int end_bonus, int zdrop, int h0, int *_qle, int *_tle, int *_gtle, int *_gscore, int *_max_off)
+{
+	eh_t *eh; // score array
+	int8_t *qp; // query profile
+	int i, j, k, oe_del = o_del + e_del, oe_ins = o_ins + e_ins, beg, end, max, max_i, max_j, max_ins, max_del, max_ie, gscore, max_off;
+	if (h0 < 0) h0 = 0;
+	// allocate memory
+	qp = malloc(qlen * m);
+	eh = calloc(qlen + 1, 8);
+	// generate the query profile
+	for (k = i = 0; k < m; ++k) {
+		const int8_t *p = &mat[k * m];
+		for (j = 0; j < qlen; ++j) qp[i++] = p[query[j]];
+	}
+	// fill the first row
+	eh[0].h = h0; eh[1].h = h0 > oe_ins? h0 - oe_ins : 0;
+	for (j = 2; j <= qlen && eh[j-1].h > e_ins; ++j)
+		eh[j].h = eh[j-1].h - e_ins;
+	// adjust $w if it is too large
+	k = m * m;
+	for (i = 0, max = 0; i < k; ++i) // get the max score
+		max = max > mat[i]? max : mat[i];
+	max_ins = (int)((double)(qlen * max + end_bonus - o_ins) / e_ins + 1.);
+	max_ins = max_ins > 1? max_ins : 1;
+	w = w < max_ins? w : max_ins;
+	max_del = (int)((double)(qlen * max + end_bonus - o_del) / e_del + 1.);
+	max_del = max_del > 1? max_del : 1;
+	w = w < max_del? w : max_del; // TODO: is this necessary?
+	// DP loop
+	max = h0, max_i = max_j = -1; max_ie = -1, gscore = -1;
+	max_off = 0;
+	beg = 0, end = qlen;
+	for (i = 0; LIKELY(i < tlen); ++i) {
+		int t, f = 0, h1, m = 0, mj = -1;
+		int8_t *q = &qp[target[i] * qlen];
+		// compute the first column
+		h1 = h0 - (o_del + e_del * (i + 1));
+		if (h1 < 0) h1 = 0;
+		// apply the band and the constraint (if provided)
+		if (beg < i - w) beg = i - w;
+		if (end > i + w + 1) end = i + w + 1;
+		if (end > qlen) end = qlen;
+		for (j = beg; LIKELY(j < end); ++j) {
+			// At the beginning of the loop: eh[j] = { H(i-1,j-1), E(i,j) }, f = F(i,j) and h1 = H(i,j-1)
+			// Similar to SSE2-SW, cells are computed in the following order:
+			//   H(i,j)   = max{H(i-1,j-1)+S(i,j), E(i,j), F(i,j)}
+			//   E(i+1,j) = max{H(i,j)-gapo, E(i,j)} - gape
+			//   F(i,j+1) = max{H(i,j)-gapo, F(i,j)} - gape
+			eh_t *p = &eh[j];
+			int h, M = p->h, e = p->e; // get H(i-1,j-1) and E(i-1,j)
+			p->h = h1;          // set H(i,j-1) for the next row
+			M += q[j];          // separating H and M to disallow a cigar like "100M3I3D20M"
+			h = M > e? M : e;
+			h = h > f? h : f;
+			h1 = h;             // save H(i,j) to h1 for the next column
+			mj = m > h? mj : j; // record the position where max score is achieved
+			m = m > h? m : h;   // m is stored at eh[mj+1]
+			t = M - oe_del;
+			t = t > 0? t : 0;
+			e -= e_del;
+			e = e > t? e : t;   // computed E(i+1,j)
+			p->e = e;           // save E(i+1,j) for the next row
+			t = M - oe_ins;
+			t = t > 0? t : 0;
+			f -= e_ins;
+			f = f > t? f : t;   // computed F(i,j+1)
+		}
+		eh[end].h = h1; eh[end].e = 0;
+		if (j == qlen) {
+			max_ie = gscore > h1? max_ie : i;
+			gscore = gscore > h1? gscore : h1;
+		}
+		if (m == 0) break;
+		if (m > max) {
+			max = m, max_i = i, max_j = mj;
+			max_off = max_off > abs(mj - i)? max_off : abs(mj - i);
+		} else if (zdrop > 0) {
+			if (i - max_i > mj - max_j) {
+				if (max - m - ((i - max_i) - (mj - max_j)) * e_del > zdrop) break;
+			} else {
+				if (max - m - ((mj - max_j) - (i - max_i)) * e_ins > zdrop) break;
+			}
+		}
+		// update beg and end for the next round
+		for (j = mj; j >= beg && eh[j].h; --j);
+		beg = j + 1;
+		for (j = mj + 2; j <= end && eh[j].h; ++j);
+		end = j;
+		//beg = 0; end = qlen; // uncomment this line for debugging
+	}
+	free(eh); free(qp);
+	if (_qle) *_qle = max_j + 1;
+	if (_tle) *_tle = max_i + 1;
+	if (_gtle) *_gtle = max_ie + 1;
+	if (_gscore) *_gscore = gscore;
+	if (_max_off) *_max_off = max_off;
+	return max;
+}
+
+int ksw_extend(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int w, int end_bonus, int zdrop, int h0, int *qle, int *tle, int *gtle, int *gscore, int *max_off)
+{
+	return ksw_extend2(qlen, query, tlen, target, m, mat, gapo, gape, gapo, gape, w, end_bonus, zdrop, h0, qle, tle, gtle, gscore, max_off);
+}
+
+/********************
+ * Global alignment *
+ ********************/
+
+#define MINUS_INF -0x40000000
+
+static inline uint32_t *push_cigar(int *n_cigar, int *m_cigar, uint32_t *cigar, int op, int len)
+{
+	if (*n_cigar == 0 || op != (int)(cigar[(*n_cigar) - 1]&0xf)) {
+		if (*n_cigar == *m_cigar) {
+			*m_cigar = *m_cigar? (*m_cigar)<<1 : 4;
+			cigar = realloc(cigar, (*m_cigar) << 2);
+		}
+		cigar[(*n_cigar)++] = len<<4 | op;
+	} else cigar[(*n_cigar)-1] += len<<4;
+	return cigar;
+}
+
+int ksw_global2(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int w, int *n_cigar_, uint32_t **cigar_)
+{
+	eh_t *eh;
+	int8_t *qp; // query profile
+	int i, j, k, oe_del = o_del + e_del, oe_ins = o_ins + e_ins, score, n_col;
+	uint8_t *z; // backtrack matrix; in each cell: f<<4|e<<2|h; in principle, we can halve the memory, but backtrack will be a little more complex
+	if (n_cigar_) *n_cigar_ = 0;
+	// allocate memory
+	n_col = qlen < 2*w+1? qlen : 2*w+1; // maximum #columns of the backtrack matrix
+	z = malloc(n_col * tlen);
+	qp = malloc(qlen * m);
+	eh = calloc(qlen + 1, 8);
+	// generate the query profile
+	for (k = i = 0; k < m; ++k) {
+		const int8_t *p = &mat[k * m];
+		for (j = 0; j < qlen; ++j) qp[i++] = p[query[j]];
+	}
+	// fill the first row
+	eh[0].h = 0; eh[0].e = MINUS_INF;
+	for (j = 1; j <= qlen && j <= w; ++j)
+		eh[j].h = -(o_ins + e_ins * j), eh[j].e = MINUS_INF;
+	for (; j <= qlen; ++j) eh[j].h = eh[j].e = MINUS_INF; // everything is -inf outside the band
+	// DP loop
+	for (i = 0; LIKELY(i < tlen); ++i) { // target sequence is in the outer loop
+		int32_t f = MINUS_INF, h1, beg, end, t;
+		int8_t *q = &qp[target[i] * qlen];
+		uint8_t *zi = &z[i * n_col];
+		beg = i > w? i - w : 0;
+		end = i + w + 1 < qlen? i + w + 1 : qlen; // only loop through [beg,end) of the query sequence
+		h1 = beg == 0? -(o_del + e_del * (i + 1)) : MINUS_INF;
+		for (j = beg; LIKELY(j < end); ++j) {
+			// At the beginning of the loop: eh[j] = { H(i-1,j-1), E(i,j) }, f = F(i,j) and h1 = H(i,j-1)
+			// Cells are computed in the following order:
+			//   M(i,j)   = H(i-1,j-1) + S(i,j)
+			//   H(i,j)   = max{M(i,j), E(i,j), F(i,j)}
+			//   E(i+1,j) = max{M(i,j)-gapo, E(i,j)} - gape
+			//   F(i,j+1) = max{M(i,j)-gapo, F(i,j)} - gape
+			// We have to separate M(i,j); otherwise the direction may not be recorded correctly.
+			// However, a CIGAR like "10M3I3D10M" allowed by local() is disallowed by global().
+			// Such a CIGAR may occur, in theory, if mismatch_penalty > 2*gap_ext_penalty + 2*gap_open_penalty/k.
+			// In practice, this should happen very rarely given a reasonable scoring system.
+			eh_t *p = &eh[j];
+			int32_t h, m = p->h, e = p->e;
+			uint8_t d; // direction
+			p->h = h1;
+			m += q[j];
+			d = m >= e? 0 : 1;
+			h = m >= e? m : e;
+			d = h >= f? d : 2;
+			h = h >= f? h : f;
+			h1 = h;
+			t = m - oe_del;
+			e -= e_del;
+			d |= e > t? 1<<2 : 0;
+			e  = e > t? e    : t;
+			p->e = e;
+			t = m - oe_ins;
+			f -= e_ins;
+			d |= f > t? 2<<4 : 0; // if we want to halve the memory, use one bit only, instead of two
+			f  = f > t? f    : t;
+			zi[j - beg] = d; // z[i,j] keeps h for the current cell and e/f for the next cell
+		}
+		eh[end].h = h1; eh[end].e = MINUS_INF;
+	}
+	score = eh[qlen].h;
+	if (n_cigar_ && cigar_) { // backtrack
+		int n_cigar = 0, m_cigar = 0, which = 0;
+		uint32_t *cigar = 0, tmp;
+		i = tlen - 1; k = (i + w + 1 < qlen? i + w + 1 : qlen) - 1; // (i,k) points to the last cell
+		while (i >= 0 && k >= 0) {
+			which = z[i * n_col + (k - (i > w? i - w : 0))] >> (which<<1) & 3;
+			if (which == 0)      cigar = push_cigar(&n_cigar, &m_cigar, cigar, 0, 1), --i, --k;
+			else if (which == 1) cigar = push_cigar(&n_cigar, &m_cigar, cigar, 2, 1), --i;
+			else                 cigar = push_cigar(&n_cigar, &m_cigar, cigar, 1, 1), --k;
+		}
+		if (i >= 0) cigar = push_cigar(&n_cigar, &m_cigar, cigar, 2, i + 1);
+		if (k >= 0) cigar = push_cigar(&n_cigar, &m_cigar, cigar, 1, k + 1);
+		for (i = 0; i < n_cigar>>1; ++i) // reverse CIGAR
+			tmp = cigar[i], cigar[i] = cigar[n_cigar-1-i], cigar[n_cigar-1-i] = tmp;
+		*n_cigar_ = n_cigar, *cigar_ = cigar;
+	}
+	free(eh); free(qp); free(z);
+	return score;
+}
+
+int ksw_global(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int w, int *n_cigar_, uint32_t **cigar_)
+{
+	return ksw_global2(qlen, query, tlen, target, m, mat, gapo, gape, gapo, gape, w, n_cigar_, cigar_);
+}
+
+/*******************************************
+ * Main function (not compiled by default) *
+ *******************************************/
+
+#ifdef _KSW_MAIN
+
+#include <unistd.h>
+#include <stdio.h>
+#include <zlib.h>
+#include "kseq.h"
+KSEQ_INIT(gzFile, err_gzread)
+
+unsigned char seq_nt4_table[256] = {
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 0, 4, 1,  4, 4, 4, 2,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  3, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 0, 4, 1,  4, 4, 4, 2,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  3, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
+	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4
+};
+
+int main(int argc, char *argv[])
+{
+	int c, sa = 1, sb = 3, i, j, k, forward_only = 0, max_rseq = 0;
+	int8_t mat[25];
+	int gapo = 5, gape = 2, minsc = 0, xtra = KSW_XSTART;
+	uint8_t *rseq = 0;
+	gzFile fpt, fpq;
+	kseq_t *kst, *ksq;
+
+	// parse command line
+	while ((c = getopt(argc, argv, "a:b:q:r:ft:1")) >= 0) {
+		switch (c) {
+			case 'a': sa = atoi(optarg); break;
+			case 'b': sb = atoi(optarg); break;
+			case 'q': gapo = atoi(optarg); break;
+			case 'r': gape = atoi(optarg); break;
+			case 't': minsc = atoi(optarg); break;
+			case 'f': forward_only = 1; break;
+			case '1': xtra |= KSW_XBYTE; break;
+		}
+	}
+	if (optind + 2 > argc) {
+		fprintf(stderr, "Usage: ksw [-1] [-f] [-a%d] [-b%d] [-q%d] [-r%d] [-t%d] <target.fa> <query.fa>\n", sa, sb, gapo, gape, minsc);
+		return 1;
+	}
+	if (minsc > 0xffff) minsc = 0xffff;
+	xtra |= KSW_XSUBO | minsc;
+	// initialize scoring matrix
+	for (i = k = 0; i < 4; ++i) {
+		for (j = 0; j < 4; ++j)
+			mat[k++] = i == j? sa : -sb;
+		mat[k++] = 0; // ambiguous base
+	}
+	for (j = 0; j < 5; ++j) mat[k++] = 0;
+	// open file
+	fpt = xzopen(argv[optind],   "r"); kst = kseq_init(fpt);
+	fpq = xzopen(argv[optind+1], "r"); ksq = kseq_init(fpq);
+	// all-pair alignment
+	while (kseq_read(ksq) > 0) {
+		kswq_t *q[2] = {0, 0};
+		kswr_t r;
+		for (i = 0; i < (int)ksq->seq.l; ++i) ksq->seq.s[i] = seq_nt4_table[(int)ksq->seq.s[i]];
+		if (!forward_only) { // reverse
+			if ((int)ksq->seq.m > max_rseq) {
+				max_rseq = ksq->seq.m;
+				rseq = (uint8_t*)realloc(rseq, max_rseq);
+			}
+			for (i = 0, j = ksq->seq.l - 1; i < (int)ksq->seq.l; ++i, --j)
+				rseq[j] = ksq->seq.s[i] == 4? 4 : 3 - ksq->seq.s[i];
+		}
+		gzrewind(fpt); kseq_rewind(kst);
+		while (kseq_read(kst) > 0) {
+			for (i = 0; i < (int)kst->seq.l; ++i) kst->seq.s[i] = seq_nt4_table[(int)kst->seq.s[i]];
+			r = ksw_align(ksq->seq.l, (uint8_t*)ksq->seq.s, kst->seq.l, (uint8_t*)kst->seq.s, 5, mat, gapo, gape, xtra, &q[0]);
+			if (r.score >= minsc)
+				err_printf("%s\t%d\t%d\t%s\t%d\t%d\t%d\t%d\t%d\n", kst->name.s, r.tb, r.te+1, ksq->name.s, r.qb, r.qe+1, r.score, r.score2, r.te2);
+			if (rseq) {
+				r = ksw_align(ksq->seq.l, rseq, kst->seq.l, (uint8_t*)kst->seq.s, 5, mat, gapo, gape, xtra, &q[1]);
+				if (r.score >= minsc)
+					err_printf("%s\t%d\t%d\t%s\t%d\t%d\t%d\t%d\t%d\n", kst->name.s, r.tb, r.te+1, ksq->name.s, (int)ksq->seq.l - r.qb, (int)ksq->seq.l - 1 - r.qe, r.score, r.score2, r.te2);
+			}
+		}
+		free(q[0]); free(q[1]);
+	}
+	free(rseq);
+	kseq_destroy(kst); err_gzclose(fpt);
+	kseq_destroy(ksq); err_gzclose(fpq);
+	return 0;
+}
+#endif

ksw.h

@@ -0,0 +1,114 @@
+#ifndef __AC_KSW_H
+#define __AC_KSW_H
+
+#include <stdint.h>
+
+#define KSW_XBYTE  0x10000
+#define KSW_XSTOP  0x20000
+#define KSW_XSUBO  0x40000
+#define KSW_XSTART 0x80000
+
+struct _kswq_t;
+typedef struct _kswq_t kswq_t;
+
+typedef struct {
+	int score; // best score
+	int te, qe; // target end and query end
+	int score2, te2; // second best score and ending position on the target
+	int tb, qb; // target start and query start
+} kswr_t;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+	/**
+	 * Aligning two sequences
+	 *
+	 * @param qlen    length of the query sequence (typically <tlen)
+	 * @param query   query sequence with 0 <= query[i] < m
+	 * @param tlen    length of the target sequence
+	 * @param target  target sequence
+	 * @param m       number of residue types
+	 * @param mat     m*m scoring matrix in one-dimension array
+	 * @param gapo    gap open penalty; a gap of length l cost "-(gapo+l*gape)"
+	 * @param gape    gap extension penalty
+	 * @param xtra    extra information (see below)
+	 * @param qry     query profile (see below)
+	 *
+	 * @return        alignment information in a struct; unset values to -1
+	 *
+	 * When xtra==0, ksw_align() uses a signed two-byte integer to store a
+	 * score and only finds the best score and the end positions. The 2nd best
+	 * score or the start positions are not attempted. The default behavior can
+	 * be tuned by setting KSW_X* flags:
+	 *
+	 *   KSW_XBYTE:  use an unsigned byte to store a score. If overflow occurs,
+	 *               kswr_t::score will be set to 255
+	 *
+	 *   KSW_XSUBO:  track the 2nd best score and the ending position on the
+	 *               target if the 2nd best is higher than (xtra&0xffff)
+	 *
+	 *   KSW_XSTOP:  stop if the maximum score is above (xtra&0xffff)
+	 *
+	 *   KSW_XSTART: find the start positions
+	 *
+	 * When *qry==NULL, ksw_align() will compute and allocate the query profile
+	 * and when the function returns, *qry will point to the profile, which can
+	 * be deallocated simply by free(). If one query is aligned against multiple
+	 * target sequences, *qry should be set to NULL during the first call and
+	 * freed after the last call. Note that qry can equal 0. In this case, the
+	 * query profile will be deallocated in ksw_align().
+	 */
+	kswr_t ksw_align(int qlen, uint8_t *query, int tlen, uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int xtra, kswq_t **qry);
+	kswr_t ksw_align2(int qlen, uint8_t *query, int tlen, uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int xtra, kswq_t **qry);
+
+	/**
+	 * Banded global alignment
+	 *
+	 * @param qlen    query length
+	 * @param query   query sequence with 0 <= query[i] < m
+	 * @param tlen    target length
+	 * @param target  target sequence with 0 <= target[i] < m
+	 * @param m       number of residue types
+	 * @param mat     m*m scoring mattrix in one-dimension array
+	 * @param gapo    gap open penalty; a gap of length l cost "-(gapo+l*gape)"
+	 * @param gape    gap extension penalty
+	 * @param w       band width
+	 * @param n_cigar (out) number of CIGAR elements
+	 * @param cigar   (out) BAM-encoded CIGAR; caller need to deallocate with free()
+	 *
+	 * @return        score of the alignment
+	 */
+	int ksw_global(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int w, int *n_cigar, uint32_t **cigar);
+	int ksw_global2(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int w, int *n_cigar, uint32_t **cigar);
+
+	/**
+	 * Extend alignment
+	 *
+	 * The routine aligns $query and $target, assuming their upstream sequences,
+	 * which are not provided, have been aligned with score $h0. In return,
+	 * region [0,*qle) on the query and [0,*tle) on the target sequences are
+	 * aligned together. If *gscore>=0, *gscore keeps the best score such that
+	 * the entire query sequence is aligned; *gtle keeps the position on the
+	 * target where *gscore is achieved. Returning *gscore and *gtle helps the
+	 * caller to decide whether an end-to-end hit or a partial hit is preferred.
+	 *
+	 * The first 9 parameters are identical to those in ksw_global()
+	 *
+	 * @param h0      alignment score of upstream sequences
+	 * @param _qle    (out) length of the query in the alignment
+	 * @param _tle    (out) length of the target in the alignment
+	 * @param _gtle   (out) length of the target if query is fully aligned
+	 * @param _gscore (out) score of the best end-to-end alignment; negative if not found
+	 *
+	 * @return        best semi-local alignment score
+	 */
+	int ksw_extend(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int gapo, int gape, int w, int end_bonus, int zdrop, int h0, int *qle, int *tle, int *gtle, int *gscore, int *max_off);
+	int ksw_extend2(int qlen, const uint8_t *query, int tlen, const uint8_t *target, int m, const int8_t *mat, int o_del, int e_del, int o_ins, int e_ins, int w, int end_bonus, int zdrop, int h0, int *qle, int *tle, int *gtle, int *gscore, int *max_off);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

list.h

@@ -0,0 +1,725 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __LIST_RJ_H
+#define __LIST_RJ_H
+
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdio.h>
+#include "sort.h"
+#include "mem_share.h"
+
+/**
+ * Heap macros
+ */
+
+//ary, size and cap must be explict variable, not expression
+#define array_heap_push(ary, size, cap, e_type, id, cmp_expr)\
+do {	\
+	e_type _pp_, _p_;	\
+	_p_ = (e_type)(id);	\
+	size_t i, j;	\
+	i = (size);	\
+	e_type a, b;	\
+	if((size_t)((size) + 1) > (size_t)(cap)){	\
+		if((size_t)((size) + 1) < 0xFFFFFFFFU){	\
+			(cap) = roundup_power2((size) + 1);	\
+		} else {	\
+			(cap) = (((size) + 1 + 0xFFFFFFFLLU - 1LLU) / 0xFFFFFFFLLU) * 0xFFFFFFFLLU;	\
+		}	\
+		(ary) = realloc((ary), sizeof(e_type) * (cap));	\
+		if((ary) == NULL){	\
+			fprintf(stderr, " -- Out of memory, try to allocate %llu bytes in %s, -- %s:%d --\n", (unsigned long long)(sizeof(e_type) * (cap)), __FUNCTION__, __FILE__, __LINE__);	\
+			print_backtrace(stderr, 20);	\
+			exit(1);	\
+		}	\
+	}	\
+	(ary)[(size)++] = _p_;	\
+	while(i){	\
+		j = (i - 1) >> 1;	\
+		a = (ary)[i]; b = (ary)[j];	\
+		if((cmp_expr) >= 0) break;	\
+		_pp_ = (ary)[i]; (ary)[i] = (ary)[j]; (ary)[j] = _pp_;	\
+		i = j;	\
+	}	\
+} while(0)
+
+#define array_heap_remove(ary, len, cap, e_type, _idx, cmp_expr)\
+do {	\
+	e_type _pp_;	\
+	size_t swap, idx;	\
+	idx = (size_t)(_idx);	\
+	(ary)[idx] = (ary)[--(len)];	\
+	e_type a, b;	\
+	while((size_t)((idx << 1) + 1) < (size_t)(len)){	\
+		swap = idx;	\
+		a = (ary)[swap]; b = (ary)[(idx << 1) + 1];	\
+		if((cmp_expr) > 0) swap = (idx << 1) + 1;	\
+		if(((idx << 1) + 2) < (size_t)(len)){	\
+			a = (ary)[swap]; b = (ary)[(idx << 1) + 2];	\
+			if((cmp_expr) > 0) swap = (idx << 1) + 2;	\
+		}	\
+		if(swap == idx) break;	\
+		_pp_ = (ary)[idx]; (ary)[idx] = (ary)[swap]; (ary)[swap] = _pp_;	\
+		idx = swap;	\
+	}	\
+} while(0)
+
+#define array_heap_replace(ary, len, cap, e_type, _idx, _val, cmp_expr)\
+do {	\
+	e_type _pp_;	\
+	size_t swap, idx;	\
+	idx = (size_t)(_idx);	\
+	(ary)[idx] = _val;	\
+	e_type a, b;	\
+	while((size_t)((idx << 1) + 1) < (size_t)(len)){	\
+		swap = idx;	\
+		a = (ary)[swap]; b = (ary)[(idx << 1) + 1];	\
+		if((cmp_expr) > 0) swap = (idx << 1) + 1;	\
+		if(((idx << 1) + 2) < (size_t)(len)){	\
+			a = (ary)[swap]; b = (ary)[(idx << 1) + 2];	\
+			if((cmp_expr) > 0) swap = (idx << 1) + 2;	\
+		}	\
+		if(swap == idx) break;	\
+		_pp_ = (ary)[idx]; (ary)[idx] = (ary)[swap]; (ary)[swap] = _pp_;	\
+		idx = swap;	\
+	}	\
+} while(0)
+
+#define array_heap_pop(ary, len, cap, e_type, cmp_expr)	\
+({	\
+	e_type _ret_;	\
+	if(len){ _ret_ = ary[0]; array_heap_remove(ary, len, cap, e_type, 0, cmp_expr); }	\
+	else memset(&_ret_, 0xFFU, sizeof(e_type));	\
+	_ret_;	\
+})
+
+/**
+ * List
+ */
+
+#define define_list_core(list_type, e_type, size_type, inc_size)	\
+	\
+typedef struct { e_type* buffer; union { size_type size; size_type length; }; size_type cap; size_type off; u4i mem_zero:1, n_head:31; } list_type;	\
+	\
+static inline size_t list_type##_obj_desc_cnt(void *list, int idx){	\
+	if(idx == 0) return ((list_type*)list)->size * sizeof(e_type);	\
+	else return 0;	\
+}	\
+	\
+static inline void list_type##_obj_desc_post_load(void *obj, size_t aux_data){	\
+	list_type *list;	\
+	UNUSED(aux_data);	\
+	list = (list_type*)obj;	\
+	list->cap = list->size;	\
+}	\
+	\
+static const obj_desc_t list_type##_obj_desc = {.tag = TOSTR(_list_##list_type), .size = sizeof(list_type), .n_child = 1, .mem_type = {1}, .addr = {offsetof(list_type, buffer)}, .desc = {(struct obj_desc_t*)&OBJ_DESC_DATA}, .cnt = list_type##_obj_desc_cnt, .post = list_type##_obj_desc_post_load};	\
+	\
+static inline void adv_##list_type##_init(list_type *list, size_type init_size, int mem_zero, u4i n_head){	\
+	if(init_size == 0) init_size = 2;	\
+	list->size = 0;	\
+	list->off  = 0;	\
+	list->cap  = init_size;	\
+	list->mem_zero = mem_zero;	\
+	list->n_head = n_head;	\
+	if(n_head){	\
+		list->buffer = (e_type*)calloc(list->cap + n_head, sizeof(e_type));	\
+		list->buffer += n_head;	\
+	} else {	\
+		list->buffer = (e_type*)calloc(list->cap, sizeof(e_type));	\
+	}	\
+}	\
+	\
+static inline list_type* adv_init_##list_type(size_type init_size, int mem_zero, u4i n_head){	\
+	list_type *list = (list_type*)malloc(sizeof(list_type));	\
+	adv_##list_type##_init(list, init_size, mem_zero, n_head);	\
+	return list;	\
+}	\
+	\
+static inline list_type* init_##list_type(size_type init_size){	\
+	return adv_init_##list_type(init_size, 0, 0);	\
+}	\
+	\
+static inline void list_type##_init(list_type *list, size_type init_size){	\
+	adv_##list_type##_init(list, init_size, 0, 0);	\
+}	\
+	\
+static inline int head_sl_##list_type(list_type *list, size_type len){	\
+	if(list->n_head < len) return 0;	\
+	list->buffer -= len;	\
+	list->n_head -= len;	\
+	list->cap    += len;	\
+	if(list->size) list->size += len;	\
+	return 1;	\
+}	\
+	\
+static inline int head_sr_##list_type(list_type *list, size_type len){	\
+	if(list->cap < len) return 0;	\
+	list->buffer += len;	\
+	list->n_head += len;	\
+	list->cap    -= len;	\
+	if(list->size > len) list->size -= len;	\
+	else list->size = 0;	\
+	return 1;	\
+}	\
+	\
+static inline void renew_##list_type(list_type *list, size_type init_size){	\
+	if(list->buffer) free(list->buffer - list->n_head);	\
+	adv_##list_type##_init(list, init_size, list->mem_zero, list->n_head);	\
+}	\
+	\
+static inline size_type count_##list_type(list_type *list){ return list->size; }	\
+	\
+static inline void clear_##list_type(list_type *list){ list->size = 0; list->off = 0; }	\
+	\
+static inline void zeros_##list_type(list_type *list){ memset(list->buffer, 0, list->cap * sizeof(e_type)); }	\
+	\
+static inline void encap_##list_type(list_type *list, size_type n){	\
+	list->cap = encap_list((void**)&list->buffer, sizeof(e_type), list->size, list->cap, n, list->mem_zero, list->n_head);	\
+}	\
+	\
+static inline void recap_##list_type(list_type *list, size_type n){	\
+	if((size_t)n == (size_t)list->cap) return;	\
+	list->cap = n;	\
+	if(list->size > n) list->size = n;	\
+	list->buffer = realloc(list->buffer - list->n_head, (list->cap + list->n_head) * sizeof(e_type)) + list->n_head;	\
+}	\
+	\
+static inline void pack_##list_type(list_type *list){	\
+	return recap_##list_type(list, list->size);	\
+}	\
+	\
+static inline void encap_and_zeros_##list_type(list_type *list, size_type n){	\
+	if(((size_t)list->size) + ((size_t)n) <= (size_t)list->cap){	\
+	} else {	\
+		if((size_t)(list->size + n) != ((size_t)list->size) + ((size_t)n)){	\
+			fprintf(stderr, " -- elements size exceed %s's data type %s in %s -- %s:%d --\n", #list_type, #size_type, __FUNCTION__, __FILE__, __LINE__);	\
+			print_backtrace(stderr, 20);	\
+			fflush(stderr);	\
+			exit(1);	\
+		}	\
+		while(list->size + n > list->cap){	\
+			if(list->cap < inc_size){	\
+				list->cap <<= 1;	\
+			} else {	\
+				list->cap += inc_size;	\
+			}	\
+		}	\
+		list->buffer = realloc(list->buffer, list->cap * sizeof(e_type));	\
+	}	\
+	memset(list->buffer + list->size, 0, n * sizeof(e_type));	\
+}	\
+	\
+static inline void clear_and_encap_##list_type(list_type *list, size_type n){	\
+	list->size = 0;	\
+	list->off  = 0;	\
+	encap_##list_type(list, n);	\
+}	\
+	\
+static inline void clear_and_inc_##list_type(list_type *list, size_type n){	\
+	list->size = 0;	\
+	list->off  = 0;	\
+	encap_##list_type(list, n);	\
+	list->size = n;	\
+}	\
+	\
+static inline void trunc_##list_type(list_type *list, size_type size){	\
+	if(size > count_##list_type(list)) size = count_##list_type(list);	\
+	list->size -= size;	\
+}	\
+	\
+static inline void set_##list_type##_size(list_type *list, size_type size){ list->size = size; }	\
+	\
+static inline void inc_##list_type(list_type *list, size_type size){	\
+	encap_##list_type(list, size);	\
+	list->size += size;	\
+}	\
+	\
+static inline void lazy_push_##list_type(list_type *list, e_type e){ list->buffer[list->size ++] = e; }	\
+	\
+static inline void push_##list_type(list_type *list, e_type e){	\
+	encap_##list_type(list, 1);	\
+	list->buffer[list->size++] = e;	\
+}	\
+	\
+static inline e_type* ring_ref_##list_type(list_type *list, size_type idx){	\
+	idx = (list->off + idx) % list->cap;	\
+	return list->buffer + idx;	\
+}	\
+	\
+static inline void ring_push_##list_type(list_type *list, e_type e){	\
+	size_type idx;	\
+	idx = (list->off + list->size) % list->cap;	\
+	list->buffer[idx] = e;	\
+	if(list->size < list->cap){	\
+		list->size ++;	\
+	}	\
+}	\
+	\
+static inline e_type* ring_peer_##list_type(list_type *list){	\
+	if(list->size){	\
+		return list->buffer + list->size - 1;	\
+	} else return NULL;	\
+}	\
+	\
+static inline e_type* ring_pop_##list_type(list_type *list){	\
+	size_type idx;	\
+	if(list->size){	\
+		list->size --;	\
+		idx = (list->off + list->size) % list->cap;	\
+		return list->buffer + idx;	\
+	} else return NULL;	\
+}	\
+	\
+static inline void ring_shift_##list_type(list_type *list, e_type e){	\
+	list->off = (list->off + list->cap - 1) % list->cap;	\
+	list->buffer[list->off] = e;	\
+	if(list->size < list->cap){	\
+		list->size ++;	\
+	}	\
+}	\
+	\
+static inline e_type* ring_unshift_##list_type(list_type *list){	\
+	size_type idx;	\
+	if(list->size){	\
+		list->size --;	\
+		list->off = (list->off + 1) % list->cap;	\
+		idx = (list->off + list->size) % list->cap;	\
+		return list->buffer + idx;	\
+	} else return NULL;	\
+}	\
+	\
+static inline int fpush_##list_type(list_type *list, FILE *inp){	\
+	int ret;	\
+	encap_##list_type(list, 1);	\
+	ret = fread(list->buffer + list->size, sizeof(e_type), 1, inp);	\
+	if(ret == 1) list->size ++;	\
+	return ret;	\
+}	\
+	\
+static inline e_type* peer_##list_type(list_type *list){	\
+	if(count_##list_type(list)){	\
+		return list->buffer + list->size - 1;	\
+	} else return NULL;	\
+}	\
+	\
+static inline e_type* head_##list_type(list_type *list){	\
+	if(list->size) return list->buffer;	\
+	else return NULL;	\
+}	\
+	\
+static inline e_type* tail_##list_type(list_type *list){	\
+	if(list->size) return list->buffer + list->size - 1;	\
+	else return NULL;	\
+}	\
+static inline int pop_##list_type(list_type *list, e_type*e){	\
+	if(count_##list_type(list)){	\
+		list->size --;	\
+		*e = list->buffer[list->size];	\
+		return 1;	\
+	} else return 0;	\
+}	\
+	\
+static inline int fpop_##list_type(list_type *list, FILE *oup){	\
+	if(list->size){	\
+		fwrite(list->buffer + list->size - 1, sizeof(e_type), 1, oup);	\
+		list->size --;	\
+		return 1;	\
+	} else {	\
+		return 0;	\
+	}	\
+}	\
+	\
+static inline void insert_##list_type(list_type *list, size_type idx, e_type e){	\
+	if(idx > list->size) return;	\
+	encap_##list_type(list, 1);	\
+	if(idx == list->size){	\
+		list->buffer[list->size] = e;	\
+	} else {	\
+		memmove(list->buffer + idx + 1, list->buffer + idx, (list->size - idx) * sizeof(e_type));	\
+		list->buffer[idx] = e;	\
+	}	\
+	list->size ++;	\
+}	\
+	\
+static inline void insert_array_##list_type(list_type *list, size_type idx, e_type *es, size_type size){	\
+	if(idx > list->size) return;	\
+	encap_##list_type(list, size);	\
+	if(idx == list->size){	\
+	} else {	\
+		memmove(list->buffer + idx + size, list->buffer + idx, (list->size - idx) * sizeof(e_type));	\
+	}	\
+	memcpy(list->buffer + idx, es, size * sizeof(e_type));	\
+	list->size += size;	\
+}	\
+	\
+static inline void remove_##list_type(list_type *list, size_type idx){	\
+	if(idx >= list->size) return;	\
+	if(idx + 1 < list->size){	\
+		memmove(list->buffer + idx, list->buffer + idx + 1, (list->size - idx - 1) * sizeof(e_type));	\
+	}	\
+	list->size --;	\
+}	\
+	\
+static inline void remove_array_##list_type(list_type *list, size_type off, size_type len){	\
+	if(off >= list->size) return;	\
+	if(off + len < list->size){	\
+		memmove(list->buffer + off, list->buffer + off + len, (list->size - off - len) * sizeof(e_type));	\
+		list->size -= len;	\
+	} else { \
+		list->size = off;	\
+	}	\
+}	\
+	\
+static inline void set_##list_type(list_type *list, size_type idx, e_type e){ list->buffer[idx] = e; }	\
+	\
+static inline e_type get_##list_type(list_type *list, size_type idx){ return list->buffer[idx]; }	\
+	\
+static inline e_type* ref_##list_type(list_type *list, size_type idx){ return list->buffer + idx; }	\
+	\
+static inline size_type offset_##list_type(list_type *list, e_type *e){ return e - list->buffer; }	\
+	\
+static inline e_type* next_ref_##list_type(list_type *list){ encap_##list_type(list, 1); list->size ++; return list->buffer + list->size - 1; }	\
+	\
+static inline e_type* ref_next_##list_type(list_type *list){ list->size ++; return list->buffer + list->size - 1; }	\
+	\
+static inline e_type* as_array_##list_type(list_type *list){ return list->buffer; }	\
+	\
+static inline void reverse_##list_type(list_type *list){	\
+	size_type i, j;	\
+	e_type t;	\
+	if(count_##list_type(list) == 0) return;	\
+	i = 0;	\
+	j = count_##list_type(list) - 1;	\
+	while(i < j){	\
+		t = get_##list_type(list, i);	\
+		set_##list_type(list, i, get_##list_type(list, j));	\
+		set_##list_type(list, j, t);	\
+		i ++;	\
+		j --;	\
+	}	\
+}	\
+	\
+static inline void sub_reverse_##list_type(list_type *list, size_type beg, size_type end){	\
+	size_type i, j;	\
+	e_type t;	\
+	if(end <= beg) return;	\
+	i = beg;	\
+	j = end - 1;	\
+	while(i < j){	\
+		t = get_##list_type(list, i);	\
+		set_##list_type(list, i, get_##list_type(list, j));	\
+		set_##list_type(list, j, t);	\
+		i ++;	\
+		j --;	\
+	}	\
+}	\
+	\
+static inline void append_##list_type(list_type *list1, list_type *list2){	\
+	encap_##list_type(list1, count_##list_type(list2));	\
+	memcpy(list1->buffer + list1->size, list2->buffer, sizeof(e_type) * list2->size);	\
+	list1->size += list2->size;	\
+}	\
+	\
+static inline void append_array_##list_type(list_type *list1, e_type *ary, size_type size){	\
+	if(size == 0) return;	\
+	encap_##list_type(list1, size);	\
+	memcpy(list1->buffer + list1->size, ary, sizeof(e_type) * size);	\
+	list1->size += size;	\
+}	\
+	\
+static inline size_type write_##list_type(list_type *list, FILE *out){	\
+	return fwrite(list->buffer, sizeof(e_type), count_##list_type(list), out);	\
+}	\
+	\
+static inline size_type dump_##list_type(list_type *list, FILE *out){	\
+	fwrite(&list->size, sizeof(size_type), 1, out);	\
+	fwrite(list->buffer, sizeof(e_type), list->size, out);	\
+	return sizeof(size_type) + sizeof(e_type) * list->size;	\
+}	\
+	\
+static inline size_t mem_size_##list_type(list_type *list){	\
+	return ((sizeof(list_type) + 7) / 8) * 8 + (list->size * sizeof(e_type) + 7) / 8 * 8;	\
+}	\
+	\
+static inline size_t mem_dump_##list_type(list_type *list, FILE *out, void *addr){	\
+	list_type clone;	\
+	size_t size;	\
+	b1i i, v;	\
+	size = mem_size_##list_type(list);	\
+	clone.size = list->size;	\
+	clone.cap  = list->size;	\
+	clone.buffer = addr + (sizeof(list_type) + 7) / 8 * 8;	\
+	fwrite(&clone, sizeof(list_type), 1, out);	\
+	v = 0;	\
+	for(i=(sizeof(list_type) + 7) / 8 * 8 - sizeof(list_type);i>0;i--) fwrite(&v, 1, 1, out);	\
+	fwrite(list->buffer, sizeof(e_type), list->size, out);	\
+	for(i=(sizeof(e_type) * list->size + 7) / 8 * 8 - sizeof(e_type) * list->size;i>0;i--) fwrite(&v, 1, 1, out);	\
+	return size;	\
+}	\
+	\
+static inline list_type* load_##list_type(FILE *inp){	\
+	list_type *list;	\
+	size_type n;	\
+	list = (list_type*)malloc(sizeof(list_type));	\
+	if((n = fread(&list->size, sizeof(size_type), 1, inp)) != 1){	\
+		free(list); return NULL;	\
+	}	\
+	list->cap = list->size;	\
+	list->buffer = (e_type*)malloc(sizeof(e_type) * list->cap);	\
+	if(list->buffer == NULL){	\
+		fprintf(stderr, " Out of memory in load_%s \n", #list_type); fflush(stderr); exit(1);	\
+		print_backtrace(stderr, 20);	\
+	}	\
+	if((n = fread(list->buffer, sizeof(e_type), list->size, inp)) != list->size){	\
+		print_backtrace(stderr, 20);	\
+		free(list->buffer); free(list); return NULL;	\
+	}	\
+	return list;	\
+}	\
+	\
+static inline void free_##list_type(list_type *list){ if(list){ free(list->buffer - list->n_head); free(list); } }	\
+	\
+static inline void list_type##_free(list_type *list){ free(list->buffer); list->buffer = NULL; }	\
+
+#define define_list_ext(list_type, e_type, size_type, cmp_func)	\
+static inline size_type delete_##list_type(list_type *list, e_type e){	\
+	size_type i, ret;	\
+	ret = 0;	\
+	for(i=list->size;i>0;i--){	\
+		if(cmp_func(list->buffer[i-1], e, NULL) == 0){	\
+			if(i < list->size){	\
+				memmove(list->buffer + i - 1, list->buffer + i, (list->size - i) * sizeof(e_type));	\
+			}	\
+			list->size --;	\
+			ret ++;	\
+		}	\
+	}	\
+	return ret;	\
+}	\
+	\
+static inline size_type occ_##list_type(list_type *list, e_type e){	\
+	size_type i, n;	\
+	for(i=0,n=0;i<list->size;i++){	\
+		if(cmp_func(list->buffer[i], e, NULL) == 0) n++;	\
+	}	\
+	return n;	\
+}	\
+	\
+static inline size_type replace_##list_type(list_type *list, e_type from, e_type to){	\
+	size_type i, ret;	\
+	ret = 0;	\
+	for(i=0;i<list->size;i++){	\
+		if(cmp_func(list->buffer[i], from, NULL) == 0){	\
+			list->buffer[i] = to;	\
+			ret ++;	\
+		}	\
+	}	\
+	return ret;	\
+}	\
+	\
+static inline size_type locate_##list_type(list_type *list, e_type e, size_type start){	\
+	size_type i;	\
+	for(i=start;i<list->size;i++){	\
+		if(cmp_func(list->buffer[i], e, NULL) == 0) return i;	\
+	}	\
+	return i;	\
+}	\
+	\
+define_quick_sort(sort_##list_type##_core, e_type, cmp_func);	\
+	\
+static inline void sort_##list_type(list_type *list){ sort_##list_type##_core(ref_##list_type(list, 0), count_##list_type(list), NULL); }
+
+#define define_list(name, e_type) define_list_core(name, e_type, size_t, 0xFFFFFU)
+
+#define native_number_cmp(e1, e2, obj) (((e1) == (e2))? 0 : (((e1) < (e2))? -1 : 1))
+
+#define define_native_list(name, e_type)	\
+define_list_core(name, e_type, size_t, 0xFFFFFU);	\
+define_list_ext(name, e_type, size_t, native_number_cmp);
+
+define_native_list(u8list,  u1i);
+define_native_list(u1v, u1i);
+define_native_list(u16list, u2i);
+define_native_list(u2v, u2i);
+define_native_list(u32list, u4i);
+define_native_list(u4v, u4i);
+define_native_list(u64list, u8i);
+define_native_list(u8v, u8i);
+
+define_native_list(b8list,  b1i);
+define_native_list(b1v, b1i);
+define_native_list(b16list, b2i);
+define_native_list(b2v, b2i);
+define_native_list(b32list, b4i);
+define_native_list(b4v, b4i);
+define_native_list(b64list, b8i);
+define_native_list(b8v, b8i);
+
+define_native_list(f4v, f4i);
+define_native_list(f8v, f8i);
+
+define_list(vplist, void*);
+define_list(cplist, char*);
+// mem_share for deep dump of cplist
+static inline size_t cplist_deep_obj_desc_cnt(void *list, int idx){
+	if(idx == 0) return ((cplist*)list)->size;
+	else return 1;
+}
+static const obj_desc_t cplist_deep_obj_desc = {.tag = "cplist_deep_dump", .size = sizeof(cplist), .n_child = 1, .mem_type = {3}, .addr = {offsetof(cplist, buffer)}, .desc = {(struct obj_desc_t*)&OBJ_DESC_CHAR_ARRAY}, .cnt = cplist_deep_obj_desc_cnt, .post=NULL};
+
+#define define_recycle_list_array(name, list_type)	\
+typedef struct {	\
+	vplist *array;	\
+	vplist *dustbin;	\
+} name;	\
+	\
+static inline name* init_##name(){	\
+	name *g;	\
+	g = (name*)malloc(sizeof(name));	\
+	g->buffer = init_vplist(4);	\
+	g->dustbin = init_vplist(4);	\
+	return g;	\
+}	\
+	\
+static inline void free_##name(name *g){	\
+	list_type *v;	\
+	size_t i;	\
+	for(i=0;i<g->array->size;i++){	\
+		v = (list_type*)get_vplist(g->array, i);	\
+		if(v) free_##list_type(v);	\
+	}	\
+	for(i=0;i<g->dustbin->size;i++){	\
+		v = (list_type*)get_vplist(g->dustbin, i);	\
+		if(v) free_##list_type(v);	\
+	}	\
+	free_vplist(g->array);	\
+	free_vplist(g->dustbin);	\
+	free(g);	\
+}	\
+	\
+static inline list_type* fetch_##name(name *g){	\
+	list_type *v;	\
+	if(g->dustbin->size) v = (list_type*)g->dustbin->buffer[--g->dustbin->size];	\
+	else v = init_##list_type(4);	\
+	return v;	\
+}	\
+	\
+static inline void recyc_##name(name *g, list_type *v){	\
+	push_vplist(g->dustbin, v);	\
+}	\
+	\
+static inline void recyc_all_##name(name *g, vplist *vs){	\
+	append_vplist(g->dustbin, vs);	\
+	vs->size = 0;	\
+}
+
+// e.g. define_recycle_list(u8r, u8i, u8i, *a = 0, NULL)
+//       u8r, when increase an element, set the to 0 (*a = 0), when free it, do nothing (NULL)
+#define define_recycle_list(list_type, e_type, size_type, e_init, e_free)	\
+typedef struct {	\
+	e_type *buffer;	\
+	size_type size, cap;	\
+	size_type *recyc;	\
+	size_type rsize, rcap;	\
+	u8i userdata;	\
+} list_type;	\
+	\
+static inline list_type* init_##list_type(size_type size){	\
+	list_type *list;	\
+	if(size < 2) size = 2;	\
+	list = malloc(sizeof(list_type));	\
+	list->size = 0;	\
+	list->cap  = size;	\
+	list->buffer = calloc(size, sizeof(e_type));	\
+	list->recyc = calloc(2, sizeof(size_type));	\
+	list->rsize = 0;	\
+	list->rcap  = 2;	\
+	list->userdata = 0;	\
+	return list;	\
+}	\
+	\
+static inline void free_##list_type(list_type *list){	\
+	e_type* a;	\
+	size_type i;	\
+	for(i=0;i<list->size;i++){	\
+		a = list->buffer + i;	\
+		UNUSED(e_free);	\
+	}	\
+	UNUSED(a);	\
+	free(list->buffer);	\
+	free(list->recyc);	\
+	free(list);	\
+}	\
+	\
+static inline void encap_##list_type(list_type *list, size_type inc){	\
+	if(list->rsize >= inc) return;	\
+	inc -= list->rsize;	\
+	list->cap = encap_list((void**)&list->buffer, sizeof(e_type), list->size, list->cap, inc, 0, 0);	\
+}	\
+	\
+static inline size_type fetch_##list_type(list_type *list){	\
+	e_type* a;	\
+	if(list->rsize){	\
+		return list->recyc[--list->rsize];	\
+	}	\
+	list->cap = encap_list((void**)&list->buffer, sizeof(e_type), list->size, list->cap, 1, 0, 0);	\
+	a = list->buffer + list->size;	\
+	UNUSED(e_init);	\
+	UNUSED(a);	\
+	return list->size ++;	\
+}	\
+	\
+static inline e_type* ref_##list_type(list_type *list, size_type idx){	\
+	return list->buffer + idx;	\
+}	\
+	\
+static inline size_type offset_##list_type(list_type *list, e_type *e){	\
+	return e - list->buffer;	\
+}	\
+	\
+static inline void recyc_##list_type(list_type *list, size_type idx){	\
+	list->rcap = encap_list((void**)&list->recyc, sizeof(size_type), list->rsize, list->rcap, 1, 0, 0);	\
+	list->recyc[list->rsize++] = idx;	\
+}	\
+	\
+static inline e_type* pop_##list_type(list_type *list){	\
+	size_type idx;	\
+	idx = fetch_##list_type(list);	\
+	return ref_##list_type(list, idx);	\
+}	\
+	\
+static inline void push_##list_type(list_type *list, e_type* e){	\
+	recyc_##list_type(list, offset_##list_type(list, e));	\
+}	\
+	\
+static inline void recyc_all_##list_type(list_type *list){	\
+	size_type i;	\
+	list->rsize = 0;	\
+	list->rcap = encap_list((void**)&list->recyc, sizeof(size_type), list->rsize, list->rcap, list->size, 0, 0);	\
+	for(i=0;i<list->size;i++){	\
+		list->recyc[i] = i;	\
+	}	\
+	list->rsize = list->size;	\
+}
+
+# endif

pgzf.c

@@ -0,0 +1,253 @@
+/*
+ *
+ * Copyright (c) 2018, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "pgzf.h"
+
+int usage(int ret){
+	fprintf(stdout,
+	"PGZF: Parallel gzip file IO\n"
+	"Author: Jue Ruan <ruanjue@gmail.com>\n"
+	"Version: 1.1\n"
+	"Usage: pgzf [options] file1 [file2 ...]\n"
+	"Options:\n"
+	" -d          Decompress mode\n"
+	" -t <int>    Number of threads, [8]\n"
+	" -f          Force to overwrite\n"
+	" -o <string> Output file name, support directory\n"
+	" -x          Delete input files after done\n"
+	" -b <int>    Block size in MB, 1 ~ 256 [16]\n"
+	" -l <int>    Compress level, 1-9, see gzip, [6]\n"
+	" -h          Show this document\n"
+	" -V          Print version information and exit\n"
+	"\n"
+	"File format:\n"
+	" PGZF fellows standard GZIP format (rfc1952), and is blocked compressed.\n"
+	" It defines two TAGs in each GZIP header, ZS: block size, ZX: random access index.\n"
+	" Program pgzf can decompress .pgzf and .gz files. When decompressing .gz files,\n"
+	" pgzf is in fact a buffered gzip reader. Also, .pgzf files can be decompressed\n"
+	" by program gzip.\n"
+	"\n"
+	"In plan to support random access\n"
+	);
+	return ret;
+}
+
+int main(int argc, char **argv){
+	PGZF *pz;
+	char *outf, *ftag;
+	FILE *in, *out;
+	void *buff;
+	u4i bufsize, nbyte;
+	int c, rw, ncpu, level, overwrite, del, is_dir;
+	rw = PGZF_MODE_W;
+	ncpu = 8;
+	bufsize = PGZF_DEFAULT_BUFF_SIZE;
+	level = 6;
+	overwrite = 0;
+	del = 0;
+	outf = NULL;
+	while((c = getopt(argc, argv, "hdxft:b:l:o:V")) != -1){
+		switch(c){
+			case 'h': return usage(0);
+			case 'd': rw = PGZF_MODE_R; break;
+			case 't': ncpu = atoi(optarg); break;
+			case 'b': bufsize = (atol(optarg) << 20); break;
+			case 'l': level = atoi(optarg); break;
+			case 'f': overwrite = 1; break;
+			case 'o': outf = optarg; break;
+			case 'x': del = 1; break;
+			case 'V': fprintf(stdout, "pgzf 1.1\n"); return 0;
+			default: return usage(1);
+		}
+	}
+	if(optind == argc){
+		return usage(1);
+	}
+	if(0 && del){
+		if(outf == NULL && overwrite == 0){
+			if(optind < argc){
+				fprintf(stderr, " ** WARNNING: won't delete input files. To force delete input files, please specify -o or/and -f\n");
+			}
+			del = 0;
+		}
+	}
+	is_dir = 0;
+	out = NULL;
+	if(outf){
+		if(file_exists(outf)){
+			if(overwrite == 0){
+				fprintf(stderr, " ** ERROR: '%s' exists\n", outf);
+				return 1;
+			} else {
+				for(c=optind;c<argc;c++){
+					if(strcmp(outf, argv[c]) == 0){
+						fprintf(stderr, " ** ERROR: The same file in INPUT and OUTPUT, '%s'\n", outf);
+						return 1;
+					}
+				}
+			}
+			out = open_file_for_write(outf, NULL, overwrite);
+		} else if(dir_exists(outf)){
+			is_dir = 1;
+		} else {
+			out = open_file_for_write(outf, NULL, overwrite);
+		}
+	}
+	buff = malloc(bufsize);
+	if(rw == PGZF_MODE_R){
+		if(outf == NULL || is_dir){
+			for(c=optind;c<argc;c++){
+				if(strlen(argv[c]) < 4 || strcasecmp(argv[c] + strlen(argv[c]) - 3, ".gz")){
+					fprintf(stderr, " ** ERROR: cannot auto generate output file name for '%s'\n", argv[c]);
+					return 1;
+				} else if(is_dir){
+					char *rtag;
+					rtag = relative_filename(argv[c]);
+					rtag[strlen(rtag) - 3] = 0;
+					ftag = malloc(strlen(outf) + 1 + strlen(rtag) + 1);
+					sprintf(ftag, "%s/%s", outf, rtag);
+					free(rtag);
+					if(overwrite == 0 && file_exists(ftag)){
+						fprintf(stderr, " ** ERROR: '%s' exists\n", ftag);
+						return 1;
+					}
+					free(ftag);
+				} else {
+					ftag = strdup(argv[optind]);
+					ftag[strlen(ftag) - 3] = 0;
+					if(overwrite == 0 && file_exists(ftag)){
+						fprintf(stderr, " ** ERROR: '%s' exists\n", ftag);
+						return 1;
+					}
+					free(ftag);
+				}
+			}
+		}
+		do {
+			in = open_file_for_read(argv[optind], NULL);
+			if(outf == NULL){
+				ftag = strdup(argv[optind]);
+				ftag[strlen(ftag) - 3] = 0;
+				out = open_file_for_write(ftag, NULL, overwrite);
+				free(ftag);
+			} else if(is_dir){
+				char *rtag;
+				rtag = relative_filename(argv[optind]);
+				rtag[strlen(rtag) - 3] = 0;
+				ftag = malloc(strlen(outf) + 1 + strlen(rtag) + 1);
+				sprintf(ftag, "%s/%s", outf, rtag);
+				free(rtag);
+				out = open_file_for_write(ftag, NULL, overwrite);
+				free(ftag);
+			}
+			pz = open_pgzf_reader(in, bufsize, ncpu);
+			while((nbyte = read_pgzf(pz, buff, bufsize))){
+				fwrite(buff, 1, nbyte, out);
+			}
+			if(pz->error){
+				fprintf(stderr, " ** ERROR: error code (%d)'\n", pz->error);
+				return 1;
+			}
+			close_pgzf(pz);
+			if(in != stdin){
+				fclose(in);
+				if(del){
+					unlink(argv[optind]);
+				}
+			}
+			optind ++;
+			if(outf == NULL || is_dir){
+				fclose(out);
+			}
+		} while(optind < argc);
+	} else {
+		if(outf && !is_dir){
+			pz = open_pgzf_writer(out, bufsize, ncpu, level);
+		} else {
+			pz = NULL;
+			for(c=optind;c<argc;c++){
+				if(strlen(argv[c]) >= 4 && strcasecmp(argv[c] + strlen(argv[c]) - 3, ".gz") == 0){
+					fprintf(stderr, " ** ERROR: file seems already compressed '%s'\n", argv[c]);
+					return 1;
+				} else if(strcmp(argv[c], "-") == 0){
+					fprintf(stderr, " ** ERROR: Please specify output file when read from STDIN '%s'\n", argv[c]);
+					return 1;
+				} else if(is_dir){
+					char *rtag;
+					rtag = relative_filename(argv[c]);
+					ftag = malloc(strlen(outf) + 1 + strlen(rtag) + 3 + 1);
+					sprintf(ftag, "%s/%s.gz", outf, rtag);
+					free(rtag);
+					if(overwrite == 0 && file_exists(ftag)){
+						fprintf(stderr, " ** ERROR: '%s' exists\n", ftag);
+						return 1;
+					}
+					free(ftag);
+				} else {
+					ftag = malloc(strlen(argv[c]) + 4);
+					sprintf(ftag, "%s.gz", argv[c]);
+					if(overwrite == 0 && file_exists(ftag)){
+						fprintf(stderr, " ** ERROR: '%s' exists\n", ftag);
+						return 1;
+					}
+					free(ftag);
+				}
+			}
+		}
+		do {
+			if(outf == NULL){
+				ftag = malloc(strlen(argv[optind]) + 4);
+				sprintf(ftag, "%s.gz", argv[optind]);
+				out = open_file_for_write(ftag, NULL, overwrite);
+				pz = open_pgzf_writer(out, bufsize, ncpu, level);
+				free(ftag);
+			} else if(is_dir){
+				char *rtag;
+				rtag = relative_filename(argv[optind]);
+				ftag = malloc(strlen(outf) + 1 + strlen(rtag) + 3 + 1);
+				sprintf(ftag, "%s/%s.gz", outf, rtag);
+				free(rtag);
+				out = open_file_for_write(ftag, NULL, overwrite);
+				pz = open_pgzf_writer(out, bufsize, ncpu, level);
+				free(ftag);
+			}
+			in = open_file_for_read(argv[optind], NULL);
+			while((nbyte = fread(buff, 1, bufsize, in))){
+				write_pgzf(pz, buff, nbyte);
+			}
+			if(in != stdin){
+				fclose(in);
+				if(del){
+					unlink(argv[optind]);
+				}
+			}
+			if(outf == NULL || is_dir){
+				close_pgzf(pz);
+				fclose(out);
+			}
+			optind ++;
+		} while(optind < argc);
+		if(outf && !is_dir){
+			close_pgzf(pz);
+		}
+	}
+	if(outf && !is_dir) fclose(out);
+	free(buff);
+	return 0;
+}

pgzf.h

@@ -0,0 +1,852 @@
+/*
+ *
+ * Copyright (c) 2018, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __PGZF_RJ_H
+#define __PGZF_RJ_H
+
+#include <zlib.h>
+#include "mem_share.h"
+#include "list.h"
+#include "thread.h"
+
+#define PGZF_DEFAULT_BUFF_SIZE	(1U << 24) // 16MB
+#define PGZF_MAX_BUFF_SIZE	(1U << 28)
+#define PGZF_HEAD_SIZE	30
+#define PGZF_HEAD_ZS_OFFSET	16
+#define PGZF_HEAD_ZX_OFFSET	24
+#define PGZF_TAIL_SIZE	8
+#define PGZF_INDEX_BIN	64
+
+#define PGZF_MODE_W	1 // pgzf writer
+#define PGZF_MODE_R	2 // pgzf reader
+#define PGZF_MODE_R_GZ	3 // gz reader
+#define PGZF_MODE_R_UNKNOWN	4 // unknown file type
+
+#define PGZF_FILETYPE_UNKNOWN	0
+#define PGZF_FILETYPE_GZ	1
+#define PGZF_FILETYPE_PGZF	2
+
+struct PGZF;
+
+#define PGZF_TASK_NULL	0
+#define PGZF_TASK_DEFLATE	1
+#define PGZF_TASK_INFLATE	2
+
+thread_beg_def(pgz);
+struct PGZF *pz;
+u4i zsval, soff;
+u8i zxval, doff;
+u1v *dst, *src;
+u4i token;
+int level;
+int task;
+thread_end_def(pgz);
+
+typedef struct PGZF {
+	u4i ncpu, ridx, widx;
+	int rw_mode, seekable;
+	u4i bufsize; // MUST be multiple of 1MB
+	u8i xsize; // total uncompressed size
+	u8v *boffs;
+	u8v *xoffs;
+	u8i tot_in, tot_out;
+	u1v **dsts, **srcs, *tmp;
+	z_stream *z;
+	u8i offset;
+	FILE *file;
+	thread_def_shared_vars(pgz);
+	int step; // used in decompress gzip file
+	int eof, error;
+} PGZF;
+
+static inline void _num2bytes_pgzf(u1i *bs, u1i bl, u8i val){
+	u1i i;
+	for(i=0;i<bl;i++){
+		bs[i] = (u1i)val;
+		val >>= 8;
+	}
+}
+
+static inline u8i _bytes2num_pgzf(u1i *bs, u1i bl){
+	u8i val;
+	u1i i;
+	val = 0;
+	for(i=0;i<bl;i++){
+		val = val | (((u8i)bs[i]) << (8 * i));
+	}
+	return val;
+}
+
+/**
+ * Please see https://tools.ietf.org/html/rfc1952
+ */
+
+static inline void _gen_pgzf_header(u1i bs[30], u4i z_size){
+	bs[0] = 0x1f; // GZIP ID1
+	bs[1] = 0x8b; // GZIP ID2
+	bs[2] = 8; // CM = deflate
+	bs[3] = 0x4; // FLG = 0b00000100, FEXTRA is ture
+	bs[4] = 0; // MTIME
+	bs[5] = 0; // MTIME
+	bs[6] = 0; // MTIME
+	bs[7] = 0; // MTIME
+	bs[8] = 0; // XFL
+	bs[9] = 3; // OS = unix
+	bs[10] = 18; // XLEN
+	bs[11] = 0; // = 18
+	bs[12] = 'Z'; // TAG ZS
+	bs[13] = 'S'; // compressed size
+	bs[14] = 4; // TAG LEN
+	bs[15] = 0; // 4
+	bs[16] = z_size >>  0; // compressed block size
+	bs[17] = z_size >>  8; //
+	bs[18] = z_size >> 16; //
+	bs[19] = z_size >> 24; // = z_size
+	bs[20] = 'Z'; //TAG ZX
+	bs[21] = 'X'; // every 64 block size
+	bs[22] = 6; // TAG LEN
+	bs[23] = 0; // 6
+	bs[24] = 0; // reserve to store the random access index
+	bs[25] = 0; //
+	bs[26] = 0; //
+	bs[27] = 0; //
+	bs[28] = 0; //
+	bs[29] = 0; //
+}
+
+static inline void _gen_pgzf_tailer(u1i bs[8], u4i crc, u4i u_size){
+	_num2bytes_pgzf(bs + 0, 4, crc);
+	_num2bytes_pgzf(bs + 4, 4, u_size);
+}
+
+static inline u4i _zlib_raw_deflate_all(u1i *dst, u4i dlen, u1i *src, u4i slen, int level){
+	z_stream Z, *z;
+	u4i ret;
+	z = &Z;
+	ZEROS(z);
+	deflateInit2(z, level, Z_DEFLATED, -15, 9, Z_DEFAULT_STRATEGY);
+	z->avail_in = slen;
+	z->next_in  = src;
+	z->avail_out = dlen;
+	z->next_out  = dst;
+	deflate(z, Z_FINISH);
+	ret = dlen - z->avail_out;
+	deflateEnd(z);
+	return ret;
+}
+
+static inline u4i _pgzf_deflate(u1v *dst, u1v *src, int level){
+	u4i z_size;
+	uLong crc;
+	clear_u1v(dst);
+	if(src->size == 0 || src->size >= MAX_U4) return 0;
+	z_size = compressBound(src->size);
+	encap_u1v(dst, z_size + PGZF_HEAD_SIZE + PGZF_TAIL_SIZE);
+	z_size = _zlib_raw_deflate_all(dst->buffer + PGZF_HEAD_SIZE, z_size, src->buffer, src->size, level);
+	_gen_pgzf_header(dst->buffer, z_size + PGZF_HEAD_SIZE + PGZF_TAIL_SIZE);
+	crc = crc32(0L, Z_NULL, 0);
+	crc = crc32(crc, src->buffer, src->size);
+	_gen_pgzf_tailer(dst->buffer + PGZF_HEAD_SIZE + z_size, crc, src->size);
+	dst->size = PGZF_HEAD_SIZE + z_size + PGZF_TAIL_SIZE;
+	return dst->size;
+}
+
+static inline int _read_pgzf_header(FILE *in, u1v *src, u4i *hoff, u4i *zsval, u8i *zxval){
+	u4i off, val, sl, end;
+	int ch, is_pgzf, xflag;
+	char si1, si2;
+	is_pgzf = 0;
+	off = *hoff;
+	*zsval = 0;
+	*zxval  = 0;
+	if(src->size < off + 10){
+		encap_u1v(src, 10);
+		src->size += fread(src->buffer + src->size, 1, off + 10 - src->size, in);
+	}
+	// At least give 10 bytes
+	if(src->size < off + 10) return PGZF_FILETYPE_UNKNOWN;
+	if(src->buffer[off + 0] != 0x1f) return PGZF_FILETYPE_UNKNOWN;
+	if(src->buffer[off + 1] != 0x8b) return PGZF_FILETYPE_UNKNOWN;
+	if((src->buffer[off + 2] != 0x08) || (src->buffer[off + 2] & 0xE0)) return PGZF_FILETYPE_UNKNOWN;
+	xflag = src->buffer[off + 3];
+	off += 10;
+	if(xflag & 0x04){
+		if(src->size < off + 2){
+			encap_u1v(src, 2);
+			sl = fread(src->buffer + src->size, 1, off + 2 - src->size, in);
+			src->size += sl;
+		}
+		if(src->size < off + 2) return PGZF_FILETYPE_UNKNOWN;
+		val = _bytes2num_pgzf(src->buffer + off, 2);
+		off += 2;
+		end = off + val;
+		if(val > 0 && val < 4) return PGZF_FILETYPE_UNKNOWN;
+		if(src->size < off + val){
+			encap_u1v(src, val);
+			sl = fread(src->buffer + src->size, 1, off + val - src->size, in);
+			src->size += sl;
+			if(src->size < off + val) return PGZF_FILETYPE_UNKNOWN;
+		}
+		//parse TAGs
+		while(off < end){
+			si1 = src->buffer[off + 0];
+			si2 = src->buffer[off + 1];
+			sl = _bytes2num_pgzf(src->buffer + off + 2, 2);
+			off += 4;
+			if(si1 == 'Z' && si2 == 'S' && sl == 4){
+				is_pgzf = 1;
+				*zsval = _bytes2num_pgzf(src->buffer + off, 4);
+			} else if(is_pgzf && si1 == 'Z' && si2 == 'X' && sl == 6){
+				*zxval = _bytes2num_pgzf(src->buffer + off, 6);
+			}
+			off += sl;
+		}
+	}
+	if(xflag & 0x08){
+		do {
+			if(off < src->size){
+				ch = src->buffer[off];
+			} else {
+				ch = getc(in);
+				if(ch == -1){
+					return PGZF_FILETYPE_UNKNOWN;
+				}
+				push_u1v(src, ch);
+			}
+			off ++;
+		} while(ch);
+	}
+	if(xflag & 0x10){
+		do {
+			if(off < src->size){
+				ch = src->buffer[off];
+			} else {
+				ch = getc(in);
+				if(ch == -1){
+					return PGZF_FILETYPE_UNKNOWN;
+				}
+				push_u1v(src, ch);
+			}
+			off ++;
+		} while(ch);
+	}
+	if(xflag & 0x02){
+		if(src->size < off + 2){
+			encap_u1v(src, 2);
+			sl = fread(src->buffer + src->size, 1, off + 2 - src->size, in);
+			src->size += sl;
+		}
+		off += 2;
+		if(src->size < off) return PGZF_FILETYPE_UNKNOWN;
+	}
+	*hoff = off;
+	return is_pgzf? PGZF_FILETYPE_PGZF : PGZF_FILETYPE_GZ;
+}
+
+int pgzf_inflate_raw_core(z_stream *z, u1i *dst, u4i *dlen, u1i *src, u4i *slen, int flush){
+	u4i dl, sl;
+	int ret;
+	ret = Z_OK;
+	dl = *dlen;
+	sl = *slen;
+	z->avail_in = sl;
+	z->next_in  = src;
+	z->avail_out = dl;
+	z->next_out  = dst;
+	ret = inflate(z, flush);
+	*dlen = dl - z->avail_out;
+	*slen = sl - z->avail_in;
+	return ret;
+}
+
+// src start just after gz_header, and include fz_tailer
+int pgzf_inflate_core(u1i *dst, u4i *dlen, u1i *src, u4i slen, int check){
+	z_stream Z, *z;
+	u4i soff, dsz;
+	uLong crc, rcr;
+	int ret;
+	z = &Z;
+	ZEROS(z);
+	inflateInit2(z, -15);
+	z->avail_in  = slen - PGZF_TAIL_SIZE;
+	z->next_in   = src;
+	z->avail_out = *dlen;
+	z->next_out  = dst;
+	ret = inflate(z, Z_FINISH);
+	*dlen -= z->avail_out;
+	soff = slen - PGZF_TAIL_SIZE - z->avail_in;
+	inflateEnd(z);
+	if(check){
+		if(soff + 8 > slen){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			return 0;
+		}
+		rcr = _bytes2num_pgzf(src + soff, 4);
+		dsz = _bytes2num_pgzf(src + soff + 4, 4);
+		if(dsz != *dlen){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			return 0;
+		}
+		crc = crc32(0L, Z_NULL, 0);
+		crc = crc32(crc, dst, *dlen);
+		if(crc != rcr){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			return 0;
+		}
+	}
+	return 1;
+}
+
+thread_beg_func(pgz);
+PGZF *pz;
+u1v *dst, *src;
+u4i bufsize, hsize, rsize, dsz, ssz, next;
+int ret;
+pz  = pgz->pz;
+dst = pgz->dst;
+src = pgz->src;
+thread_beg_loop(pgz);
+if(pgz->task == PGZF_TASK_DEFLATE){
+	if(src->size == 0) continue;
+	clear_u1v(dst);
+	_pgzf_deflate(dst, src, pgz->level);
+	while(pz->ridx != pgz->token){
+		nano_sleep(1);
+	}
+	{
+		pz->tot_out += pgz->dst->size;
+		push_u8v(pz->boffs, pz->tot_out);
+		fwrite(pgz->dst->buffer, 1, pgz->dst->size, pz->file);
+		clear_u1v(pgz->dst);
+		clear_u1v(pgz->src);
+	}
+	pz->ridx ++;
+} else if(pgz->task == PGZF_TASK_INFLATE){
+	pgz->doff = 0;
+	clear_u1v(pgz->dst);
+	while((pz->ridx % pz->ncpu) != UInt(pgz->t_idx)){
+		nano_sleep(10);
+		if(pz->error) break;
+		//if(pz->eof){
+			//if(pz->rw_mode != PGZF_MODE_R_GZ){
+				//break;
+			//}
+		//}
+		if(pgz->running == 0){
+			break;
+		}
+	}
+	if(pz->error) break;
+	if(pz->rw_mode == PGZF_MODE_R){
+		if(pgz->src->size){ // loaded header, had set zsval and zxval
+		} else {
+			pgz->zsval = pgz->zxval = 0;
+			pgz->soff = pgz->src->size = 0;
+			ret = _read_pgzf_header(pz->file, pgz->src, &pgz->soff, &pgz->zsval, &pgz->zxval);
+			if(pgz->src->size == 0){
+				pz->eof = 1;
+				pz->ridx ++;
+				break;
+			}
+			if(ret != PGZF_FILETYPE_PGZF){
+				fprintf(stderr, " -- Error: not a PGZF format at %u block, in %s -- %s:%d --\n", pz->ridx, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+				pz->error = 1;
+				pz->ridx ++;
+				break;
+			}
+		}
+		hsize = pgz->soff;
+		encap_u1v(pgz->src, pgz->zsval - pgz->src->size);
+		rsize = fread(pgz->src->buffer + hsize, 1, pgz->zsval - pgz->src->size, pz->file);
+		if(rsize < pgz->zsval - pgz->src->size){
+			fprintf(stderr, " -- Error: read %u < %u at %u block, in %s -- %s:%d --\n", UInt(pgz->src->size + rsize), pgz->zsval, pz->ridx, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			pz->error = 1;
+			pz->ridx ++;
+			break;
+		}
+		pgz->src->size += rsize;
+		pz->tot_in += pgz->zsval;
+		pz->ridx ++;
+		dsz = _bytes2num_pgzf(pgz->src->buffer + pgz->zsval - 4, 4);
+		encap_u1v(pgz->dst, dsz);
+		pgz->soff = 0;
+		if(pgzf_inflate_core(pgz->dst->buffer, &dsz, pgz->src->buffer + hsize, pgz->zsval - hsize, 1) == 0){
+			clear_u1v(pgz->src);
+			pz->error = 1;
+			break;
+		}
+		pgz->dst->size = dsz;
+		clear_u1v(pgz->src);
+	} else if(pz->rw_mode == PGZF_MODE_R_GZ){
+		u4i bsz;
+		bsz = 1024 * 1024;
+		bufsize = pz->bufsize? pz->bufsize : PGZF_DEFAULT_BUFF_SIZE;
+		encap_u1v(pgz->dst, bufsize);
+		while(!pz->error){
+			if(pgz->src->size == pgz->soff){
+				pgz->soff = pgz->src->size = 0;
+			}
+			if(pgz->src->size < bsz){
+				encap_u1v(pgz->src, bsz - pgz->src->size);
+				rsize = fread(pgz->src->buffer + pgz->src->size, 1, bsz - pgz->src->size, pz->file);
+				if(rsize < bsz - pgz->src->size){
+					pz->eof = 1;
+				}
+				pz->tot_in += rsize;
+				pgz->src->size += rsize;
+			}
+			if(pgz->src->size == pgz->soff){
+				break;
+			}
+			if(pz->step == 0){
+				u4i tsz;
+				tsz = pgz->src->size;
+				ret = _read_pgzf_header(pz->file, pgz->src, &pgz->soff, &pgz->zsval, &pgz->zxval);
+				if(ret != PGZF_FILETYPE_GZ && ret != PGZF_FILETYPE_PGZF){
+					if(pgz->src->size == pgz->soff){
+						pz->eof = 1;
+					} else {
+						fprintf(stderr, " -- failed in read gzip header, ret = %d in %s -- %s:%d --\n", ret, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+						pz->error = 1;
+					}
+					break;
+				} else {
+					pz->tot_in += pgz->src->size - tsz;
+				}
+				pz->step = 1;
+				continue;
+			} else if(pz->step == 2){
+				if(pgz->src->size >= pgz->soff + PGZF_TAIL_SIZE){
+					pgz->soff += PGZF_TAIL_SIZE;
+					pz->step = 0;
+					inflateReset(pz->z);
+					continue;
+				} else if(pz->eof){
+					pz->error = 2;
+					break;
+				} else {
+					memmove(pgz->src->buffer, pgz->src->buffer + pgz->soff, pgz->src->size - pgz->soff);
+					pgz->src->size -= pgz->soff;
+					pgz->soff = 0;
+				}
+			}
+			while(pgz->dst->size < bufsize && pgz->soff < pgz->src->size){
+				dsz = bufsize - pgz->dst->size;
+				ssz = pgz->src->size - pgz->soff;
+				ret = pgzf_inflate_raw_core(pz->z, pgz->dst->buffer + pgz->dst->size, &dsz, pgz->src->buffer + pgz->soff, &ssz, Z_NO_FLUSH);
+				pgz->dst->size += dsz;
+				pgz->soff += ssz;
+				if(ret == Z_STREAM_END){
+					pz->step = 2;
+					break;
+				} else if(ret != Z_OK){
+					fprintf(stderr, " -- ZERROR: %d in %s -- %s:%d --\n", ret, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+					pz->error = 1;
+					break;
+				}
+			}
+			if(pgz->dst->size == bufsize){
+				if(pgz->soff < pgz->src->size){
+					if(pz->ncpu > 1){
+						next = (pz->ridx + 1) % pz->ncpu;
+						if(pz->srcs[next]->size != 0){
+							fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+							abort();
+						}
+						append_array_u1v(pz->srcs[next], pgz->src->buffer + pgz->soff, pgz->src->size - pgz->soff);
+					}
+				}
+				pgz->soff = pgz->src->size = 0;
+				break;
+			}
+		}
+		pz->ridx ++;
+	} else if(pz->rw_mode == PGZF_MODE_R_UNKNOWN){
+		bufsize = pz->bufsize? pz->bufsize : PGZF_DEFAULT_BUFF_SIZE;
+		encap_u1v(pgz->dst, bufsize);
+		if(pgz->src->size > bufsize){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			pz->error = 1;
+			pz->ridx ++;
+			break;
+		} else if(pgz->src->size){
+			append_u1v(pgz->dst, pgz->src);
+			clear_u1v(pgz->src);
+		}
+		rsize = fread(pgz->dst->buffer + pgz->dst->size, 1, bufsize - pgz->dst->size, pz->file);
+		if(rsize < bufsize - pgz->dst->size){
+			pz->eof = 1;
+		}
+		pgz->dst->size += rsize;
+		pz->tot_in += pgz->dst->size;
+		pz->ridx ++;
+	}
+}
+thread_end_loop(pgz);
+thread_end_func(pgz);
+
+static inline PGZF* open_pgzf_writer(FILE *out, u4i buffer_size, int ncpu, int level){
+	PGZF *pz;
+	u4i i;
+	b8i offset;
+	thread_prepare(pgz);
+	pz = malloc(sizeof(PGZF));
+	if(ncpu < 1){
+		get_linux_sys_info(NULL, NULL, &ncpu);
+		if(ncpu < 1) ncpu = 8;
+	}
+	pz->ncpu = ncpu;
+	pz->ridx = 0;
+	pz->widx = 0;
+	offset = ftell(out);
+	if(offset == -1){
+		pz->offset = 0;
+		pz->seekable = 0;
+	} else {
+		pz->offset = offset;
+		pz->seekable = 1;
+	}
+	pz->file = out;
+	pz->error = 0;
+	pz->eof = 0;
+	pz->step = 0;
+	pz->rw_mode = 1; // write
+	if(buffer_size == 0) buffer_size = PGZF_DEFAULT_BUFF_SIZE;
+	pz->bufsize = (buffer_size + 0xFFFFFU) & 0xFFF00000U;
+	pz->xsize = 0;
+	pz->boffs = init_u8v(32);
+	pz->xoffs = init_u8v(32);
+	pz->z = NULL;
+	pz->dsts = calloc(pz->ncpu, sizeof(u1v*));
+	for(i=0;i<pz->ncpu;i++){
+		pz->dsts[i] = init_u1v(pz->bufsize);
+	}
+	pz->srcs = calloc(pz->ncpu, sizeof(u1v*));
+	for(i=0;i<pz->ncpu;i++){
+		pz->srcs[i] = init_u1v(pz->bufsize);
+	}
+	pz->tmp = init_u1v(32);
+	pz->tot_in  = 0;
+	pz->tot_out = 0;
+	if(level == 0) level = Z_DEFAULT_COMPRESSION; // disable level 0, set to default level 6
+	thread_beg_init(pgz, pz->ncpu);
+	pgz->pz = pz;
+	pgz->zsval = 0;
+	pgz->zxval = 0;
+	pgz->soff = 0;
+	pgz->doff = 0;
+	pgz->dst = pz->dsts[pgz->t_idx];
+	pgz->src = pz->srcs[pgz->t_idx];
+	pgz->token = 0;
+	pgz->level = level;
+	pgz->task = PGZF_TASK_NULL;
+	thread_end_init(pgz);
+	thread_export(pgz, pz);
+	return pz;
+}
+
+static inline size_t write_pgzf(PGZF *pz, void *dat, size_t len){
+	size_t off, cnt;
+	thread_prepare(pgz);
+	thread_import(pgz, pz);
+	off = 0;
+	while(off < len){
+		thread_beg_operate(pgz, pz->widx % pz->ncpu);
+		thread_wait(pgz);
+		/*
+		if(pgz->dst->size){
+			pz->tot_out += pgz->dst->size;
+			push_u8v(pz->boffs, pz->tot_out);
+			fwrite(pgz->dst->buffer, 1, pgz->dst->size, pz->file);
+			clear_u1v(pgz->dst);
+			clear_u1v(pgz->src);
+		}
+		*/
+		cnt = num_min(len - off, pz->bufsize - pgz->src->size);
+		append_array_u1v(pgz->src, dat + off, cnt);
+		off += cnt;
+		if(pgz->src->size == pz->bufsize){
+			pz->tot_in += pgz->src->size;
+			pgz->task = PGZF_TASK_DEFLATE;
+			pgz->token = pz->widx;
+			thread_wake(pgz);
+			pz->widx ++;
+		}
+	}
+	thread_export(pgz, pz);
+	return len;
+}
+
+static inline void _end_pgzf_writer(PGZF *pz){
+	u4i i, widx;
+	thread_prepare(pgz);
+	thread_import(pgz, pz);
+	widx = pz->widx;
+	for(i=0;i<=pz->ncpu;i++){ // (pz->tidx + ncpu + 1) % ncpu
+		thread_beg_operate(pgz, widx % pz->ncpu);
+		thread_wait(pgz);
+		/*
+		if(pgz->dst->size){
+			pz->tot_out += pgz->dst->size;
+			push_u8v(pz->boffs, pz->tot_out);
+			fwrite(pgz->dst->buffer, 1, pgz->dst->size, pz->file);
+			clear_u1v(pgz->dst);
+			clear_u1v(pgz->src);
+		}
+		*/
+		if(pgz->src->size){ // will force to write un-full block
+			pz->tot_in += pgz->src->size;
+			pgz->task = PGZF_TASK_DEFLATE;
+			pgz->token = pz->widx;
+			thread_wake(pgz);
+			pz->widx ++;
+		}
+		widx ++;
+	}
+	thread_export(pgz, pz);
+}
+
+static inline int write_index_pgzf(PGZF *pz){
+	u8i i, x;
+	u1i bs[6];
+	pz->xsize = pz->tot_in;
+	if(!pz->seekable) return 0;
+	if(fseek(pz->file, pz->offset + PGZF_HEAD_ZX_OFFSET, SEEK_SET) == -1){
+		perror("fseek error in write_index_pgzf");
+		return 0;
+	}
+	_num2bytes_pgzf(bs, 6, pz->xsize);
+	fwrite(bs, 1, 6, pz->file);
+	for(i=64,x=1;i+PGZF_INDEX_BIN<=pz->boffs->size;i+=PGZF_INDEX_BIN,x++){
+		push_u8v(pz->xoffs, pz->boffs->buffer[i+PGZF_INDEX_BIN]);
+		_num2bytes_pgzf(bs, 6, pz->boffs->buffer[i+PGZF_INDEX_BIN]);
+		if(fseek(pz->file, pz->offset + pz->boffs->buffer[x] + PGZF_HEAD_ZX_OFFSET, SEEK_SET) == -1){
+			perror("fseek error in write_index_pgzf");
+			return 0;
+		}
+		fwrite(bs, 1, 6, pz->file);
+	}
+	fseek(pz->file, 0, SEEK_END);
+	return 1;
+}
+
+static inline PGZF* open_pgzf_reader(FILE *in, u4i bufsize, int ncpu){
+	PGZF *pz;
+	u8i zxval;
+	b8i offset;
+	u4i i, zsval, hoff;
+	int ftype;
+	thread_prepare(pgz);
+	pz = malloc(sizeof(PGZF));
+	pz->ncpu = ncpu;
+	pz->ridx = 0;
+	pz->widx = 0;
+	offset = ftell(in);
+	if(offset == -1){
+		pz->offset = 0;
+		pz->seekable = 0;
+	} else {
+		pz->offset = offset;
+		pz->seekable = 1;
+	}
+	pz->file = in;
+	pz->eof = 0;
+	pz->error = 0;
+	pz->step = 0;
+	pz->dsts = calloc(pz->ncpu, sizeof(u1v*));
+	pz->srcs = calloc(pz->ncpu, sizeof(u1v*));
+	pz->tmp = init_u1v(32);
+	pz->tot_in  = 0;
+	pz->tot_out = 0;
+	pz->boffs = init_u8v(32);
+	pz->xoffs = init_u8v(32);
+	// recognize PGZF
+	zsval = zxval = 0;
+	hoff = 0;
+	pz->srcs[0] = init_u1v(1024);
+	ftype = _read_pgzf_header(pz->file, pz->srcs[0], &hoff, &zsval, &zxval);
+	pz->tot_in = pz->srcs[0]->size;
+	switch(ftype){
+		case PGZF_FILETYPE_GZ: pz->step = 1; pz->rw_mode = PGZF_MODE_R_GZ; break;
+		case PGZF_FILETYPE_PGZF: pz->rw_mode = PGZF_MODE_R; break;
+		default:
+			fprintf(stderr, " ** WARNNING: input file is not in gzip format **\n");
+			pz->rw_mode = PGZF_MODE_R_UNKNOWN; break;
+	}
+	if(pz->rw_mode == PGZF_MODE_R){
+		pz->z = NULL;
+		pz->xsize = zxval;
+		push_u8v(pz->boffs, zsval);
+		if(pz->seekable){
+			u8i foff;
+			foff = ftell(pz->file);
+			if(fseek(pz->file, pz->offset + zsval - 4, SEEK_SET) == -1){
+				fprintf(stderr, " ** ERROR: failed to read uncompress block size in the first block ERR(1) **\n");
+				return NULL;
+			}
+			if(fread(&pz->bufsize, 4, 1, pz->file) == 0){
+				fprintf(stderr, " ** ERROR: failed to read uncompress block size in the first block ERR(2) **\n");
+				return NULL;
+			}
+			if(fseek(pz->file, foff, SEEK_SET) == -1){
+				fprintf(stderr, " ** ERROR: failed to read uncompress block size in the first block ERR(3) **\n");
+				return NULL;
+			}
+		} else {
+			pz->bufsize = bufsize;
+		}
+	} else if(pz->rw_mode == PGZF_MODE_R_GZ){
+		pz->z = calloc(1, sizeof(z_stream));
+		inflateInit2(pz->z, -15);
+		pz->bufsize = bufsize;
+	} else {
+		pz->z = NULL;
+		pz->bufsize = bufsize;
+	}
+	if(pz->bufsize == 0) pz->bufsize = PGZF_DEFAULT_BUFF_SIZE;
+	pz->bufsize = (pz->bufsize + 0xFFFFFU) & 0xFFF00000U;
+	for(i=0;i<pz->ncpu;i++){
+		pz->dsts[i] = init_u1v(pz->bufsize);
+	}
+	if(pz->bufsize > pz->srcs[0]->size){
+		encap_u1v(pz->srcs[0], pz->bufsize - pz->srcs[0]->size);
+	}
+	for(i=1;i<pz->ncpu;i++){
+		pz->srcs[i] = init_u1v(pz->bufsize);
+	}
+	thread_beg_init(pgz, pz->ncpu);
+	pgz->pz = pz;
+	pgz->zsval = pgz->t_idx? 0 : zsval;
+	pgz->zxval = pgz->t_idx? 0 : zxval;
+	pgz->soff = pgz->t_idx? 0 : hoff;
+	pgz->doff = 0;
+	pgz->src = pz->srcs[pgz->t_idx];
+	pgz->dst = pz->dsts[pgz->t_idx];
+	pgz->level = Z_DEFAULT_COMPRESSION; // useless in inflating
+	pgz->task = PGZF_TASK_INFLATE;
+	thread_end_init(pgz);
+	thread_wake_all(pgz);
+	thread_export(pgz, pz);
+	return pz;
+}
+
+/*
+static inline _clear_pgzf_reader(PGZF *pz){
+	UNUSED(pz);
+}
+
+static inline off_t seek_pgzf(PGZF *pz, u8i offset){
+	u4i bidx, boff, xidx, xoff;
+	if(offset > pz->xsize) return -1;
+	else if(offset == pz->xsize){
+		pz->eof = 1;
+		return offset;
+	}
+	if(!pz->seekable) return -1;
+	bidx = offset / pz->bufsize;
+	boff = offset % pz->>bufsize;
+	xidx = bidx / PGZF_INDEX_BIN;
+	xoff = bidx % PGZF_INDEX_BIN;
+	if(xidx > pz->xoffs->size){
+		
+	}
+	return 0;
+}
+*/
+
+static inline size_t read_pgzf(PGZF *pz, void *dat, size_t len){
+	size_t off;
+	u4i nrun;
+	thread_prepare(pgz);
+	thread_import(pgz, pz);
+	nrun = 0;
+	for(off=0;off<len;){
+		thread_beg_operate(pgz, pz->widx % pz->ncpu);
+		thread_wait(pgz);
+		if(pz->error) break;
+		if(len - off < pgz->dst->size - pgz->doff){
+			memcpy(dat + off, pgz->dst->buffer + pgz->doff, len - off);
+			pz->tot_out += len - off;
+			pgz->doff += len - off;
+			off = len;
+			break;
+		} else if(pgz->dst->size){
+			memcpy(dat + off, pgz->dst->buffer + pgz->doff, pgz->dst->size - pgz->doff);
+			pz->tot_out += pgz->dst->size - pgz->doff;
+			off += pgz->dst->size - pgz->doff;
+			pgz->doff = pgz->dst->size;
+			pgz->task = PGZF_TASK_INFLATE;
+			thread_wake(pgz);
+			pz->widx ++;
+		} else if(pz->eof){
+			nrun ++;
+			if(nrun >= pz->ncpu){
+				break;
+			}
+		}
+	}
+	return off;
+}
+
+static inline void close_pgzf(PGZF *pz){
+	thread_prepare(pgz);
+	if(pz->rw_mode == PGZF_MODE_W){
+		_end_pgzf_writer(pz);
+	}
+	thread_import(pgz, pz);
+	thread_beg_close(pgz);
+	free_u1v(pgz->dst);
+	free_u1v(pgz->src);
+	thread_end_close(pgz);
+	free(pz->dsts);
+	free(pz->srcs);
+	free_u1v(pz->tmp);
+	switch(pz->rw_mode){
+		case PGZF_MODE_W: write_index_pgzf(pz); fflush(pz->file); break;
+		case PGZF_MODE_R: break;
+		case PGZF_MODE_R_GZ:
+			if(pz->z){
+				inflateEnd(pz->z);
+				free(pz->z);
+			}
+			break;
+	}
+	free_u8v(pz->boffs);
+	free_u8v(pz->xoffs);
+	free(pz);
+}
+
+static inline size_t read_pgzf4filereader(void *obj, void *dat, size_t len){ return read_pgzf((PGZF*)obj, dat, len); }
+static inline void close_pgzf4filereader(void *obj){
+	PGZF *pz;
+	pz = (PGZF*)obj;
+	if(pz->file != stdin){
+		fclose(pz->file);
+	}
+	return close_pgzf(pz);
+}
+
+static inline size_t write_pgzf4filewriter(void *obj, void *dat, size_t len){ return write_pgzf((PGZF*)obj, dat, len); }
+static inline void close_pgzf4filewriter(void *obj){
+	PGZF *pz;
+	pz = (PGZF*)obj;
+	return close_pgzf(pz);
+}
+
+#endif

queue.h

@@ -0,0 +1,107 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __FIFO_QUEUE_RJ_H
+#define __FIFO_QUEUE_RJ_H
+
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdio.h>
+#include "sort.h"
+#include "mem_share.h"
+#include "list.h"
+
+#define define_queue_core(list_type, e_type, size_type, inc_size)	\
+	\
+typedef struct { e_type* buffer; size_type* links; size_type head, tail, size; size_type* trash; size_type trash_size, trash_cap; size_type off; size_type cap; } list_type;	\
+	\
+static inline size_t list_type##_obj_desc_cnt(void *list, int idx){	\
+	if(idx == 0) return ((list_type*)list)->size * sizeof(e_type);	\
+	else if(idx == 1) return ((list_type*)list)->size * sizeof(size_type);	\
+	else if(idx == 2) return ((list_type*)list)->trash_size * sizeof(size_type);	\
+	else return 1;	\
+}	\
+	\
+static const obj_desc_t list_type##_obj_desc = {.tag = TOSTR(list_type##_obj_desc), .size = sizeof(list_type), .n_child = 3, .addr = {offsetof(list_type, buffer), offsetof(list_type, links), offsetof(list_type, trash)}, .desc = {(struct obj_desc_t*)&OBJ_DESC_DATA, (struct obj_desc_t*)&OBJ_DESC_DATA, (struct obj_desc_t*)&OBJ_DESC_DATA}, .cnt = list_type##_obj_desc_cnt, .post = NULL};	\
+	\
+static inline list_type* init_##list_type(){	\
+	list_type *list = (list_type*)malloc(sizeof(list_type));	\
+	list->off = 0;	\
+	list->cap = 0;	\
+	list->buffer = NULL;	\
+	list->links  = NULL;	\
+	list->trash  = NULL;	\
+	list->trash_size = 0;	\
+	list->trash_cap = 0;	\
+	list->head = 0;	\
+	list->tail = 0;	\
+	list->size = 0;	\
+	return list;	\
+}	\
+	\
+static inline size_type count_##list_type(list_type *list){ return list->size; }	\
+	\
+static inline void clear_##list_type(list_type *list){ list->trash_size = 0; list->off = 0; list->head = 0; list->tail = 0; list->size = 0; }	\
+	\
+static inline void push_##list_type(list_type *list, e_type e){	\
+	size_type idx;	\
+	if(list->trash_size){	\
+		idx = list->trash[--list->trash_size];	\
+	} else {	\
+		encap_list((void**)&(list->buffer), sizeof(e_type), list->off, list->cap, 1, 0, 0);	\
+		list->cap = encap_list((void**)&(list->links), sizeof(size_type), list->off, list->cap, 1, 0, 0);	\
+		idx = list->off ++;	\
+	}	\
+	list->buffer[idx] = e;	\
+	if(list->size == 0){	\
+		list->head = list->tail = idx;	\
+	} else {	\
+		list->links[list->tail] = idx;	\
+	}	\
+	list->links[idx] = 0;	\
+	list->tail = idx;	\
+	list->size ++;	\
+}	\
+	\
+static inline int pop_##list_type(list_type *list, e_type*e){	\
+	if(list->size == 0) return 0;	\
+	*e = list->buffer[list->head];	\
+	list->trash_cap = encap_list((void**)&list->trash, sizeof(size_type), list->trash_size, list->trash_cap, 1, 0, 0);	\
+	list->trash[list->trash_size ++] = list->head;	\
+	list->head = list->links[list->head];	\
+	list->size --;	\
+	return 1;	\
+}	\
+	\
+static inline void free_##list_type(list_type *list){	\
+	if(list->buffer) free(list->buffer);	\
+	if(list->links) free(list->links);	\
+	if(list->trash) free(list->trash);	\
+	free(list);	\
+}
+
+#define define_queue(name, e_type) define_queue_core(name, e_type, size_t, 0xFFFFFU)
+
+define_queue(u32fifo, uint32_t);
+define_queue(s32fifo, int32_t);
+define_queue(u64fifo, uint64_t);
+define_queue(s64fifo, int64_t);
+
+#endif

scripts/best_kbm_hit.pl

@@ -0,0 +1,20 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan
+#
+use strict;
+
+my $rs = [""];
+
+while(<>){
+	my @ts = split;
+	if($ts[0] eq $rs->[0]){
+		$rs = \@ts if($ts[10] > $rs->[10]);
+	} else {
+		print join("\t", @{$rs}), "\n" if($rs->[10]);
+		$rs = \@ts;
+	}
+}
+print join("\t", @{$rs}), "\n" if($rs->[10]);
+
+1;

scripts/best_minimap_hit.pl

@@ -0,0 +1,26 @@
+#!/usr/bin/perl -w
+#
+#Author: Jue Ruan
+#
+use strict;
+
+my $rs = [""];
+
+while(<>){
+	my @ts = split;
+	if($ts[0] ne $rs->[0]){
+		if(length $rs->[0]){
+			print join("\t", @{$rs}), "\n";
+		}
+		$rs = \@ts;
+	} else {
+		if($ts[9] > $rs->[9]){
+			$rs = \@ts;
+		}
+	}
+}
+if(length $rs->[0]){
+	print join("\t", @{$rs}), "\n";
+}
+
+1;

scripts/best_sam_hits4longreads.pl

@@ -0,0 +1,111 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan
+#
+use Getopt::Std;
+use strict;
+
+our ($opt_h, $opt_l, $opt_f);
+
+getopts("hl:f:");
+&usage if($opt_h);
+
+my $map_len = $opt_l || 100;
+my $map_cov = $opt_f || 0.75;
+
+my %refs = ();
+
+my %cigar_hash = (M=>[1,1], I=>[1,0], D=>[0,1], N=>[0,1], S=>[1,0], H=>[1,0], P=>[0,0], '='=>[1,1], X=>[1,1]);
+
+my @hits = ();
+
+while(<>){
+	if(/^@/){
+		print;
+		if(/^\@SQ\sSN:(\S+)\sLN:(\d+)/){
+			$refs{$1} = $2;
+		}
+		next;
+	}
+	my $str = $_;
+	my @ts = split;
+	my $tag = $ts[0];
+	my $flg = $ts[1];
+	my $ref = $ts[2];
+	next if($ref eq '*');
+	my $rlen = $refs{$ref};
+	if(not defined $rlen){
+		print;
+		die("Cannot find '$ref' in SAM Header");
+	}
+	my $tb = $ts[3];
+	my $te = $tb;
+	my $qb = 0;
+	my $qe = 0;
+	my $len = 0;
+	my $tmp = 0;
+	my $cnt = 0;
+	my $cgr = $ts[5];
+	while($cgr=~/(\d+)(\D)/g){
+		my $mov = $cigar_hash{$2};
+		if($mov->[1]){
+			$te += $1;
+			$qe += $tmp;
+			$tmp = 0;
+			if($cnt == 0){
+				$qb = $qe;
+				$cnt = 1;
+			}
+			if($mov->[0]){
+				$qe += $1;
+				$len += $1;
+			}
+		} elsif($mov->[0]){
+			$tmp += $1;
+			$len += $1;
+		}
+	}
+	if($flg & 0x10){
+		$tmp = $len - $qb;
+		$qb = $len - $qe;
+		$qe = $tmp;
+	}
+	next unless($qe - $qb >= $map_len and $te - $tb >= $map_len);
+	next unless($qe - $qb >= $map_cov * $len or $te - $tb >= $map_cov * $rlen);
+	my $hit = [$tag, $len, $qb, $qe, $ref, $rlen, $tb, $te, $str];
+	&select_best_hit($hit);
+}
+&select_best_hit(undef);
+
+1;
+
+sub usage {
+	print STDERR qq{Usage: $0 [-l 100:map_len] [-f 0.75:map_cov] <sam_file_in_query_order_with_header>\n};
+	exit 1;
+}
+
+sub select_best_hit {
+	my $hit = shift;
+	if(@hits == 0){
+		push(@hits, $hit) if($hit);
+	} elsif(not defined $hit or $hit->[0] ne $hits[-1][0]){
+		@hits = sort {$b->[3] - $b->[2] <=> $a->[3] - $a->[2]} @hits;
+		for(my $i=0;$i<@hits;$i++){
+			my $pass = 1;
+			for(my $j=0;$j<$i;$j++){
+				my $x = $hits[$i][2] < $hits[$j][2]? $hits[$j][2] : $hits[$i][2];
+				my $y = $hits[$i][3] > $hits[$j][3]? $hits[$j][3] : $hits[$i][3];
+				if($y - $x >= (1 - $map_cov) * ($hits[$i][3] - $hits[$i][2])){
+					$pass = 0;
+					last;
+				}
+			}
+			print $hits[$i][8] if($pass);
+		}
+		@hits = ();
+		push(@hits, $hit) if($hit);
+	} else {
+		push(@hits, $hit);
+	}
+}
+

scripts/dbm_index_dot.pl

@@ -0,0 +1,70 @@
+#!/usr/bin/perl -w
+#
+#Author: Ruan Jue
+#
+use strict;
+use DB_File;
+
+my $dot_file = shift or die("Usage: $0 <dot_file>\n");
+
+die("$dot_file.dbm already exists!!!") if(-e "$dot_file.dbm");
+
+open(IN, "<", $dot_file) or die;
+my %hash;
+
+tie %hash, 'DB_File', "$dot_file.dbm", O_RDWR | O_CREAT, 0644, $DB_HASH or die "Cannot open $dot_file.dbm: $!";
+
+my %nodes = ();
+my %link = ();
+
+while(<IN>){
+	s/^\s+//;
+	s/\s+$//;
+	if(/^rankdir\s*=\s*(\S+)/){
+		$hash{"rankdir"} = $1;
+		next;
+	}
+	my $desc = '';
+	while(1){
+		if(/\s*(\[[^]]+\]);?$/){
+			$_ = substr($_, 0, length($_) - length($1));
+			$desc .= $1;
+			s/\s+$//;
+		} else {
+			last;
+		}
+	}
+	my @ts = split;
+	if(@ts == 1 and length $desc){
+		$nodes{$ts[0]} = $desc;
+	} elsif(@ts >= 3 and ($ts[1] eq '->' or $ts[1] eq '-')){
+		my ($lnk1, $lnk2) = ("", "");
+		if($ts[0]=~/^(\S+?):(\S+)$/){ $ts[0] = $1; $lnk1 = $2; }
+		if($ts[2]=~/^(\S+?):(\S+)$/){ $ts[2] = $1; $lnk2 = $2; }
+		push(@{$link{$ts[0]}{$ts[2]}}, [$desc, $lnk1, $lnk2]);
+		$link{$ts[2]}{$ts[0]} = [] unless(defined $link{$ts[2]}{$ts[0]});
+	}
+}
+
+close IN;
+
+foreach my $n1 (keys %nodes){
+	$hash{$n1} = $nodes{$n1};
+}
+
+foreach my $n1 (keys %link){
+	my $hx = $link{$n1};
+	my $str = (defined $hash{$n1})? $hash{$n1} : "";
+	foreach my $n2 (keys %{$hx}){
+		my $ls = $hx->{$n2};
+		$str .= "\nN\t$n2";
+		foreach my $lk (@{$ls}){
+			$str .= "\n" . join("\t", "L", @{$lk});
+		}
+	}
+	$hash{$n1} = $str;
+}
+
+untie %hash;
+
+1;

scripts/dbm_index_fa.pl

@@ -0,0 +1,43 @@
+#!/usr/bin/perl -w
+#
+#Author: Ruan Jue
+#
+use strict;
+use DB_File;
+
+my $fasta_file = shift or die("Usage: $0 <fasta_file>\n");
+
+die("$fasta_file.dbm already exists!!!") if(-e "$fasta_file.dbm");
+
+open(IN, "<", $fasta_file) or die;
+my %seqs;
+
+tie %seqs, 'DB_File', "$fasta_file.dbm", O_RDWR | O_CREAT, 0644, $DB_HASH or die "Cannot open $fasta_file.dbm: $!";
+
+my $name = '';
+#my $seq = '';
+
+my $off = 0;
+my $len = 0;
+
+while(<IN>){
+	if(/^>(\S+)/){
+		#$seqs{$name} = $seq if($name);
+		$seqs{$name} = $off;
+		$name = $1;
+		$off += $len;
+		$len = 0;
+		#$seq = '';
+	#} else {
+		#chomp;
+		#$seq .= $_;
+	}
+	$len += length($_);
+}
+$seqs{$name} = $off if($name);
+
+untie %seqs;
+
+close IN;
+
+1;

scripts/dbm_read_dot.pl

@@ -0,0 +1,74 @@
+#!/usr/bin/perl -w
+#
+#Author: Ruan Jue
+#
+use strict;
+use Getopt::Std;
+use DB_File;
+
+our ($opt_l, $opt_h);
+
+getopts("l:h");
+
+my $level = $opt_l || 0;
+&usage if($opt_h);
+
+my $dbf = shift or &usage;
+if($dbf!~/\.dbm$/){
+	$dbf .= ".dbm" if(-e "$dbf.dbm");
+}
+
+my @nodes = @ARGV;
+
+if(@nodes == 0){
+	while(<>){
+		chomp;
+		push(@nodes, $_);
+	}
+}
+
+my %hash;
+
+tie %hash, 'DB_File', $dbf, O_RDONLY or die "Cannot open $dbf: $!";
+
+my %levels = map {$_=>0} @nodes;
+
+print "digraph {\n";
+
+print "rank=$hash{rank}\n" if(exists $hash{"rank"});
+print "node $hash{node}\n" if(exists $hash{"node"});
+print "edge $hash{edge}\n" if(exists $hash{"edge"});
+
+while(@nodes){
+	my $nd = shift @nodes;
+	my $str = $hash{$nd};
+	my @rs = split /\n/, $str;
+	print "$nd $rs[0]";
+	if($levels{$nd} == 0){
+		print " [style=filled fillcolor=yellow]\n"
+	} else {
+		print "\n";
+	}
+	my $n2 = '';
+	for(my $i=1;$i<@rs;$i++){
+		if($rs[$i]=~/^N\s(\S+)/){
+			$n2 = $1;
+			if($levels{$nd} < $level and not exists $levels{$n2}){
+				$levels{$n2} = $levels{$nd} + 1;
+				push(@nodes, $n2);
+			}
+		} elsif($rs[$i]=~/^L\s(.+)$/){
+			print " $nd -> $n2 $1\n"
+		}
+	}
+}
+
+print "}\n";
+
+untie %hash;
+
+1;
+
+sub usage {
+	die("Usage: $0 [-l TRACE_LEVEL:0] <dot_dbm_file> <node1> ...\n");
+}

scripts/dbm_read_fa.pl

@@ -0,0 +1,135 @@
+#!/usr/bin/perl -w
+#
+#Author: Ruan Jue
+#
+use strict;
+use DB_File;
+
+my $dbf = shift or die("Usage: $0 <dbm_file> [read_name1 ...]\n");
+my $faf = undef;
+if($dbf!~/\.dbm$/){
+	$dbf .= ".dbm" if(-e "$dbf.dbm");
+}
+
+my @names = @ARGV;
+
+my $list_all = 0;
+
+if(@names == 1 and $names[0] eq '#LIST'){
+	$list_all = 1;
+	@names = ();
+} elsif(@names == 0){
+	while(<>){
+		chomp;
+		push(@names, $_);
+	}
+}
+
+my @tags = ();
+foreach my $tag (@names){
+	if($tag=~/^(.+?)(\[([+-])(:(-?\d+),(-?\d+))?\])$/){
+		push(@tags, [$1, $3 eq '+'? 1:2, (defined $5)? $5:1, (defined $6)? $6:-1, 1]);
+	} elsif($tag=~/^(.+?)_([FR])(_(\d+)(_(\d+))?)?$/){
+		push(@tags, [$1, $2 eq 'F'? 1:2, (defined $4)? $4:1, (defined $6)? $4 + $6 - 1 : 0, 1]);
+	} else {
+		push(@tags, [$tag, 1, 1, -1, 0]);
+	}
+}
+
+foreach my $tag (@tags){
+	$tag->[2] = 1 if($tag->[2] < 1);
+}
+
+my %seqs;
+
+tie %seqs, 'DB_File', $dbf, O_RDONLY or die "Cannot open $dbf: $!";
+
+if($dbf=~/^(.+?)\.dbm$/){
+	$faf = $1;
+}
+
+my $fa_file = undef;
+
+foreach my $tag (@tags){
+	if(exists $seqs{$tag->[0]}){
+		#my $seq = $seqs{$tag->[0]};
+		my $seq = &read_fasta($tag->[0], $tag->[4]);
+		if($tag->[4]){
+			$tag->[3] = length($seq) if($tag->[3] <= $tag->[2]);
+			if($tag->[4]){
+				print ">", join("_", $tag->[0], $tag->[1] == 1? "F":"R", $tag->[2], $tag->[3] + 1 - $tag->[2]), "\n";
+			} else {
+				print ">$tag->[0]\n";
+			}
+			if($tag->[2] < $tag->[3]){
+				my $ss = substr($seq, $tag->[2] - 1, $tag->[3] - $tag->[2] + 1);
+				if($tag->[1] == 2){
+					$ss =~tr/ACGTacgt/TGCAtgca/;
+					$ss = reverse $ss;
+				}
+				while($ss=~/(.{1,100})/g){ print "$1\n"; }
+			}
+		} else {
+			print $seq;
+		}
+	} else {
+		warn("'$tag->[0]' was not found\n");
+	}
+}
+
+if($list_all){
+	&list_all_seqs;
+}
+
+untie %seqs;
+
+if($fa_file){
+	close $fa_file;
+}
+
+1;
+
+sub read_fasta {
+	my $tag = shift;
+	my $tidy = shift || 0;
+	my $obj = $seqs{$tag};
+	if($obj!~/^[0-9]/){
+		return $obj;
+	}
+	my $off = $obj;
+	if(not defined $fa_file){
+		if(not defined $faf){
+			die("Cannot find fasta file");
+		} else {
+			open($fa_file, "<", $faf) or die $!;
+		}
+	}
+	seek($fa_file, $off, 0);
+	my $nam = '';
+	my $seq = '';
+	while(<$fa_file>){
+		if(/^>(\S+)/){
+			last if(length $nam);
+			$nam = $1;
+			if(!$tidy){
+				$seq .= $_;
+			}
+		} else {
+			if($tidy){
+				chomp;
+			}
+			$seq .= $_;
+		}
+	}
+	if($nam ne $tag){
+		die("Broken dbm index, \"$nam\" ne \"$tag\", offset = $off");
+	}
+	return $seq;
+}
+
+sub list_all_seqs {
+	while(my ($tag, $seq) = each %seqs){
+		print "$tag\t$seq\n";
+	}
+}
+

scripts/fa2fq.pl

@@ -0,0 +1,36 @@
+#!/usr/bin/perl -w
+#
+#
+#
+use strict;
+
+my $Q = '!';
+
+my $tag = '';
+my $seq = '';
+while(<>){
+	if(/^>(.+?)$/){
+		if(length $seq){
+			print "\@$tag\n";
+			print "$seq\n";
+			print "+\n";
+			print $Q x length($seq);
+			print "\n";
+		}
+		$tag = $1;
+		$seq = '';
+	} else {
+		chomp;
+		$seq .= $_;
+	}
+}
+
+if(length $seq){
+	print "\@$tag\n";
+	print "$seq\n";
+	print "+\n";
+	print $Q x length($seq);
+	print "\n";
+}
+
+1;

scripts/fa2tab.pl

@@ -0,0 +1,20 @@
+#!/usr/bin/perl -w
+#
+use strict;
+
+my $name;
+my $seq = '';
+
+while(<>){
+	if(/>(\S+)/){
+		print "$name\t$seq\n" if(defined $name);
+		$name = $1;
+		$seq = '';
+	} else {
+		chomp;
+		$seq .= $_;
+	}
+}
+print "$name\t$seq\n" if(defined $name);
+
+1;

scripts/first_n_bases.pl

@@ -0,0 +1,18 @@
+#!/usr/bin/perl -w
+#
+use strict;
+
+my $n = shift or die("Usage: $0 <num_of_bases> [fasta_file]\n");
+
+my $m = 0;
+
+while(<>){
+	if(/^>/){
+		last if($m >= $n);
+	} else {
+		$m += length($_) - 1;
+	}
+	print;
+}
+
+1;

scripts/first_n_seqs.pl

@@ -0,0 +1,17 @@
+#!/usr/bin/perl -w
+#
+use strict;
+
+my $n = shift or die("Usage: $0 <num_of_seqs> [fasta_file]\n");
+
+my $m = 0;
+
+while(<>){
+	if(/^>/){
+		last if($m == $n);
+		$m ++;
+	}
+	print;
+}
+
+1;

scripts/fq2fa.pl

@@ -0,0 +1,12 @@
+#!/usr/bin/perl -w
+my $count=0;
+while(<>){
+	if($count%4==0){
+		print ">", substr($_, 1);
+	} elsif($count%4==1){
+		print $_;
+	}
+	$count++;
+}
+
+1;

scripts/hlcolor

@@ -0,0 +1,66 @@
+#!/usr/bin/env perl
+#
+# Author: Jue Ruan
+#
+use strict;
+
+my $color = 35;
+
+my %fg_color_map = (black=>30, red=>31, green=>32, yellow=>33, blue=>34, magenta=>35, cyan=>36);
+
+my @pats = ();
+foreach my $pat (@ARGV){
+	if($pat=~/^(\S+)=(\S+?)$/){
+		if(exists $fg_color_map{$2}){
+			$color = $fg_color_map{$2};
+			push(@pats, [$1, $color]);
+		} else {
+			print STDERR "[hlcolor] unknown color \"$2\"\n";
+			push(@pats, [$1, $color]);
+		}
+	} else {
+		push(@pats, [$pat, $color]);
+	}
+}
+
+$| = 1;
+
+while(<STDIN>){
+	foreach my $pat (@pats){
+		s/($pat->[0])/\e[1;$pat->[1]m\1\e[0m/g;
+	}
+	print;
+}
+
+1;
+
+=pod
+function highlight() {
+	declare -A fg_color_map
+	fg_color_map[black]=30
+	fg_color_map[red]=31
+	fg_color_map[green]=32
+	fg_color_map[yellow]=33
+	fg_color_map[blue]=34
+	fg_color_map[magenta]=35
+	fg_color_map[cyan]=36
+
+	fg_c=$(echo -e "\e[1;${fg_color_map[$2]}m")
+	c_rs=$'\e[0m'
+	sed -u "s/$1/$fg_c\0$c_rs/g"
+}
+
+if [ -z $1 ]; then
+	echo "Usage: $0 <sed_regex_pattern> <color:black,red,green,yellow,blue,magenta,cyan>"
+	exit
+fi
+
+PAT=$1
+COLOR=$2
+
+if [ -z $COLOR ]; then
+	COLOR=magenta
+fi
+
+highlight $PAT $COLOR
+=cut

scripts/longest_pacbio_subreads.pl

@@ -0,0 +1,45 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan
+#
+use strict;
+
+my $tag = '';
+my @seqs = (['', '']);
+
+while(<>){
+	chomp;
+	if(/^>(\S+)/){
+		my $name = $1;
+		my $comm = substr($_, length($name) + 1);
+		my $subr = '';
+		if($name=~/^(.+?)(\/\d+_\d+)$/){
+			$name = $1;
+			$subr = $2;
+		}
+		if($name eq $tag){
+			push(@seqs, [$subr, '', $comm]);
+		} else {
+			&print_longest_seq;
+			$tag = $name;
+			@seqs = ([$subr, '', $comm]);
+		}
+	} else {
+		$seqs[-1][1] .= $_;
+	}
+}
+&print_longest_seq;
+
+1;
+
+sub print_longest_seq {
+	my $idx = 0;
+	my $max = 0;
+	for(my $i=0;$i<@seqs;$i++){
+		if(length($seqs[$i][1]) > $max){
+			$idx = $i; $max = length($seqs[$i][1]);
+		}
+	}
+	return unless($max);
+	print ">$tag$seqs[$idx][0]$seqs[$idx][2] len=$max\n$seqs[$idx][1]\n";
+}

scripts/mmpoa.pl

@@ -0,0 +1,54 @@
+#!/usr/bin/env perl
+#
+# Author: Jue Ruan <ruanjue@gmail.com>
+#
+use Getopt::Std;
+use strict;
+
+our ($opt_h, $opt_t, $opt_s, $opt_m, $opt_x, $opt_p);
+
+getopts("ht:s:m:x:p:");
+
+&usage if($opt_h);
+&usage(1) if(@ARGV < 2);
+my $ref = shift || &usage(1);
+my $rst = $ref;
+$rst=~s/\.fa$//;
+$rst=~s/\.fasta$//;
+
+my $ncpu = 4;
+$ncpu = $opt_t if(defined $opt_t);
+
+my $MM = $opt_m || "minimap2";
+my $MX = $opt_x || "pb";
+my $ST = $opt_s || "samtools";
+my $WP = $opt_p || "wtpoa-cns";
+
+my $cmd = '';
+
+$cmd = "$MM -t $ncpu -x map-$MX -a $ref @ARGV | $ST view -Sb - > $rst.bam";
+&run($cmd);
+
+$cmd = "$ST sort $rst.bam $rst.srt";
+&run($cmd);
+
+$cmd = "$ST view $rst.srt.bam | $WP -t $ncpu -d $ref -i - -fo $rst.mmpoa.fa";
+&run($cmd);
+
+1;
+
+sub usage {
+	my $ret = shift || 0;
+	print qq{Usage: $0 [-t n_cpu:4] [-s samtools] [-m minimap2] [-p wtpoa-cns] [-x pb|ont] <ref.fa> <reads1.fa> [reads2.fa ...]\n};
+	exit $ret;
+}
+
+sub run {
+	my $cmd = shift;
+	system("date");
+	print "# $cmd\n";
+	if(system($cmd)){
+		die("$cmd");
+	}
+	system("date");
+}

scripts/mum_assess.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+#
+# Author: Jue Ruan
+#
+REF=$1
+CTG=$2
+# change this path to your mummer
+MUM=/public/software/mummer-323-intel
+
+if [ -z $REF ] || [ -z $CTG ]; then
+	echo "Usage: $0 <ref> <ctg>"
+	exit
+fi
+
+echo "REF:$REF"
+echo "CTG:$REF"
+
+$MUM/nucmer --mumreference -l 100 -c 1000 -d 10 --banded -D 5 $REF $CTG
+
+$MUM/delta-filter -i 95 -o 95 out.delta > out.best.delta
+
+$MUM/dnadiff -d out.best.delta
+
+$MUM/mummerplot out.best.delta --fat -f -png
+

scripts/num_n50.pl

@@ -0,0 +1,39 @@
+#!/usr/bin/perl -w
+#
+# Author: Ruan Jue
+#
+use strict;
+
+my $total = 0;
+my @nums = ();
+
+my $len = 0;
+while(<>){
+	next unless(/(\d+)/);
+	push(@nums, $1),$total+=$1 if($1);
+}
+
+print "Total: $total\n";
+
+my @nxxs = ();
+for(my $i=0;$i<=10;$i++){
+	push(@nxxs, int($total*$i*0.1));
+}
+push(@nxxs, $total + 1);
+
+my $i = 0;
+my $j = 0;
+
+@nums = sort {$b <=> $a} @nums;
+
+$len = 0;
+
+for(;$i<@nums;$i++){
+	$len += $nums[$i];
+	while($nxxs[$j] <= $len){
+		print "N".$j."0: $nums[$i]\t" . ($i + 1) . "\n";
+		$j ++;
+	}
+}
+
+1;

scripts/rename_fa.pl

@@ -0,0 +1,51 @@
+#!/usr/bin/perl -w
+#
+# Author: Ruan Jue
+#
+use strict;
+use Getopt::Std;
+
+my $prefix = $ENV{'PARAM_RENAME_FA_PREFIX'} || 'S';
+my $suffix = $ENV{'PARAM_RENAME_FA_SUFFIX'} || '';
+
+our ($opt_p, $opt_s, $opt_h, $opt_f, $opt_b, $opt_I);
+
+getopts("hp:s:f:b:I");
+die("Usage: $0 [-p name_prefix] [-s name_suffix] [-f trans_file] [-b begin_idx] [-I:include orignial_name] <fasta_file>\n") if($opt_h);
+$prefix = $opt_p if(defined $opt_p);
+$suffix = $opt_s if(defined $opt_s);
+my %hash;
+if(defined $opt_f){
+	open(IN, "<", $opt_f) or die;
+	%hash = ();
+	while(<IN>){
+		my @ts = split;
+		$hash{$ts[0]} = $ts[1];
+		#print STDERR "$ts[0]\t$ts[1]\n";
+	}
+	close IN;
+}
+
+my $idx = $opt_b || 0;
+
+while(<>){
+	if(/^>(\S+)/){
+		my $desc = substr($_, length($1) + 1);
+		my $ori = $opt_I? " $1" : "";
+		if(%hash){
+			if(exists $hash{$1}){
+				my $tag = $hash{$1};
+				print ">$tag$ori$desc", substr($_, length($1) + 1);
+			} else {
+				print;
+			}
+		} else {
+			printf(">$prefix%010d$suffix$ori$desc", $idx);
+		}
+		$idx ++;
+	} else {
+		print;
+	}
+}
+
+1;

scripts/rename_fq.pl

@@ -0,0 +1,51 @@
+#!/usr/bin/perl -w
+#
+# Author: Ruan Jue
+#
+use strict;
+use Getopt::Std;
+
+my $prefix = $ENV{'PARAM_RENAME_FA_PREFIX'} || 'S';
+my $suffix = $ENV{'PARAM_RENAME_FA_SUFFIX'} || '';
+
+our ($opt_p, $opt_s, $opt_h, $opt_f);
+
+getopts("hp:s:f:");
+die("Usage: $0 [-p name_prefix] [-s name_suffix] [-f trans_file] <fastq_file>\n") if($opt_h);
+$prefix = $opt_p if(defined $opt_p);
+$suffix = $opt_s if(defined $opt_s);
+my %hash;
+if(defined $opt_f){
+	open(IN, "<", $opt_f) or die;
+	%hash = ();
+	while(<IN>){
+		my @ts = split;
+		$hash{$ts[0]} = $ts[1];
+		print STDERR "$ts[0]\t$ts[1]\n";
+	}
+	close IN;
+}
+
+my $idx = 0;
+my $line = 0;
+while(<>){
+	if(($line % 4) == 0 and /^@(\S+)/){
+		my $comm = substr($_, length($1) + 1);
+		$idx ++;
+		if(%hash){
+			if(exists $hash{$1}){
+				my $tag = $hash{$1};
+				print "\@$tag", $comm;
+			} else {
+				print;
+			}
+		} else {
+			printf("\@$prefix%012d$suffix$comm", $idx);
+		}
+	} else {
+		print;
+	}
+	$line ++;
+}
+
+1;

scripts/rev_seq.pl

@@ -0,0 +1,34 @@
+#!/usr/bin/perl -w
+#
+#AUthor: Ruan Jue
+#
+use strict;
+
+my $seq = shift;
+
+if($seq){
+	rev_seq($seq);
+} else {
+	$seq = '';
+	while(<>){
+		if(/^>/){
+			&rev_seq($seq);
+			print; next;
+			$seq = '';
+		}
+		chomp;
+		$seq .= $_;
+	}
+	&rev_seq($seq);
+}
+
+1;
+
+sub rev_seq {
+	my $s = shift;
+	$s =~tr/ACGTacgt/TGCAtgca/;
+	$s = reverse $s;
+	while($s=~/(.{1,100})/g){
+		print "$1\n";
+	}
+}

scripts/runit.pl

@@ -0,0 +1,206 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan <ruanjue@gmail.com>
+#
+use strict;
+use warnings;
+use POSIX;
+use POSIX ":sys_wait_h";
+use Time::HiRes;
+
+my $USER = $ENV{USER};
+if(not defined $USER){
+	$USER = `whoami`;
+	chomp $USER;
+}
+my $RUNITALL = $ENV{RUNIT_ALL} || 0;
+
+my $cmd = join(" ", @ARGV);
+$cmd=~s/^\s+//;
+$cmd=~s/\s+$//;
+&usage unless($cmd);
+
+my $pagesize = POSIX::sysconf(POSIX::_SC_PAGESIZE);
+my $sleep_inv = 1000000; # in mircoseconds
+
+my $maxram = 0; # kb
+my $maxcpu = 0;
+my $retval = 0;
+my $utime = 0;
+my $stime = 0;
+my $rtime = 0;
+my $maxrss = 0;
+my $maxvsz = 0;
+
+my %exclusive = ();
+if($RUNITALL){
+	for my $proc (&get_all_process()){
+		$exclusive{$proc->[0]} = 1;
+	}
+}
+my %procs = ();
+
+&get_linux_sys_info();
+
+print STDERR " --------------------------------------------------------------------------------\n";
+print STDERR " -- runit.pl is a program launcher and minitor written by Jue Ruan <ruanjue\@gmail.com>\n";
+print STDERR " -- RAM : $maxram kB\n";
+print STDERR " -- CPU : $maxcpu cores\n";
+print STDERR " -- SYS : ", `uname -a`;
+print STDERR " -- USER: $USER\n";
+print STDERR " -- DATE: ", `date`;
+print STDERR " -- PWD : ", `pwd`;
+print STDERR " -- CMD : $cmd\n";
+print STDERR " --------------------------------------------------------------------------------\n";
+
+my $PID = open(READ, "$cmd |") or die("Cannot invoke commands: $!");
+
+my $begtime = Time::HiRes::time;
+
+if(fork() == 0){
+	while(<READ>){
+		print;
+	}
+	exit;
+}
+
+while(1){
+	my $mrss = 0;
+	my $mvsz = 0;
+	foreach my $proc (&get_all_process()){
+		my ($pid, $cm, $ppid) = @{$proc};
+		next if($exclusive{$pid});
+		my ($fail, $ut, $st, $rss, $vsz) = &get_linux_proc_info($pid);
+		next if($fail);
+		if(exists $procs{$pid}){
+			$procs{$pid}[0] = $ut;
+			$procs{$pid}[1] = $st;
+		} else {
+			print STDERR " -- RUNIT PID($pid): $ppid\t$cm\n";
+			$procs{$pid} = [$ut, $st, $cm, $ppid];
+		}
+		$mrss += $rss;
+		$mvsz += $vsz;
+	}
+	#print STDERR "$mrss\t$mvsz\n";
+	$maxrss = $mrss if($mrss > $maxrss);
+	$maxvsz = $mvsz if($mvsz > $maxvsz);
+	my $res = waitpid($PID, WNOHANG);
+	if($res == -1){
+		print STDERR " ** Error ", ($? >> 8), "\n";
+		last;
+	} elsif($res){
+		$retval = $? >> 8;
+		last;
+	} else {
+		Time::HiRes::usleep($sleep_inv);
+	}
+}
+my $endtime = Time::HiRes::time;
+$rtime = $endtime - $begtime;
+
+foreach my $pid (sort {$a <=> $b} keys %procs){
+	next if($exclusive{$pid});
+	print STDERR " -- STAT PID($pid): $procs{$pid}[3]\t$procs{$pid}[0]\t$procs{$pid}[1]\t$procs{$pid}[2]\n";
+	$utime += $procs{$pid}[0];
+	$stime += $procs{$pid}[1];
+}
+
+print STDERR " --------------------------------------------------------------------------------\n";
+printf STDERR " -- retval    %16d\n", $retval;
+printf STDERR " -- real      %16.3f\n", $rtime;
+printf STDERR " -- user      %16.3f\n", $utime;
+printf STDERR " -- sys       %16.3f\n", $stime;
+printf STDERR " -- maxrss    %16.3f kB\n", $maxrss;
+printf STDERR " -- maxvsz    %16.3f kB\n", $maxvsz;
+print STDERR " --------------------------------------------------------------------------------\n";
+
+1;
+
+sub usage {
+	print qq{Launch program and minitor the cputime and ram usage of it and its childs\n};
+	print qq{Usage: $0 \$'commands'\n};
+	exit 1;
+}
+
+sub get_linux_sys_info {
+	open(IN, "/proc/meminfo") or die;
+	while(<IN>){
+		my @ts = split;
+		if($ts[0]=~/^MemTotal/){
+			$maxram = $ts[1];
+			last;
+		}
+	}
+	close IN;
+	open(IN, "/proc/cpuinfo") or die;
+	$maxcpu = 0;
+	while(<IN>){
+		if(/^processor/){
+			$maxcpu ++;
+		}
+	}
+	close IN;
+}
+
+sub get_all_process {
+	my %ps = ();
+	my @pids = ();
+	if(my $psid = open(IN, "ps -o ppid,pid,cmd --no-headers --user $USER 2>/dev/null |")){
+		while(<IN>){
+			chomp;
+			my @ts = split /\s+/, $_, 3;
+			next if($ts[1] == $psid);
+			if($RUNITALL){
+				push(@pids, [$ts[1], $ts[2], $ts[0]]);
+			} else {
+				$ps{$ts[1]}[0] = $ts[2];
+				$ps{$ts[1]}[1] = $ts[0];
+				$ps{$ts[1]}[2] = [] if(not defined $ps{$ts[1]}[2]);
+				push(@{$ps{$ts[0]}[2]}, $ts[1]);
+			}
+		}
+		close IN;
+	}
+	if($RUNITALL){
+	} else {
+		my @stack = ($PID);
+		while(@stack){
+			my $pid = pop @stack;
+			next unless(exists $ps{$pid});
+			my $p = $ps{$pid};
+			push(@pids, [$pid, $p->[0], $p->[1]]);
+			push(@stack, @{$p->[2]});
+		}
+	}
+	return @pids;
+}
+
+sub get_linux_proc_info {
+	my $pid = shift;
+	my ($fail, $ut, $st, $rss, $vsz);
+	if(open(IN, "/proc/$pid/stat")){
+		if($_ = <IN>){
+			my @ts = split /\s/, $_;
+			#print STDERR join("|", @ts), "\n";
+			#for(my $i=0;$i<@ts;$i++){
+			#	print STDERR ($i+1), "\t", $ts[$i], "\n";
+			#}
+			$ut = $ts[13] / 100;
+			$st = $ts[14] / 100;
+			$vsz = $ts[22] / 1024;
+			$rss = $ts[23] * $pagesize / 1024;
+			$fail = 0;
+		} else {
+			$fail = 1;
+		}
+		close IN;
+	} else {
+		$fail = 1;
+	}
+	if($fail){
+		print STDERR " ** FAIL STAT($pid)\n";
+	}
+	return ($fail, $ut, $st, $rss, $vsz);
+}
+

scripts/sam2dbgcns.pl

@@ -0,0 +1,151 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan <ruanjue@gmail.com>
+#
+use strict;
+use Getopt::Std;
+
+our ($opt_h, $opt_w, $opt_s, $opt_d, $opt_q, $opt_t);
+
+getopts("hw:s:dq:t");
+
+&usage if($opt_h);
+$opt_w = 2000 unless($opt_w);
+$opt_s = 1000 unless($opt_s);
+$opt_d = 0 unless($opt_d);
+$opt_q = 0 unless($opt_q);
+$opt_t = 0 unless($opt_t);
+
+&usage if($opt_w <= $opt_s);
+
+my $ctgf = shift or &usage;
+
+my %refs = ();
+
+my $ctag = "";
+my $cdes = "";
+my $cseq = "";
+
+open(IN, "<", $ctgf) or die;
+while(<IN>){
+	chomp;
+	if(/^>(\S+)/){
+		$refs{$ctag} = [$cseq, $cdes] if(length $cseq);
+		$ctag = $1;
+		$cdes = substr($_, 1 + length($1));
+		$cseq = '';
+	} else {
+		$cseq .= $_;
+	}
+}
+$refs{$ctag} = [$cseq, $cdes] if(length $cseq);
+close IN;
+
+$ctag = "";
+my $coff = 0;
+my $wlen = $opt_w;
+my $slen = $opt_s;
+my @rseqs = ();
+my $ref = undef;
+my $rsize = 0;
+
+my $nid = 0;
+
+my $line = 0;
+
+while(<>){
+	$line ++;
+	next if(/^@/);
+	my @ts= split;
+	next if($ts[2] eq '*');
+	next if($opt_q and $ts[4] < $opt_q);
+	my $rtag = $ts[0];
+	my $rdir = (($ts[1] >> 4) & 0x01)? "-" : "+";
+	my $gtag = $ts[2];
+	my $gpos = $ts[3];
+	my $rseq = $ts[9];
+	if($opt_t){
+		my $l = 0;
+		my $r = 0;
+		if($ts[5]=~/^((\d+)S)/){
+			$l = $2;
+		}
+		if($ts[5]=~/((\d+)S)$/){
+			$r = $2;
+		}
+		$rseq = substr($rseq, $l, length($rseq) - $r - $l) if($l or $r);
+	}
+	if($gtag ne $ctag){
+		while(length $ctag and $coff < $rsize){
+			&print_frag;
+		}
+		$ctag = $gtag;
+		#print STDERR "$ctag begin at SAM line $line\n";
+		$ref = $refs{$ctag};
+		$rsize = length $ref->[0];
+		$coff = 0;
+		@rseqs = ();
+		&print_header;
+	} else {
+		while($coff + $wlen < $gpos){
+			&print_frag;
+		}
+	}
+	my $dup = 0;
+	if($opt_d){
+		for(my $i=@rseqs-1;$i>=0;$i--){
+			last if($rseqs[$i][0] != $gpos);
+			if($rseqs[$i][3] eq $rseq){
+				$dup = 1;
+				last;
+			}
+		}
+	}
+	push(@rseqs, [$gpos, $rtag, $rdir, $rseq]) unless($dup);
+}
+
+while(length $ctag and $coff < $rsize){
+	&print_frag;
+}
+
+1;
+
+sub print_header {
+	return unless(length $ctag and exists $refs{$ctag});
+	print ">$ctag$refs{$ctag}->[1]\n";
+}
+
+sub print_frag {
+	my $cend = $coff + $wlen;
+	my $cnxt = $coff + $slen;
+	$cend = $rsize if($cend > $rsize);
+	$cnxt = $rsize if($cnxt > $rsize);
+	print "E\t$coff\tN$nid\t+\t";
+	$nid ++;
+	print "N$nid\t+\n";
+	$nid ++;
+	print "S\t$ctag\_F_$coff\_", ($cend - $coff), "\t+\t$coff\t", ($cend - $coff),"\t", substr($ref->[0], $coff, $cend - $coff), "\n";
+	my @nseqs = ();
+	foreach my $r (@rseqs){
+		next if($r->[0] + length($r->[3]) < $coff);
+		if($r->[0] < $cend){
+			print "S\t$r->[1]\t$r->[2]\t0\t", length($r->[3]), "\t$r->[3]\n";
+			if($r->[0] >= $cnxt){
+				push(@nseqs, $r);
+			}
+		}
+	}
+	@rseqs = @nseqs;
+	$coff = $cnxt;
+}
+
+sub usage {
+	die(qq{Usage: $0 [options] <ctg_fa_file> [srt_sam_file]
+	Option:
+	-w <int> window size, [2000]
+	-s <int> silding size, [1000]
+	-d       remove duplication
+	-q <int> filter by mapQ, [0]
+	-t <int> trim tailing clip
+	});
+}

scripts/seq_n50.pl

@@ -0,0 +1,63 @@
+#!/usr/bin/perl -w
+#
+# Author: Ruan Jue
+#
+use strict;
+use Getopt::Std;
+
+our ($opt_h, $opt_s);
+
+getopts("hs");
+
+my $total = 0;
+my @nums = ();
+
+my $len = 0;
+while(<>){
+	if(/^>(\S+)/){
+		if($opt_s){
+			print "\t$len\n" if($len);
+			print $1;
+		}
+		push(@nums, $len),$total+=$len if($len);
+		$len = 0;
+	} else {
+		$len += length($_) - 1;
+	}
+}
+print "\t$len\n" if($opt_s and $len);
+push(@nums, $len),$total+=$len if($len);
+
+my $n_seq = @nums;
+my $avg = sprintf("%0.2f", $total / $n_seq);
+
+print "Total: $total\n";
+print "Count: $n_seq\n";
+print "Average: $avg\n";
+
+my @nxxs = ();
+for(my $i=0;$i<=10;$i++){
+	push(@nxxs, int($total*$i*0.1));
+}
+push(@nxxs, $total + 1);
+
+my $i = 0;
+my $j = 0;
+
+@nums = sort {$b <=> $a} @nums;
+
+my $median = $nums[int($n_seq / 2)];
+
+print "Median: $median\n";
+
+$len = 0;
+
+for(;$i<@nums;$i++){
+	$len += $nums[$i];
+	while($nxxs[$j] <= $len){
+		print "N".$j."0: $nums[$i]\t". ($i + 1) . "\n";
+		$j ++;
+	}
+}
+
+1;

scripts/split_seqs_2.pl

@@ -0,0 +1,40 @@
+#!/usr/bin/perl -w
+#
+#
+use strict;
+
+my $cnt = shift or die("Usage: $0 <n_files> <fasta_file> [gzip:0]\n");
+my $inf = shift or die("Usage: $0 <n_files> <fasta_file> [gzip:0]\n");
+my $zip = shift || 0;
+$zip = 0 unless($zip eq '1' or $zip eq 'y');
+my $ouf = $inf;
+if($ouf eq '-'){
+	$ouf = 'unnamed';
+} else {
+	unshift @ARGV, $inf;
+}
+
+my @fhs = ();
+
+for(my $i=1;$i<=$cnt;$i++){
+	my $fh;
+	if($zip){
+		open($fh, "| gzip -c >$ouf.shuffle$i.gz") or die;
+	} else {
+		open($fh, ">", "$ouf.shuffle$i") or die;
+	}
+	push(@fhs, $fh);
+}
+
+my $n = -1;
+while(<>){
+	$n ++ if(/^>/);
+	my $fh = $fhs[$n % $cnt];
+	print $fh $_;
+}
+
+foreach my $fh (@fhs) {
+	close $fh;
+}
+
+1;

scripts/split_seqs_3.pl

@@ -0,0 +1,26 @@
+#!/usr/bin/perl -w
+#
+#
+use strict;
+
+my $cnt = shift or die("Usage: $0 <parts> <index> <fasta_file>\n");
+my $idx = shift or die("Usage: $0 <parts> <index> <fasta_file>\n");
+
+$idx --;
+
+my $ns = 0;
+my $print = 0;
+
+while(<>){
+	if(/^>/){
+		if($ns % $cnt == $idx){
+			$print = 1;
+		} else {
+			$print = 0;
+		}
+		$ns ++;
+	}
+	print if($print);
+}
+
+1;

scripts/wtdbg-dot2gfa.pl

@@ -0,0 +1,42 @@
+#!/usr/bin/perl -w
+#
+# Author: Jue Ruan <ruanjue@gmail.com>
+#
+use strict;
+
+=pod
+Transform <dbg>.[1/2/3/frg/ctg].dot into GFA format. <dbg>.[1/2/3/frg/ctg].dot is the assembly graph from wtdbg
+TAG:
+	gl: gap length
+	rc: read count to support the link
+	sq: sequence can be substr from <rd_name>_<FR>_<off>_<len>, F fwd, R rev, off is based on the fwd strand
+=cut
+
+while(<>){
+	chomp;
+	next unless(/^([NF]\d+)\s(->\s([NF]\d+)\s)?\[([^\]]+)\]$/);
+	my $n1 = $1;
+	my $n2 = $3;
+	my $label = $4;
+	if(defined $n2){
+		if($label=~/label=\"(\S)(\S):(\d+):(\-?\d+)\"/){
+			if($4 >= 0){
+				print "L\t$n1\t$1\t$n2\t$2\t0S\tgl:i:$4\trc:i:$3\n";
+			} else {
+				print "L\t$n1\t$1\t$n2\t$2\t". (0 - $4) . "M\trc:i:$3\n";
+			}
+		} else {
+			die("Bad format: $_");
+		}
+	} else {
+		if($label=~/\{N\d+\s(\d+)\s\|\s(\S+)\s\|\s([FR])_(\d+)_(\d+)\}/){
+			print "S\t$n1\t*\tLN:i:$5\tsq:Z:$2_$3_$4_$5\n"
+		} elsif($label=~/\{F\d+\s(\d+)\s(\d+)\/(\d+)\s\|/){
+			print "S\t$n1\t*\tLN:i:$2\n";
+		} else {
+			die("Bad format: $_");
+		}
+	}
+}
+
+1;

sort.h

@@ -0,0 +1,559 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+ 
+#ifndef __SORT_RJ_H
+#define __SORT_RJ_H
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#define cmp_2nums_proc(a, b) if((a) < (b)) return -1; else if((a) > (b)) return 1;
+#define num_cmp_script(e1, e2, obj, val_macro) ((val_macro(e1, obj) == val_macro(e2, obj))? 0 : ((val_macro(e1, obj) < val_macro(e2, obj))? -1 : 1))
+#define cmpgt_2nums_proc(a, b) if((a) < (b)) return 0; else if((a) > (b)) return 1;
+
+#define define_bubble_sort(name, e_type, is_greater_func)	\
+static inline void name(e_type* list, size_t size, void *ref){	\
+	size_t i, j, n;	\
+	e_type t;	\
+	i = 0;	\
+	while(i < size){	\
+		n = 0;	\
+		for(j=size-1;j>i;j--){	\
+			if(is_greater_func(list[j-1], list[j], ref) > 0){	\
+				t = list[j-1]; list[j-1] = list[j]; list[j] = t;	\
+				n = 1;	\
+			}	\
+		}	\
+		if(n == 0) break;	\
+		i ++;	\
+	}	\
+	if(ref == ref) return;	\
+}
+
+#define bubble_sort_array(rs, size, e_type, is_a_greater_than_b)	\
+do {	\
+	size_t bubble_i, bubble_j, bubble_n, bubble_size;	\
+	e_type a, b;	\
+	bubble_size = size;	\
+	for(bubble_i=0;bubble_i<bubble_size;bubble_i++){	\
+		bubble_n = 0;	\
+		for(bubble_j=bubble_size-1;bubble_j>bubble_i;bubble_j--){	\
+			a = (rs)[bubble_j - 1];	\
+			b = (rs)[bubble_j];	\
+			if((int)(is_a_greater_than_b) > 0){	\
+				(rs)[bubble_j] = a; (rs)[bubble_j - 1] = b;	\
+				bubble_n = 1;	\
+			}	\
+		}	\
+		if(bubble_n == 0) break;	\
+	}	\
+} while(0)
+
+#define divide_array(rs_ary, rs_size, e_type, is_a_greater_than_b, ret_val)	\
+do {	\
+	e_type *_rs;	\
+	_rs = (e_type*)(rs_ary);	\
+	size_t s, e, i, j, m;	\
+	e_type p, t, a, b;	\
+	if((rs_size) < 2){ (ret_val) = 0; break; }	\
+	{	\
+		s = 0;	\
+		e = (rs_size) - 1;	\
+		m = s + (e - s) / 2;	\
+		a = _rs[s]; b = _rs[m];	\
+		if(is_a_greater_than_b){ t = _rs[s]; _rs[s] = _rs[m]; _rs[m] = t; }	\
+		a = _rs[m]; b = _rs[e];	\
+		if(is_a_greater_than_b){	\
+			t = _rs[e]; _rs[e] = _rs[m]; _rs[m] = t;	\
+			a = _rs[s]; b = _rs[m];	\
+			if(is_a_greater_than_b){ t = _rs[s]; _rs[s] = _rs[m]; _rs[m] = t; }	\
+		}	\
+		p = _rs[m];	\
+		i = s + 1; j = e - 1;	\
+		while(1){	\
+			a = p;	\
+			while(b = _rs[i], (is_a_greater_than_b)) i ++;	\
+			b = p;	\
+			while(a = _rs[j], (is_a_greater_than_b)) j --;	\
+			if(i < j){	\
+				t = _rs[i]; _rs[i] = _rs[j]; _rs[j] = t;	\
+				i ++; j --;	\
+			} else break;	\
+		}	\
+		if(i == j){ i ++; j --; }	\
+		(ret_val) = i;	\
+	}	\
+} while(0)
+
+#define sort_array(rs_ary, rs_size, e_type, is_a_greater_than_b)	\
+do {	\
+	e_type *_rs;	\
+	_rs = (e_type*)(rs_ary);	\
+	size_t _qsort_n;	\
+	_qsort_n = rs_size;	\
+	size_t s, e, i, j, m, stack[64][2], x;	\
+	e_type p, t, a, b;	\
+	if(_qsort_n < 2) break;	\
+	x = 0;	\
+	stack[x][0] = 0; stack[x][1] = _qsort_n - 1; x ++;	\
+	while(x){	\
+		x --; s = stack[x][0]; e = stack[x][1];	\
+		m = s + (e - s) / 2;	\
+		a = _rs[s]; b = _rs[m];	\
+		if(is_a_greater_than_b){ t = _rs[s]; _rs[s] = _rs[m]; _rs[m] = t; }	\
+		a = _rs[m]; b = _rs[e];	\
+		if(is_a_greater_than_b){	\
+			t = _rs[e]; _rs[e] = _rs[m]; _rs[m] = t;	\
+			a = _rs[s]; b = _rs[m];	\
+			if(is_a_greater_than_b){ t = _rs[s]; _rs[s] = _rs[m]; _rs[m] = t; }	\
+		}	\
+		p = _rs[m];	\
+		i = s + 1; j = e - 1;	\
+		while(1){	\
+			a = p;	\
+			while(b = _rs[i], (is_a_greater_than_b)) i ++;	\
+			b = p;	\
+			while(a = _rs[j], (is_a_greater_than_b)) j --;	\
+			if(i < j){	\
+				t = _rs[i]; _rs[i] = _rs[j]; _rs[j] = t;	\
+				i ++; j --;	\
+			} else break;	\
+		}	\
+		if(i == j){ i ++; j --; }	\
+		if(j - s > e - i){	\
+			if(s + 4 < j){ stack[x][0] = s; stack[x][1] = j; x ++; }	\
+			if(i + 4 < e){ stack[x][0] = i; stack[x][1] = e; x ++; }	\
+		} else {	\
+			if(i + 4 < e){ stack[x][0] = i; stack[x][1] = e; x ++; }	\
+			if(s + 4 < j){ stack[x][0] = s; stack[x][1] = j; x ++; }	\
+		}	\
+	}	\
+	for(i=0;i<_qsort_n;i++){	\
+		x = 0;	\
+		for(j=_qsort_n-1;j>i;j--){	\
+			a = _rs[j - 1]; b = _rs[j];	\
+			if(is_a_greater_than_b){ t = _rs[j - 1]; _rs[j - 1] = _rs[j]; _rs[j] = t; x = 1; }	\
+		}	\
+		if(x == 0) break;	\
+	}	\
+} while(0)
+
+#define sort_array_adv(rs_size, is_a_greater_than_b, swap_expr)	\
+do {	\
+	size_t _qsort_n;	\
+	_qsort_n = rs_size;	\
+	size_t s, e, i, j, m, stack[64][2], x, a, b;	\
+	if(_qsort_n < 2) break;	\
+	x = 0;	\
+	stack[x][0] = 0; stack[x][1] = _qsort_n - 1; x ++;	\
+	while(x){	\
+		x --; s = stack[x][0]; e = stack[x][1];	\
+		m = s + (e - s) / 2;	\
+		a = s; b = m;	\
+		if(is_a_greater_than_b){ swap_expr; }	\
+		a = m; b = e;	\
+		if(is_a_greater_than_b){	\
+			swap_expr;	\
+			a = s; b = m;	\
+			if(is_a_greater_than_b){ swap_expr; }	\
+		}	\
+		i = s + 1; j = e - 1;	\
+		while(1){	\
+			a = m;	\
+			while(b = i, (is_a_greater_than_b)) i ++;	\
+			b = m;	\
+			while(a = j, (is_a_greater_than_b)) j --;	\
+			if(i < j){	\
+				if(m == i) m = j;	\
+				else if(m == j) m = i;	\
+				a = i; b = j; \
+				swap_expr; \
+				i ++; j --;	\
+			} else break;	\
+		}	\
+		if(i == j){ i ++; j --; }	\
+		if(s + 4 < j){ stack[x][0] = s; stack[x][1] = j; x ++; }	\
+		if(i + 4 < e){ stack[x][0] = i; stack[x][1] = e; x ++; }	\
+	}	\
+	for(i=0;i<_qsort_n;i++){	\
+		x = 0;	\
+		for(j=_qsort_n-1;j>i;j--){	\
+			a = j - 1; b = j;	\
+			if(is_a_greater_than_b){ swap_expr; x = 1; }	\
+		}	\
+		if(x == 0) break;	\
+	}	\
+} while(0)
+
+// Must #include "thread.h" and "list.h"
+#define psort_array(rs_ary, rs_size, e_type, ncpu, is_a_greater_than_b)	\
+do {	\
+	thread_beg_def(psrt);	\
+	e_type *rs;	\
+	size_t beg, end, div;	\
+	int divide;	\
+	thread_end_def(psrt);	\
+	thread_beg_func_inline(psrt);	\
+	thread_beg_loop(psrt);	\
+	if(psrt->divide){	\
+		divide_array(psrt->rs + psrt->beg, psrt->end - psrt->beg, e_type, is_a_greater_than_b, psrt->div);	\
+		psrt->div += psrt->beg;	\
+	} else {	\
+		sort_array(psrt->rs + psrt->beg, psrt->end - psrt->beg, e_type, is_a_greater_than_b);	\
+	}	\
+	thread_end_loop(psrt);	\
+	thread_end_func(psrt);	\
+	thread_preprocess(psrt);	\
+	e_type *_rs;	\
+	_rs = (e_type*)(rs_ary);	\
+	size_t _qsort_n, _min_blk_size;	\
+	_qsort_n = rs_size;	\
+	_min_blk_size = _qsort_n / ncpu / 8;	\
+	if(_min_blk_size < 1024) _min_blk_size = 1024;	\
+	if(_qsort_n < ((uint32_t)(ncpu)) * 32){	\
+		sort_array(rs_ary, rs_size, e_type, is_a_greater_than_b);	\
+		break;	\
+	}	\
+	thread_beg_init(psrt, (int)(ncpu));	\
+	psrt->rs = _rs; psrt->beg = psrt->end = psrt->div = 0; psrt->divide = 0;	\
+	thread_end_init(psrt);	\
+	u64list *stacks[2];	\
+	int x;	\
+	stacks[0] = init_u64list(32);	\
+	stacks[1] = init_u64list(32);	\
+	push_u64list(stacks[0], 0);	\
+	push_u64list(stacks[1], _qsort_n);	\
+	x = 0;	\
+	while(stacks[0]->size || x > 0){	\
+		thread_waitfor_one_idle(psrt);	\
+		if(psrt->divide){	\
+			if(psrt->div - psrt->beg <= psrt->end - psrt->div){	\
+				push_u64list(stacks[0], psrt->beg);	\
+				push_u64list(stacks[1], psrt->div);	\
+				push_u64list(stacks[0], psrt->div);	\
+				push_u64list(stacks[1], psrt->end);	\
+			} else {	\
+				push_u64list(stacks[0], psrt->div);	\
+				push_u64list(stacks[1], psrt->end);	\
+				push_u64list(stacks[0], psrt->beg);	\
+				push_u64list(stacks[1], psrt->div);	\
+			}	\
+			x --;	\
+			psrt->divide = 0;	\
+		} else if(stacks[0]->size){	\
+			psrt->beg = stacks[0]->buffer[--stacks[0]->size];	\
+			psrt->end = stacks[1]->buffer[--stacks[1]->size];	\
+			psrt->divide = (psrt->end - psrt->beg > _min_blk_size);	\
+			if(psrt->divide) x ++;	\
+			thread_wake(psrt);	\
+		}	\
+	}	\
+	thread_waitfor_all_idle(psrt);	\
+	thread_beg_close(psrt);	\
+	thread_end_close(psrt);	\
+	free_u64list(stacks[0]);	\
+	free_u64list(stacks[1]);	\
+} while(0)
+
+#define quick_median_array(_rs, _rs_size, e_type, expr)	\
+({	\
+	e_type key;	\
+	do {	\
+		e_type *rs;	\
+		rs = (e_type*)(_rs);	\
+		size_t size;	\
+		size = (size_t)(_rs_size);	\
+		size_t i, j, beg, mid, end;	\
+		if(size == 0){	\
+			memset(&key, 0, sizeof(e_type));	\
+			break;	\
+		}	\
+		beg = 0;	\
+		end = size - 1;	\
+		e_type tmp, a, b;	\
+		while(beg < end){	\
+			mid = beg + (end - beg) / 2;	\
+			a = rs[beg]; b = rs[mid];	\
+			if(expr){ tmp = rs[beg]; rs[beg] = rs[mid]; rs[mid] = tmp; }	\
+			a = rs[mid]; b = rs[end];	\
+			if(expr){	\
+				tmp = rs[end]; rs[end] = rs[mid]; rs[mid] = tmp;	\
+				a = rs[beg]; b = rs[mid];	\
+				if(expr){ tmp = rs[beg]; rs[beg] = rs[mid]; rs[mid] = tmp; }	\
+			}	\
+			key = rs[mid];	\
+			i = beg + 1; j = end - 1;	\
+			while(1){	\
+				a = key;	\
+				while(b = rs[i], (expr)) i ++;	\
+				b = key;	\
+				while(a = rs[j], (expr)) j --;	\
+				if(i < j){	\
+					tmp = rs[i]; rs[i] = rs[j]; rs[j] = tmp;	\
+					i ++; j --;	\
+				} else break;	\
+			}	\
+			if(i == j){ i ++; j --; }	\
+			if(i <= size / 2) beg = i;	\
+			else end = j;	\
+		}	\
+		key = rs[size/2];	\
+	} while(0);	\
+	key;	\
+})
+
+
+#define apply_array(rs, rs_size, e_type, expression)	\
+do {	\
+	size_t _i, _rs_size;	\
+	e_type a;	\
+	_rs_size = rs_size;	\
+	for(_i=0;_i<_rs_size;_i++){	\
+		a = (rs)[_i];	\
+		expression;	\
+	}	\
+} while(0)
+
+#define ref_apply_array(rs, rs_size, e_type, expression)	\
+do {	\
+	size_t _i, _rs_size;	\
+	e_type *a;	\
+	_rs_size = rs_size;	\
+	for(_i=0;_i<_rs_size;_i++){	\
+		a = (rs) + _i;	\
+		(expression);	\
+	}	\
+} while(0)
+
+#define locate_array(rs, rs_size, e_type, expr)		\
+({	\
+	 size_t _i, _size;	\
+	 e_type a;	\
+	 _size = rs_size;	\
+	 for(_i=0;_i<_size;_i++){	\
+		 a = rs[_i];	\
+		 if(expr) break;	\
+	 };	\
+	 _i;	\
+ })
+
+// sort the array according to bool value (true then flase), and return the size of trues
+#define apply_xchg_array(rs, rs_size, e_type, expr)	\
+({	\
+	size_t _i, _j, _size;	\
+	e_type a;	\
+	_size = rs_size;	\
+	for(_i=_j=0;_i<_size;_i++){	\
+		a = (rs)[_i];	\
+		if(!(expr)) continue;	\
+		if(_j < _i){	\
+			a = (rs)[_j];	\
+			(rs)[_j] = (rs)[_i];	\
+			(rs)[_i] = a;	\
+		}	\
+		_j ++;	\
+	}	\
+	_j;	\
+})
+
+#define define_quick_sort(name, e_type, is_greater_func)	\
+static inline void name(e_type *rs, size_t n, void *obj){	\
+	size_t s, e, i, j, m, stack[64][2], x;	\
+	e_type p, t;	\
+	if(n < 2) return;	\
+	x = 0;	\
+	stack[x][0] = 0; stack[x][1] = n - 1; x ++;	\
+	while(x){	\
+		x --; s = stack[x][0]; e = stack[x][1];	\
+		m = s + (e - s) / 2;	\
+		if(is_greater_func(rs[s], rs[m], obj) > 0){ t = rs[s]; rs[s] = rs[m]; rs[m] = t; }	\
+		if(is_greater_func(rs[m], rs[e], obj) > 0){	\
+			t = rs[e]; rs[e] = rs[m]; rs[m] = t;	\
+			if(is_greater_func(rs[s], rs[m], obj) > 0){ t = rs[s]; rs[s] = rs[m]; rs[m] = t; }	\
+		}	\
+		p = rs[m];	\
+		i = s + 1; j = e - 1;	\
+		while(1){	\
+			while(is_greater_func(p, rs[i], obj) > 0) i ++;	\
+			while(is_greater_func(rs[j], p, obj) > 0) j --;	\
+			if(i < j){	\
+				t = rs[i]; rs[i] = rs[j]; rs[j] = t;	\
+				i ++; j --;	\
+			} else break;	\
+		}	\
+		if(i == j){ i ++; j --; }	\
+		if(j - s > e - i){	\
+			if(s + 4 < j){ stack[x][0] = s; stack[x][1] = j; x ++; }	\
+			if(i + 4 < e){ stack[x][0] = i; stack[x][1] = e; x ++; }	\
+		} else {	\
+			if(i + 4 < e){ stack[x][0] = i; stack[x][1] = e; x ++; }	\
+			if(s + 4 < j){ stack[x][0] = s; stack[x][1] = j; x ++; }	\
+		}	\
+	}	\
+	for(i=0;i<n;i++){	\
+		x = 0;	\
+		for(j=n-1;j>i;j--){	\
+			if(is_greater_func(rs[j - 1], rs[j], obj) > 0){ t = rs[j - 1]; rs[j - 1] = rs[j]; rs[j] = t; x = 1; }	\
+		}	\
+		if(x == 0) break;	\
+	}	\
+	if(obj == obj) return;	\
+}
+
+#define define_merge(name, e_type, cmp_func, output_func)	\
+static inline void name(e_type *list1, size_t size1, e_type *list2, size_t size2, void *ref){	\
+	size_t i, j;	\
+	i = j = 0;	\
+	while(i < size1 && j < size2){	\
+		if(cmp_func(list1[i], list2[j], ref) <= 0){	\
+			output_func(list1[i], ref);	\
+			i ++;	\
+		} else {	\
+			output_func(list2[j], ref);	\
+			j ++;	\
+		}	\
+	}	\
+	while(i < size1){ output_func(list1[i++], ref); }	\
+	while(j < size2){ output_func(list2[j++], ref); }	\
+}	\
+	\
+static inline size_t name##_files(FILE **files, int n, void *ref){	\
+	e_type *es;	\
+	int *flags, i, min;	\
+	size_t ret;	\
+	ret = 0;	\
+	es = malloc(sizeof(e_type) * n);	\
+	flags = malloc(sizeof(int) * n);	\
+	for(i=0;i<n;i++) flags[i] = 0;	\
+	while(1){	\
+		min = -1;	\
+		for(i=0;i<n;i++){	\
+			if(flags[i] == 0){	\
+				flags[i] = (fread(es + i, sizeof(e_type), 1, files[i]) == 1)? 1 : 2;	\
+			}	\
+			if(flags[i] == 1){	\
+				if(min == -1){	\
+					min = i;	\
+				} else if(cmp_func(es[i], es[min], ref) <= 0){	\
+					min = i;	\
+				}	\
+			}	\
+		}	\
+		if(min == -1) break;	\
+		output_func(es[min], ref);	\
+		flags[min] = 0;	\
+		ret ++;	\
+	}	\
+	free(es);	\
+	free(flags);	\
+	return ret;	\
+}
+
+#define define_unique_merge(name, e_type, cmp_func, output_func)	\
+static inline void name(e_type *list1, size_t size1, e_type *list2, size_t size2, void *ref){	\
+	size_t i, j;	\
+	i = j = 0;	\
+	while(i < size1 && j < size2){	\
+		switch(cmp_func(list1[i], list2[j])){	\
+			case 0:  output_func(list1[i++], ref); j ++; break;	\
+			case -1: output_func(list1[i++], ref); break;	\
+			default: output_func(list2[j++], ref);	\
+		}	\
+	}	\
+	while(i < size1){ output_func(list1[i++], ref); }	\
+	while(j < size2){ output_func(list2[j++], ref); }	\
+}
+
+#define reverse_array(_rs, _rs_size, e_type)	\
+do {	\
+	size_t i, n;	\
+	n = (_rs_size);	\
+	e_type* __rs = (_rs);	\
+	e_type e;	\
+	for(i=0;i<n>>1;i++){	\
+		e = __rs[i]; __rs[i] = __rs[n-1-i]; __rs[n-1-i] = e;	\
+	}	\
+} while(0)
+
+#define define_reverse_array(name, e_type)	\
+static inline void name(e_type *list, size_t size){	\
+	size_t i, j;	\
+	e_type t;	\
+	if(size == 0) return;	\
+	i = 0;	\
+	j = size - 1;	\
+	while(i < j){	\
+		t = list[i]; list[i] = list[j]; list[j] = t;	\
+		i ++; j --;	\
+	}	\
+}
+
+#define define_apply_array(name, e_type, apply_func)	\
+static inline size_t name(e_type *list, size_t size, void *ref){	\
+	size_t i, ret;	\
+	ret = 0;	\
+	for(i=0;i<size;i++){	\
+		ret += apply_func(list[i], ref);	\
+	}	\
+	return ret;	\
+	ref = NULL;	\
+}
+
+#define define_search_array(name, e_type, cmp_func)	\
+static inline long long name(e_type *array, long long size, e_type key, void *ref){	\
+	long long i, j, m;	\
+	i = 0;	\
+	j = size;	\
+	while(i < j){	\
+		m = i + (j - i) / 2;	\
+		if(cmp_func(array[m], key, ref) < 0){	\
+			i = m + 1;	\
+		} else {	\
+			j = m;	\
+		}	\
+	}	\
+	if(i < size && cmp_func(array[i], key, ref) == 0) return i;	\
+	else return - (i + 1);	\
+	if(ref) return 0;	\
+}
+
+#define bsearch_array(_rs_array, _rs_size, e_type, _bs_ret, is_a_less_than_your)	\
+	\
+do {	\
+	e_type*_bs_rs;	\
+	e_type a;	\
+	_bs_rs = (e_type*)(_rs_array);	\
+	size_t _bs_size;	\
+	_bs_size = _rs_size;	\
+	size_t _bs_i, _bs_j, _bs_m;	\
+	_bs_i = 0;	\
+	_bs_j = _bs_size;	\
+	while(_bs_i < _bs_j){	\
+		_bs_m = _bs_i + (_bs_j - _bs_i) / 2;	\
+		a = _bs_rs[_bs_m];	\
+		if(is_a_less_than_your){	\
+			_bs_i = _bs_m + 1;	\
+		} else {	\
+			_bs_j = _bs_m;	\
+		}	\
+	}	\
+	_bs_ret = _bs_i;	\
+} while(0)
+
+#endif

thread.h

@@ -0,0 +1,401 @@
+/*
+ * 
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __THEAD_RJ_H
+#define __THEAD_RJ_H
+
+#include <pthread.h>
+#include <assert.h>
+#include <stdlib.h>
+
+#define lock_cmpxchg(location, value, comparand)    \
+({	\
+__typeof (*location) _result;	\
+if(sizeof(*location) == 1){	\
+	__asm__ __volatile__ (	\
+		"lock\n\t"	\
+		"cmpxchgb %b2,(%1)"	\
+		:"=a" (_result)	\
+		:"r"  (location), "r" (value), "a" (comparand)	\
+		:"memory", "cc");	\
+} else if(sizeof(*location) == 2){	\
+	__asm__ __volatile__ (	\
+		"lock\n\t"	\
+		"cmpxchgw %w2,(%1)"	\
+		:"=a" (_result)	\
+		:"r"  (location), "r" (value), "a" (comparand)	\
+		:"memory", "cc");	\
+} else if(sizeof(*location) == 4){	\
+	__asm__ __volatile__ (	\
+		"lock\n\t"	\
+		"cmpxchgl %w2,(%1)"	\
+		:"=a" (_result)	\
+		:"r"  (location), "r" (value), "a" (comparand)	\
+		:"memory", "cc");	\
+} else {	\
+	__asm__ __volatile__ (	\
+		"lock\n\t"	\
+		"cmpxchgq %2,(%1)"	\
+		:"=a" (_result)	\
+		:"r"  (location), "r" (value), "a" (comparand)	\
+		:"memory", "cc");	\
+}	\
+_result;	\
+})
+
+#define thread_typeof(tname) (struct tname##_struct)
+
+#define thread_begin_def(tname)	\
+struct tname##_struct {		\
+	struct tname##_struct *tname##_params;	\
+	struct tname##_struct *tname##_array;	\
+	int n_cpu;	\
+	int t_idx;	\
+	volatile int running;	\
+	volatile int state;	\
+	volatile int once;	\
+	pthread_mutex_t *mutex_lock;	\
+	pthread_rwlock_t *rw_lock;	\
+	pthread_mutex_t _COND_LOCK;	\
+	pthread_cond_t _COND
+#define thread_end_def(tname) }
+
+#define thread_def_shared_vars(tname)	\
+	struct tname##_struct *tname##_params;	\
+	pthread_t *tname##_pids;	\
+	pthread_mutex_t *tname##_mlock;	\
+	pthread_rwlock_t *tname##_rwlock;	\
+	struct tname##_struct *tname;	\
+	int tname##_i;	\
+	int tname##_j;	\
+	int tname##_var_next
+
+#define thread_beg_def(tname) thread_begin_def(tname)
+
+#define thread_begin_func_core(tname) inline void* thread_##tname##_func(void *obj){\
+	struct tname##_struct * tname = (struct tname##_struct *)obj;\
+	int tname##_var_i;\
+	struct tname##_struct * tname##_params;	\
+	tname##_params = tname->tname##_params;	\
+	if(tname##_params + tname->t_idx != tname){	\
+		fprintf(stderr, " --  Unexcepted error in thread [%s] in %s -- %s:%d --\n", #tname, __FUNCTION__, __FILE__, __LINE__);	\
+	}
+#define thread_begin_func(tname) static thread_begin_func_core(tname)
+#define thread_beg_func(tname) thread_begin_func(tname)
+#define thread_beg_func_inline(tname) inline void* thread_##tname##_func(void *obj){\
+	struct tname##_struct * tname = (struct tname##_struct *)obj;\
+	int tname##_var_i;\
+	struct tname##_struct * tname##_params;	\
+	tname##_params = tname->tname##_params;	\
+	if(tname##_params + tname->t_idx != tname){	\
+		fprintf(stderr, " --  Unexcepted error in thread [%s] in %s -- %s:%d --\n", #tname, __FUNCTION__, __FILE__, __LINE__);	\
+	}
+#define thread_begin_loop(tname)	\
+	pthread_mutex_lock(&tname->_COND_LOCK);	\
+	tname->state = 0;	\
+	pthread_cond_signal(&tname->_COND);	\
+	pthread_mutex_unlock(&tname->_COND_LOCK);	\
+	while(tname->running){	\
+		if(tname->state != 1){	\
+			struct timespec _timeout;	\
+			pthread_mutex_lock(&tname->_COND_LOCK);	\
+			clock_gettime(CLOCK_REALTIME, &_timeout);	\
+			_timeout.tv_nsec += 1000000;	\
+			pthread_cond_timedwait(&tname->_COND, &tname->_COND_LOCK, &_timeout);	\
+			pthread_mutex_unlock(&tname->_COND_LOCK);	\
+			continue;	\
+		}	\
+		for(tname##_var_i=0;tname##_var_i<1;tname##_var_i++){
+#define thread_beg_loop(tname) thread_begin_loop(tname)
+#define thread_begin_syn(tname) pthread_mutex_lock(tname->mutex_lock)
+#define thread_beg_syn(tname) thread_begin_syn(tname)
+#define thread_end_syn(tname) pthread_mutex_unlock(tname->mutex_lock)
+#define thread_beg_syn_read(tname) pthread_rwlock_rdlock(tname->rw_lock)
+#define thread_end_syn_read(tname) pthread_rwlock_unlock(tname->rw_lock)
+#define thread_beg_syn_write(tname) pthread_rwlock_wrlock(tname->rw_lock)
+#define thread_end_syn_write(tname) pthread_rwlock_unlock(tname->rw_lock)
+#define thread_end_loop(tname)	\
+		}	\
+		if(tname->once){	\
+			pthread_mutex_lock(&tname->_COND_LOCK);	\
+			tname->state = 2;	\
+			pthread_cond_signal(&tname->_COND);	\
+			pthread_mutex_unlock(&tname->_COND_LOCK);	\
+		}	\
+	}	\
+	pthread_mutex_lock(&tname->_COND_LOCK);	\
+	tname->state = 2;	\
+	pthread_cond_signal(&tname->_COND);	\
+	pthread_mutex_unlock(&tname->_COND_LOCK)
+#define thread_end_func(tname) return NULL; }
+
+#define thread_preprocess(tname)		\
+	thread_def_shared_vars(tname);	\
+	(void)(tname##_params);	\
+	(void)(tname##_pids);	\
+	(void)(tname##_mlock);	\
+	(void)(tname##_rwlock);	\
+	(void)(tname);	\
+	tname##_i = 0;	\
+	tname##_j = 0;	\
+	tname##_var_next = 0
+
+#define thread_prepare(tname) thread_preprocess(tname)
+#define thread_begin_init(tname, n_thread) assert(n_thread > 0);\
+	tname##_params = (struct tname##_struct *)malloc(sizeof(struct tname##_struct) * n_thread);\
+	tname##_pids = (pthread_t *)malloc(sizeof(pthread_t) * n_thread); \
+	tname##_mlock = calloc(1, sizeof(pthread_mutex_t));	\
+	*tname##_mlock = (pthread_mutex_t)PTHREAD_MUTEX_INITIALIZER;	\
+	tname##_rwlock = calloc(1, sizeof(pthread_rwlock_t));	\
+	*tname##_rwlock = (pthread_rwlock_t)PTHREAD_RWLOCK_INITIALIZER;	\
+	for(tname##_i=0,tname##_j=0;tname##_i<(int)(n_thread);tname##_i++){ \
+		tname = tname##_params + tname##_i;\
+		tname->mutex_lock = tname##_mlock;\
+		tname->rw_lock = tname##_rwlock;\
+		tname->_COND_LOCK = (pthread_mutex_t)PTHREAD_MUTEX_INITIALIZER;	\
+		tname->_COND = (pthread_cond_t)PTHREAD_COND_INITIALIZER;	\
+		tname->n_cpu      = n_thread;\
+		tname->t_idx      = tname##_i;\
+		tname->running    = 1;\
+		tname->state      = 3;\
+		tname->once       = 1;\
+		tname->tname##_params = (struct tname##_struct*)tname##_params;\
+		tname->tname##_array  = (struct tname##_struct*)tname##_params
+#define thread_beg_init(tname, n_thread) thread_begin_init(tname, n_thread)
+
+#define thread_end_init(tname) \
+		if(pthread_create(tname##_pids + tname##_i, NULL, thread_##tname##_func, (void*)tname) != 0){\
+			fprintf(stderr, " -- Failed to create thread [%s, %04d] in %s -- %s:%d --\n", #tname, tname##_i, __FUNCTION__, __FILE__, __LINE__);\
+			exit(1);\
+		}\
+		while(1){	\
+			int _stop;	\
+			_stop = 0;	\
+			if(tname->state == 0) _stop = 1;	\
+			else {	\
+				struct timespec _timeout;	\
+				pthread_mutex_lock(&tname->_COND_LOCK);	\
+				clock_gettime(CLOCK_REALTIME, &_timeout);	\
+				_timeout.tv_nsec += 1000000;	\
+				pthread_cond_timedwait(&tname->_COND, &tname->_COND_LOCK, &_timeout);	\
+				if(tname->state == 0) _stop = 1;	\
+				pthread_mutex_unlock(&tname->_COND_LOCK);	\
+			}	\
+			if(_stop) break;	\
+		}	\
+	}\
+	tname = tname##_params + 0;	\
+	tname##_var_next = 0;	\
+	tname##_i = 0;	\
+	tname##_j = 0
+
+#define thread_export_core(tname, _tname, params, pids, mlock, rwlock, i, j, next)	\
+do {	\
+	_tname = tname;	\
+	params = tname##_params;	\
+	pids   = tname##_pids;	\
+	mlock  = tname##_mlock;	\
+	rwlock = tname##_rwlock;	\
+	i      = tname##_i;	\
+	j      = tname##_j;	\
+	next   = tname##_var_next;	\
+} while(0)
+
+#define thread_export(tname, obj) thread_export_core(tname, (obj)->tname, (obj)->tname##_params, (obj)->tname##_pids, (obj)->tname##_mlock, (obj)->tname##_rwlock, (obj)->tname##_i, (obj)->tname##_j, (obj)->tname##_var_next)
+
+#define thread_import_core(tname, _tname, params, pids, mlock, rwlock, i, j, next)	\
+do {	\
+	tname           = _tname;	\
+	tname##_params = params;	\
+	tname##_pids   = pids;	\
+	tname##_mlock  = mlock;	\
+	tname##_rwlock = rwlock;	\
+	tname##_i      = i;	\
+	tname##_j      = j;	\
+	tname##_var_next   = next;	\
+} while(0)
+
+#define thread_import(tname, obj) thread_import_core(tname, (obj)->tname, (obj)->tname##_params, (obj)->tname##_pids, (obj)->tname##_mlock, (obj)->tname##_rwlock, (obj)->tname##_i, (obj)->tname##_j, (obj)->tname##_var_next)
+
+#define thread_begin_operate(tname, idx) tname = tname##_params + idx
+#define thread_beg_operate(tname, idx) thread_begin_operate(tname, idx)
+#define thread_wake(tname)  do { pthread_mutex_lock(&tname->_COND_LOCK); tname->state = 1; pthread_cond_signal(&tname->_COND); pthread_mutex_unlock(&tname->_COND_LOCK); } while(0)
+#define thread_wake_all(tname)  do { thread_beg_iter(tname); thread_wake(tname); thread_end_iter(tname); } while(0);
+#define thread_waitfor_idle(tname)	\
+	while(1){	\
+		int _stop;	\
+		_stop = 0;	\
+		pthread_mutex_lock(&tname->_COND_LOCK);	\
+		if(tname->state != 1) _stop = 1;	\
+		else {	\
+			struct timespec _timeout;	\
+			clock_gettime(CLOCK_REALTIME, &_timeout);	\
+			_timeout.tv_nsec += 1000;	\
+			pthread_cond_timedwait(&tname->_COND, &tname->_COND_LOCK, &_timeout);	\
+			if(tname->state != 1) _stop = 1;	\
+		}	\
+		pthread_mutex_unlock(&tname->_COND_LOCK);	\
+		if(_stop) break;	\
+	}	\
+	tname->state = 0
+#define thread_wait(tname) thread_waitfor_idle(tname)
+#define thread_wait_next(tname) do { thread_beg_operate(tname, tname##_var_next); thread_wait(tname); tname##_var_next = (tname##_var_next + 1) % tname##_params[0].n_cpu; } while(0)
+#define thread_end_operate(tname, idx)   tname = NULL
+#define thread_begin_iter(tname) { struct tname##_struct * tname = NULL; int tname##_i; for(tname##_i=0;tname##_i<tname##_params[0].n_cpu;tname##_i++){ tname = tname##_params + tname##_i
+#define thread_beg_iter(tname) thread_begin_iter(tname)
+#define thread_is_idle(tname) (tname->state != 1)
+#define thread_n_cpus(tname) (tname->n_cpu)
+#define thread_index(tname) (tname->t_idx)
+#define thread_end_iter(tname) } }
+#define thread_access(tname, idx) (tname##_params + idx)
+
+#define thread_beg_monitor(tname, usec)	\
+while(1){	\
+	nano_sleep(usec)
+
+#define thread_end_monitor(tname)	\
+	for(tname##_j=0;tname##_j<tname##_params[0].n_cpu;tname##_j++){	\
+		if(tname##_params[tname##_j].state == 1) break;	\
+	}	\
+	if(tname##_j==tname##_params[0].n_cpu) break;	\
+}
+
+#define thread_waitfor_one_idle(tname)	\
+while(1){	\
+	for(;tname##_j<tname##_params[0].n_cpu;tname##_j++){	\
+		if(tname##_params[tname##_j].state != 1){	\
+			tname = tname##_params + tname##_j;	\
+			break;	\
+		}	\
+	}	\
+	if(tname##_j >= tname##_params[0].n_cpu){	\
+		tname##_j = 0;	\
+		nano_sleep(10);	\
+	} else {	\
+		tname##_j = (tname##_j + 1) % tname##_params[0].n_cpu;	\
+		break;	\
+	}	\
+}	\
+tname->state = 0
+#define thread_wait_one(tname) thread_waitfor_one_idle(tname)
+
+#define thread_wait_done(tname)	\
+while(1){	\
+	int _nrun_;	\
+	_nrun_ = 0;	\
+	for(tname##_j=0;tname##_j<tname##_params[0].n_cpu;tname##_j++){	\
+		if(tname##_params[tname##_j].state == 2){	\
+			tname = tname##_params + tname##_j;	\
+			break;	\
+		} else if(tname##_params[tname##_j].state == 1){	\
+			_nrun_ ++;	\
+		}	\
+	}	\
+	if(tname##_j >= tname##_params[0].n_cpu){	\
+		tname##_j = 0;	\
+		if(_nrun_ == 0){	\
+			tname = tname##_params + tname##_j;	\
+			tname##_j = (tname##_j + 1) % tname##_params[0].n_cpu;	\
+			break;	\
+		} else {	\
+			nano_sleep(10);	\
+		}	\
+	} else {	\
+		tname##_j = (tname##_j + 1) % tname##_params[0].n_cpu;	\
+		break;	\
+	}	\
+}	\
+tname->state = 0
+
+#define thread_test_all(tname, expr) ({int ret = 1; thread_beg_iter(tname); if(!(expr)){ ret = 0; break;} thread_end_iter(tname); ret;})
+
+#define thread_count_all(tname, expr) ({int ret = 0; thread_beg_iter(tname); if((expr)){ ret ++; } thread_end_iter(tname); ret;})
+
+#define thread_all_idle(tname) thread_test_all(tname, (tname)->state != 1)
+
+#define thread_waitfor_all_idle(tname) { thread_begin_iter(tname); thread_waitfor_idle(tname); thread_end_iter(tname); }
+#define thread_wait_all(tname) thread_waitfor_all_idle(tname)
+
+#define thread_apply_all(tname, expr) { thread_begin_iter(tname); (expr); thread_wake(tname); thread_end_iter(tname); thread_waitfor_all_idle(tname); }
+
+#define thread_begin_close(tname) for(tname##_i=0;tname##_i<tname##_params[0].n_cpu;tname##_i++){ \
+		tname = tname##_params + tname##_i;\
+		pthread_mutex_lock(&tname->_COND_LOCK);	\
+		tname->running = 0; \
+		pthread_cond_signal(&tname->_COND);	\
+		pthread_mutex_unlock(&tname->_COND_LOCK);	\
+		thread_wait(tname);\
+		pthread_join(tname##_pids[tname##_i], NULL)
+#define thread_beg_close(tname) thread_begin_close(tname)
+#define thread_end_close(tname) }  free((void*)tname##_params); free(tname##_pids); free(tname##_mlock); free(tname##_rwlock)
+
+#define thread_run(tname, ncpu, vars_expr, init_expr, free_expr, pre_expr, loop_expr, post_expr, invoke_expr)	\
+{	\
+	thread_beg_def(tname);	\
+	vars_expr	\
+	thread_end_def(tname);	\
+	thread_begin_func_core(tname);	\
+	pre_expr	\
+	thread_beg_loop(tname);	\
+	loop_expr	\
+	thread_end_loop(tname);	\
+	post_expr	\
+	thread_end_func(tname);	\
+	{	\
+		thread_preprocess(tname);	\
+		thread_beg_init(tname, ncpu);	\
+		init_expr	\
+		thread_end_init(tname);	\
+		invoke_expr	\
+		thread_beg_close(tname);	\
+		free_expr	\
+		thread_end_close(tname);	\
+	}	\
+}
+
+#define THREAD_EXPR(...) __VA_ARGS__
+
+#define thread_fast_run(tname, ncpu, loop_expr) thread_run(tname, ncpu, , , , THREAD_EXPR(int NCPU; int TIDX; NCPU = tname->n_cpu; TIDX = tname->t_idx;), THREAD_EXPR(loop_expr;), , THREAD_EXPR(thread_wake_all(tname); thread_wait_all(tname);))
+
+#define thread_fast_run2(tname, ncpu, expr)	\
+{	\
+	thread_beg_def(tname);	\
+	thread_end_def(tname);	\
+	thread_begin_func_core(tname);	\
+	int NCPU, TIDX;	\
+	NCPU = tname->n_cpu;	\
+	TIDX = tname->t_idx;	\
+	UNUSED(NCPU);	\
+	UNUSED(TIDX);	\
+	thread_beg_loop(tname);	\
+	(expr);	\
+	thread_end_loop(tname);	\
+	thread_end_func(tname);	\
+	{	\
+		thread_preprocess(tname);	\
+		thread_beg_init(tname, ncpu);	\
+		thread_end_init(tname);	\
+		thread_wake_all(tname);	\
+		thread_wait_all(tname);	\
+		thread_beg_close(tname);	\
+		thread_end_close(tname);	\
+	}	\
+}
+
+#endif

tripoa.h

@@ -0,0 +1,565 @@
+/*
+ * 
+ * Copyright (c) 2018, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef TRI_PO_MSA_CNS_RJ_H
+#define TRI_PO_MSA_CNS_RJ_H
+
+#include "poacns.h"
+#include "ksw.h"
+
+typedef struct {
+	SeqBank *seqs;
+	u2v *rbegs, *rends;
+	u4i longest_idx, seqmax;
+	int8_t matrix[16];
+	int fail_skip;
+	u1i ksize; // 11
+	float kdup; // 0.1
+	float keqs; // 0.2
+	uuhash *khash;
+	u1v *qry, *ref;
+	u2i winlen, winmin;
+	u2v *regs[2];
+	POG *pogs[3];
+	u4v *kidxs;
+	f4v *kords;
+	BaseBank *cns;
+	int is_tripog, refmode;
+	int shuffle; // 0: no shuffling, 1: by kmer, 2, random
+} TriPOG;
+
+//static inline TriPOG* init_tripog(u4i seqmax, int refmode, int winlen, int winmin, int fail_skip, int M, int X, int I, int D, int W, int use_sse, int rW, u4i min_cnt, float min_freq){
+static inline TriPOG* init_tripog(u4i seqmax, int shuffle, int winlen, int winmin, int fail_skip, POGPar *par){
+	TriPOG *tp;
+	u4i i;
+	tp = malloc(sizeof(TriPOG));
+	tp->seqs = init_seqbank();
+	tp->rbegs = init_u2v(32);
+	tp->rends = init_u2v(32);
+	tp->seqmax = seqmax;
+	tp->longest_idx = 0;
+	tp->ksize = 11;
+	tp->kdup = 0.1;
+	tp->keqs = 0.2;
+	tp->khash = init_uuhash(13);
+	tp->ref  = init_u1v(winlen);
+	tp->qry  = init_u1v(1024);
+	tp->regs[0] = init_u2v(32);
+	tp->regs[1] = init_u2v(32);
+	tp->refmode = par->refmode;
+	tp->fail_skip = fail_skip;
+	tp->winlen = winlen;
+	tp->winmin = winmin;
+	tp->pogs[0] = init_pog(*par);
+#if 0
+	if(winlen > 0){
+		tp->winlen = winlen;
+		tp->winmin = winmin;
+		tp->pogs[0] = init_pog(refmode, M, X, I, D, 0, 0, use_sse, 0, min_cnt, min_freq);
+	} else {
+		tp->winlen = 0;
+		tp->winmin = 0;
+		tp->pogs[0] = init_pog(refmode, M, X, I, D, W, - winlen, use_sse, rW, min_cnt, min_freq);
+	}
+#endif
+	tp->pogs[1] = init_pog(*par);
+	tp->pogs[1]->par->W_score = 0;
+	tp->pogs[2] = init_pog(*par);
+	tp->pogs[2]->par->W_score = 0;
+	//tp->pogs[1] = init_pog(refmode, M, X, I, D, W, 0, use_sse, rW, min_cnt, min_freq);
+	//tp->pogs[2] = init_pog(refmode, M, X, I, D, W, 0, use_sse, rW, min_cnt, min_freq);
+	//tp->pogs[1]->near_dialog = 1;
+	//tp->pogs[2]->near_dialog = 1;
+	tp->kidxs = init_u4v(32);
+	tp->kords = init_f4v(32);
+	tp->cns = init_basebank();
+	for(i=0;i<16;i++){
+		tp->matrix[i] = ((i / 4) == (i % 4))? par->M : par->X;
+	}
+	tp->is_tripog = 0;
+	tp->shuffle = shuffle;
+	return tp;
+}
+
+static inline void free_tripog(TriPOG *tp){
+	free_seqbank(tp->seqs);
+	free_u2v(tp->rbegs);
+	free_u2v(tp->rends);
+	free_uuhash(tp->khash);
+	free_u1v(tp->qry);
+	free_u1v(tp->ref);
+	free_u2v(tp->regs[0]);
+	free_u2v(tp->regs[1]);
+	free_pog(tp->pogs[0]);
+	free_pog(tp->pogs[1]);
+	free_pog(tp->pogs[2]);
+	free_u4v(tp->kidxs);
+	free_f4v(tp->kords);
+	free_basebank(tp->cns);
+	free(tp);
+}
+
+static inline void beg_tripog(TriPOG *tp){
+	clear_seqbank(tp->seqs);
+	clear_u2v(tp->rbegs);
+	clear_u2v(tp->rends);
+	tp->longest_idx = 0;
+	tp->is_tripog = 0;
+}
+
+static inline void push_tripog(TriPOG *tp, char *seq, u4i len, u2i refbeg, u2i refend){
+	if(!tp->shuffle && tp->seqs->nseq >= tp->seqmax) return;
+	push_seqbank(tp->seqs, NULL, 0, seq, len);
+	push_u2v(tp->rbegs, refbeg);
+	push_u2v(tp->rends, refend);
+	if(tp->seqs->rdlens->buffer[tp->longest_idx] < len){
+		tp->longest_idx = tp->seqs->nseq - 1;
+	}
+}
+
+static inline void fwdbitpush_tripog(TriPOG *tp, u8i *bits, u8i off, u4i len, u2i refbeg, u2i refend){
+	if(!tp->shuffle && tp->seqs->nseq >= tp->seqmax) return;
+	fwdbitpush_seqbank(tp->seqs, NULL, 0, bits, off, len);
+	push_u2v(tp->rbegs, refbeg);
+	push_u2v(tp->rends, refend);
+	if(tp->seqs->rdlens->buffer[tp->longest_idx] < len){
+		tp->longest_idx = tp->seqs->nseq - 1;
+	}
+}
+
+static inline void revbitpush_tripog(TriPOG *tp, u8i *bits, u8i off, u4i len, u2i refbeg, u2i refend){
+	if(!tp->shuffle && tp->seqs->nseq >= tp->seqmax) return;
+	revbitpush_seqbank(tp->seqs, NULL, 0, bits, off, len);
+	push_u2v(tp->rbegs, refbeg);
+	push_u2v(tp->rends, refend);
+	if(tp->seqs->rdlens->buffer[tp->longest_idx] < len){
+		tp->longest_idx = tp->seqs->nseq - 1;
+	}
+}
+
+static inline void direct_run_tripog(TriPOG *tp){
+	POG *g;
+	u4i ridx;
+	clear_basebank(tp->cns);
+	if(tp->refmode){
+	} else {
+		tp->pogs[0]->par->W = tp->pogs[1]->par->W;
+		tp->pogs[0]->par->rW = tp->pogs[1]->par->rW;
+		tp->pogs[0]->par->W_score = tp->pogs[1]->par->W * tp->pogs[1]->par->M * 8;
+	}
+	g = tp->pogs[0];
+	beg_pog(g);
+	for(ridx=0;ridx<tp->seqs->nseq;ridx++){
+		fwdbitpush_pog_core(g, tp->seqs->rdseqs->bits, tp->seqs->rdoffs->buffer[ridx], tp->seqs->rdlens->buffer[ridx], tp->rbegs->buffer[ridx], tp->rends->buffer[ridx]);
+	}
+	end_pog(g);
+	if(g->cns->size == 0){
+		fast_fwdbits2basebank(tp->cns, tp->seqs->rdseqs->bits, tp->seqs->rdoffs->buffer[tp->longest_idx], tp->seqs->rdlens->buffer[tp->longest_idx]);
+	} else {
+		fast_fwdbits2basebank(tp->cns, g->cns->bits, 0, g->cns->size);
+	}
+	if(tp->refmode){
+	} else {
+		tp->pogs[0]->par->W = 0;
+		tp->pogs[0]->par->rW = 0;
+		tp->pogs[0]->par->W_score = 0;
+	}
+}
+
+static inline void shuffle_reads_by_kmers_tripog(TriPOG *tp){
+	SeqBank *sb;
+	uuhash *khash;
+	u4v *kidxs;
+	f4v *kords;
+	u2v *rbegs, *rends;
+	uuhash_t *u;
+	u8i roff;
+	u4i ridx, i, ksize, kmer, kmask, rlen, khit, mincnt;
+	double logv;
+	int exists;
+	sb = tp->seqs;
+	if(sb->nseq == 0) return;
+	khash = tp->khash;
+	kidxs = tp->kidxs;
+	kords = tp->kords;
+	rbegs = tp->rbegs;
+	rends = tp->rends;
+	ksize = tp->ksize;
+	clear_u4v(kidxs);
+	clear_f4v(kords);
+	clear_uuhash(khash);
+	kmask = MAX_U4 >> ((16 - ksize) << 1);
+	mincnt = tp->refmode? 1 : 2;
+	for(ridx=0;ridx<sb->nseq;ridx++){
+		rlen = sb->rdlens->buffer[ridx];
+		kmer = 0;
+		roff = sb->rdoffs->buffer[ridx];
+		for(i=0;i<rlen;i++){
+			kmer = ((kmer << 2) | get_basebank(sb->rdseqs, roff + i)) & kmask;
+			if(i + 1 < ksize) continue;
+			u = prepare_uuhash(khash, kmer, &exists);
+			if(exists){
+				if(((u->val >> 16) & 0x7FFFU) == ridx + 1){
+					u->val |= 1U << 31;
+				} else {
+					u->val = (u->val & 0x8000FFFFU) | ((ridx + 1) << 16);
+				}
+				u->val ++;
+			} else {
+				u->key = kmer;
+				u->val = (0U << 31) | ((ridx + 1) << 16) | 1;
+			}
+		}
+		if(tp->refmode) break;
+	}
+	logv = log(1.2);
+	for(ridx=0;ridx<sb->nseq;ridx++){
+		rlen = sb->rdlens->buffer[ridx];
+		kmer = 0;
+		roff = sb->rdoffs->buffer[ridx];
+		khit = 0;
+		for(i=0;i<rlen;i++){
+			kmer = ((kmer << 2) | get_basebank(sb->rdseqs, roff + i)) & kmask;
+			if(i + 1 < ksize) continue;
+			u = get_uuhash(khash, kmer);
+			if(u && (u->val & 0x80000000U) == 0 && (u->val & 0xFFFFU) >= mincnt){
+				khit ++;
+			}
+		}
+		if(tp->refmode){
+			if(ridx == 0){
+				push_f4v(kords, 3e+38F);
+			} else {
+				push_f4v(kords, ((double)khit) * logv / log(num_max(rlen, sb->rdlens->buffer[0])));
+			}
+		} else {
+			push_f4v(kords, ((double)khit) * logv / log(rlen));
+		}
+		push_u4v(kidxs, ridx);
+	}
+	sort_array(kidxs->buffer, kidxs->size, u4i, num_cmpgt(kords->buffer[b], kords->buffer[a]));
+	if(cns_debug > 1){
+		for(i=0;i<kidxs->size;i++){
+			fprintf(stderr, "SHUFFLE[%u] %u\t%u\t%0.4f\n", i, kidxs->buffer[i], sb->rdlens->buffer[kidxs->buffer[i]], kords->buffer[kidxs->buffer[i]]);
+		}
+	}
+	for(i=0;i<kidxs->size;i++){
+		push_u8v(sb->rdoffs, sb->rdoffs->buffer[kidxs->buffer[i]]);
+		push_u4v(sb->rdlens, sb->rdlens->buffer[kidxs->buffer[i]]);
+		push_u2v(rbegs, rbegs->buffer[kidxs->buffer[i]]);
+		push_u2v(rends, rends->buffer[kidxs->buffer[i]]);
+	}
+	remove_array_u8v(sb->rdoffs, 0, kidxs->size);
+	remove_array_u4v(sb->rdlens, 0, kidxs->size);
+	remove_array_u2v(rbegs, 0, kidxs->size);
+	remove_array_u2v(rends, 0, kidxs->size);
+	if(tp->seqmax && sb->nseq > tp->seqmax){
+		if(cns_debug > 1){
+			fprintf(stderr, "SEQMAX: %u -> %u\n", sb->nseq, tp->seqmax);
+		}
+		sb->nseq = tp->seqmax;
+		sb->rdoffs->size = tp->seqmax;
+		sb->rdlens->size = tp->seqmax;
+		rbegs->size = tp->seqmax;
+		rends->size = tp->seqmax;
+	}
+}
+
+static inline void subsample_reads_tripog(TriPOG *tp){
+	SeqBank *sb;
+	u4v *kidxs;
+	u2v *rbegs, *rends;
+	f4v *kords;
+	u4i ridx, i;
+	f4i cutoff;
+	sb = tp->seqs;
+	if(sb->nseq == 0 || sb->nseq <= tp->seqmax) return;
+	kidxs = tp->kidxs;
+	kords = tp->kords;
+	rbegs = tp->rbegs;
+	rends = tp->rends;
+	clear_u4v(kidxs);
+	clear_f4v(kords);
+	if(tp->refmode){
+		push_u4v(kidxs, 0);
+		push_f4v(kords, 200.0);
+		ridx = 1;
+	} else {
+		ridx = 0;
+	}
+	for(;ridx<sb->nseq;ridx++){
+		push_u4v(kidxs, ridx);
+		push_f4v(kords, 100.0 * drand48());
+	}
+	sort_array(kidxs->buffer, kidxs->size, u4i, num_cmpgt(kords->buffer[b], kords->buffer[a]));
+	cutoff = kords->buffer[kidxs->buffer[tp->seqmax - 1]];
+	clear_u4v(kidxs);
+	for(i=0;i<kords->size;i++){
+		if(kords->buffer[i] >= cutoff){
+			push_u4v(kidxs, i);
+		}
+	}
+	if(cns_debug > 1){
+		fprintf(stderr, "RANDOM CUTOFF = %f\n", cutoff);
+		for(i=0;i<kidxs->size;i++){
+			fprintf(stderr, "SHUFFLE[%u] %u\t%0.4f\n", i, kidxs->buffer[i], kords->buffer[kidxs->buffer[i]]);
+		}
+	}
+	for(i=0;i<kidxs->size;i++){
+		push_u8v(sb->rdoffs, sb->rdoffs->buffer[kidxs->buffer[i]]);
+		push_u4v(sb->rdlens, sb->rdlens->buffer[kidxs->buffer[i]]);
+		push_u2v(rbegs, rbegs->buffer[kidxs->buffer[i]]);
+		push_u2v(rends, rends->buffer[kidxs->buffer[i]]);
+	}
+	remove_array_u8v(sb->rdoffs, 0, kords->size);
+	remove_array_u4v(sb->rdlens, 0, kords->size);
+	remove_array_u2v(rbegs, 0, kords->size);
+	remove_array_u2v(rends, 0, kords->size);
+	if(cns_debug > 1){
+		fprintf(stderr, "SEQMAX: %u -> %u\n", sb->nseq, tp->seqmax);
+	}
+	sb->nseq = tp->seqmax;
+}
+
+static inline void end_tripog(TriPOG *tp){
+	POG *g;
+	kswr_t R;
+	u4i ridx, rlen, b, e, failed;
+	switch(tp->shuffle){
+		case 1: shuffle_reads_by_kmers_tripog(tp); break;
+		case 2: subsample_reads_tripog(tp); break;
+	}
+	if(tp->winlen == 0){
+		return direct_run_tripog(tp);
+	}
+	tp->longest_idx = 0;
+	for(ridx=1;ridx<tp->seqs->nseq;ridx++){
+		if(tp->seqs->rdlens->buffer[ridx] > tp->seqs->rdlens->buffer[tp->longest_idx]){
+			tp->longest_idx = ridx;
+		}
+	}
+	if(tp->seqs->nseq < 2){
+		return direct_run_tripog(tp);
+	}
+	rlen = tp->seqs->rdlens->buffer[0];
+	if(rlen < 2 * tp->winlen){
+		return direct_run_tripog(tp);
+	}
+	clear_u2v(tp->regs[0]);
+	clear_u2v(tp->regs[1]);
+	// selecting unique window
+	{
+		uuhash_t *u;
+		u4i i, j, hit, nb, kmer, kcnt, kdup, keqs, ktot, kmask, roff;
+		reset_iter_uuhash(tp->khash);
+		while((u = ref_iter_uuhash(tp->khash))){
+			u->val &= 0xFFFFU;
+		}
+		kmask = MAX_U4 >> ((16 - tp->ksize) << 1);
+		for(ridx=0;ridx<1;ridx++){
+			rlen = tp->seqs->rdlens->buffer[ridx];
+			kmer = 0;
+			roff = tp->seqs->rdoffs->buffer[ridx];
+			for(i=0;i<rlen;i++){
+				kmer = ((kmer << 2) | get_basebank(tp->seqs->rdseqs, roff + i)) & kmask;
+				if(i + 1 < tp->ksize) continue;
+				u = get_uuhash(tp->khash, kmer);
+				u->val += 1U << 16;
+			}
+		}
+		nb = ((rlen - tp->winlen) * 2 / 3 / tp->winlen / 2) * 2;
+		b = (rlen - tp->winlen) / 2;
+		hit = 0;
+		roff = tp->seqs->rdoffs->buffer[0];
+		for(j=0;j<nb;){
+			e = b + tp->winlen;
+			kmer = 0;
+			kdup = 0;
+			keqs = 0;
+			ktot = e - b + 1 - tp->ksize;
+			for(i=b;i<e;i++){
+				kmer = ((kmer << 2) | get_basebank(tp->seqs->rdseqs, roff + i)) & kmask;
+				if(i + 1 - b < tp->ksize) continue;
+				kcnt = getval_uuhash(tp->khash, kmer);
+				if((kcnt >> 16) > 1){
+					kdup ++;
+				} else if((kcnt & 0xFFFFU) > 1){
+					keqs ++;
+				}
+			}
+			if(cns_debug > 1){
+				fprintf(stderr, "Selecting anchor[%4d,%4d]: ktot=%d keqs=%d kdup=%d\n", b, e, ktot, keqs, kdup);
+			}
+			if(kdup < UInt(tp->kdup * ktot) && keqs >= UInt(tp->keqs * ktot)){
+				hit = 1;
+				break;
+			}
+			j ++;
+			if(j & 0x01){
+				b = (rlen - tp->winlen) / 2 + tp->winlen * ((j + 1) >> 1);
+			} else {
+				b = (rlen - tp->winlen) / 2 - tp->winlen * ((j + 1) >> 1);
+			}
+		}
+		if(hit == 0){
+			return direct_run_tripog(tp);
+		}
+	}
+	push_u2v(tp->regs[0], b);
+	push_u2v(tp->regs[1], e);
+	failed = 0;
+	for(ridx=1;ridx<tp->seqs->nseq;ridx++){
+		clear_and_encap_u1v(tp->qry, tp->winlen);
+		bitseq_basebank(tp->seqs->rdseqs, tp->seqs->rdoffs->buffer[0] + b, tp->winlen, tp->qry->buffer);
+		rlen = tp->seqs->rdlens->buffer[ridx];
+		clear_and_encap_u1v(tp->ref, rlen);
+		bitseq_basebank(tp->seqs->rdseqs, tp->seqs->rdoffs->buffer[ridx], rlen, tp->ref->buffer);
+		R = ksw_align(tp->winlen, tp->qry->buffer, rlen, tp->ref->buffer, 4, tp->matrix, - (tp->pogs[0]->par->I + tp->pogs[0]->par->D)/2, 1, KSW_XSTART, NULL);
+		if(R.qb <= -1 || R.tb <= -1 || R.qe <= -1 || R.te <= -1){
+			if(cns_debug > 1){
+				fprintf(stderr, "FAILED_ALIGN: READ%u [%d,%d=%d][%d,%d=%d]\n", ridx, R.qb, R.qe, R.qe - R.qb, R.tb, R.te, R.te - R.tb);
+			}
+			if(tp->fail_skip){
+				push_u2v(tp->regs[0], MAX_U2);
+				push_u2v(tp->regs[1], MAX_U2);
+				failed ++;
+				continue;
+			} else {
+				return direct_run_tripog(tp);
+			}
+		}
+		if(R.qe + 1 - R.qb < tp->winmin || R.te + 1 - R.tb < tp->winmin){
+			if(cns_debug > 1){
+				fprintf(stderr, "FAILED_ALIGN: READ%u [%d,%d=%d][%d,%d=%d]\n", ridx, R.qb, R.qe, R.qe - R.qb, R.tb, R.te, R.te - R.tb);
+			}
+			if(tp->fail_skip){
+				push_u2v(tp->regs[0], MAX_U2);
+				push_u2v(tp->regs[1], MAX_U2);
+				failed ++;
+				continue;
+			} else {
+				return direct_run_tripog(tp);
+			}
+		}
+		push_u2v(tp->regs[0], R.tb);
+		push_u2v(tp->regs[1], R.te + 1);
+	}
+	if(failed * 2 >= tp->seqs->nseq){
+		return direct_run_tripog(tp);
+	}
+	// building cns for fast aligned regions
+	g = tp->pogs[0];
+	g->par->alnmode = POG_ALNMODE_GLOBAL;
+	beg_pog(g);
+	for(ridx=0;ridx<tp->seqs->nseq;ridx++){
+		if(tp->regs[0]->buffer[ridx] == MAX_U2) continue;
+		fwdbitpush_pog(g, tp->seqs->rdseqs->bits, tp->seqs->rdoffs->buffer[ridx] + tp->regs[0]->buffer[ridx], tp->regs[1]->buffer[ridx] - tp->regs[0]->buffer[ridx]);
+	}
+	if(0){
+		print_seqs_pog(g, "p0.fa", NULL);
+	}
+	end_pog(g);
+	g->par->alnmode = POG_ALNMODE_OVERLAP;
+	// finding a best break point
+	{
+		u2v *rs;
+		u1i *s;
+		u2i i, idx, bst, bsi, cnt;
+		rs = init_u2v(256);
+		bst = 0;
+		bsi = MAX_U2;
+		s = g->msa->buffer + g->msa_len * g->seqs->nseq;
+		for(idx=0;idx<g->msa_len;idx++){
+			if(s[idx] == 4) continue;
+			cnt = 0;
+			for(ridx=0;ridx<g->seqs->nseq;ridx++){
+				if(g->msa->buffer[g->msa_len * ridx + idx] == s[idx]) cnt ++;
+			}
+			if(cnt > bst){
+				bst = cnt;
+				bsi = idx;
+				clear_u2v(rs);
+				push_u2v(rs, idx);
+			} else if(cnt == bst){
+				push_u2v(rs, idx);
+			}
+		}
+		if(rs->size){
+			bsi = rs->buffer[rs->size / 2];
+		}
+		free_u2v(rs);
+		if(bsi == MAX_U2){
+			return direct_run_tripog(tp);
+		}
+		if(cns_debug > 1){
+			fprintf(stderr, "BREAKPOINT[MSA]: %u/%u\n", bsi, g->msa_len);
+		}
+		// transform coordinate
+		i = 0;
+		for(ridx=0;ridx<tp->seqs->nseq;ridx++){
+			if(tp->regs[0]->buffer[ridx] == MAX_U2) continue;
+			cnt = 0;
+			s = g->msa->buffer + g->msa_len * i;
+			for(idx=0;idx<bsi;idx++){
+				if(s[idx] != 4) cnt ++;
+			}
+			tp->regs[0]->buffer[ridx] = tp->regs[0]->buffer[ridx] + cnt;
+			tp->regs[1]->buffer[ridx] = tp->regs[0]->buffer[ridx];
+			if(cns_debug > 1){
+				fprintf(stderr, "BREAKPOINT[%u:%u]: %u/%u\n", i, ridx, tp->regs[0]->buffer[ridx], tp->seqs->rdlens->buffer[ridx]);
+			}
+			i ++;
+		}
+	}
+	// forward cns to seqs' ends
+	g = tp->pogs[2];
+	beg_pog(g);
+	for(ridx=0;ridx<tp->seqs->nseq;ridx++){
+		if(tp->regs[0]->buffer[ridx] == MAX_U2) continue;
+		fwdbitpush_pog(g, tp->seqs->rdseqs->bits, tp->seqs->rdoffs->buffer[ridx] + tp->regs[1]->buffer[ridx], tp->seqs->rdlens->buffer[ridx] - tp->regs[1]->buffer[ridx]);
+	}
+	if(0){
+		print_seqs_pog(g, "p2.fa", NULL);
+	}
+	end_pog(g);
+	if(g->cns->size == 0){
+		return direct_run_tripog(tp);
+	}
+	// backward cns to seqs' begs
+	g = tp->pogs[1];
+	beg_pog(g);
+	for(ridx=0;ridx<tp->seqs->nseq;ridx++){
+		if(tp->regs[0]->buffer[ridx] == MAX_U2) continue;
+		revbitpush_pog(g, tp->seqs->rdseqs->bits, tp->seqs->rdoffs->buffer[ridx], tp->regs[0]->buffer[ridx]);
+	}
+	if(0){
+		print_seqs_pog(g, "p1.fa", NULL);
+	}
+	end_pog(g);
+	if(g->cns->size == 0){
+		return direct_run_tripog(tp);
+	}
+	// merge two parts
+	clear_basebank(tp->cns);
+	fast_revbits2basebank(tp->cns, tp->pogs[1]->cns->bits, 0, tp->pogs[1]->cns->size);
+	fast_fwdbits2basebank(tp->cns, tp->pogs[2]->cns->bits, 0, tp->pogs[2]->cns->size);
+	tp->is_tripog = 1;
+}
+
+#endif

txtplot.h

@@ -0,0 +1,87 @@
+/*
+ *
+ * Copyright (c) 2018, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __TEXT_PLOT_RJ_H
+#define __TEXT_PLOT_RJ_H
+
+#include "list.h"
+
+static inline char* barplot_txt_u4_simple(u4i w, u4i h, u4i *vals, u4i size, u4i max_val){
+	char *g;
+	double hdiv;
+	u4i wdiv;
+	u4i i, j, x, y;
+	g = malloc((w + 1) * h + 1);
+	memset(g, ' ', (w + 1) * h);
+	for(i=1;i<=h;i++){
+		g[(w + 1) * i - 1] = '\n';
+	}
+	g[(w + 1) * h] = 0;
+	if(max_val == 0){
+		max_val = 1;
+		for(i=0;i<size;i++){
+			if(vals[i] > max_val) max_val = vals[i];
+		}
+	}
+	hdiv = 1.0 * max_val / h;
+	wdiv = size / w;
+	if(wdiv == 0) wdiv = 1;
+	for(i=x=0;i<size;i+=wdiv,x++){
+		y = UInt(vals[i] / hdiv);
+		if(y == 0 && vals[i]) y = 1;
+		if(y > h) y = h;
+		for(j=1;j<=y;j++){
+			g[(h - j) * (w + 1) + x] = '|';
+		}
+	}
+	return g;
+}
+
+static inline char* barplot_txt_u8_simple(u4i w, u4i h, u8i *vals, u4i size, u8i max_val){
+	char *g;
+	double hdiv;
+	u8i wdiv;
+	u4i i, j, x, y;
+	g = malloc((w + 1) * h + 1);
+	memset(g, ' ', (w + 1) * h);
+	for(i=1;i<=h;i++){
+		g[(w + 1) * i - 1] = '\n';
+	}
+	g[(w + 1) * h] = 0;
+	if(max_val == 0){
+		max_val = 1;
+		for(i=0;i<size;i++){
+			if(vals[i] > max_val) max_val = vals[i];
+		}
+	}
+	hdiv = 1.0 * max_val / h;
+	wdiv = size / w;
+	if(wdiv == 0) wdiv = 1;
+	for(i=x=0;i<size;i+=wdiv,x++){
+		y = UInt(vals[i] / hdiv);
+		if(y == 0 && vals[i]) y = 1;
+		if(y > h) y = h;
+		for(j=1;j<=y;j++){
+			g[(h - j) * (w + 1) + x] = '|';
+		}
+	}
+	return g;
+}
+
+#endif

wtdbg-cns.c

@@ -0,0 +1,523 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "kswx.h"
+#include "dbgcns.h"
+#include "dagcns.h"
+#include "filereader.h"
+
+static int cns_debug = 0;
+
+#define MCNS_TASK_DBGCNS	1
+#define MCNS_TASK_DAGCNS	2
+#define MCNS_TASK_OVERLAP	3
+
+thread_beg_def(mcns);
+uint32_t eid;
+CNS *cns;
+u1v *seq1, *seq2;
+u4i node1, node2;
+int reglen;
+int W, M, X, I, D, E;
+int candidate_mode;
+int corr_struct;
+f4i pM, pX, pI, pD;
+kswx_t ret;
+u32list *cigars;
+int task;
+thread_end_def(mcns);
+
+thread_beg_func(mcns);
+CNS *cns;
+DAGCNS *dag;
+GEGraph *g;
+bdpnodev *bnodes;
+bdpedgev *bedges;
+bdplinkv *linkstack;
+u8list *mem_cache;
+u1v    *mem_buffer;
+u32list *cigars[2];
+kswx_t *xs[2];
+u4i i, nbeg;
+blk_t *blk;
+int qb, qe, tb, te;
+mem_cache = init_u8list(1024);
+cigars[0] = mcns->cigars;
+cigars[1] = NULL;
+xs[0] = malloc(sizeof(kswx_t));
+xs[1] = NULL;
+cns = mcns->cns;
+dag = init_dagcns(mcns->W, mcns->M, mcns->X, mcns->I, mcns->D, mcns->E, mcns->pM, mcns->pX, mcns->pI, mcns->pD);
+g = init_gegraph();
+bnodes = init_bdpnodev(32);
+bedges = init_bdpedgev(32);
+linkstack = init_bdplinkv(32);
+mem_buffer = init_u1v(1024);
+thread_beg_loop(mcns);
+if(mcns->task == MCNS_TASK_DBGCNS){
+	ready_cns(cns);
+	run_cns(cns, mcns->candidate_mode, mcns->corr_struct);
+} else if(mcns->task == MCNS_TASK_DAGCNS){
+	// force mcns->candidate_mode to 2
+	// use data from DBGCNS struct
+	ready_cns(cns);
+	clear_u8list(dag->cns);
+	blk = ref_blkv(cns->qblks, 0);
+	append_array_u8list(dag->cns, cns->qseqs->buffer + blk->off, blk->len);
+	gen_pregraph_dagcns(dag);
+	for(i=1;i<cns->qblks->size;i++){
+		blk = ref_blkv(mcns->cns->qblks, i);
+		nbeg = branched_dynamic_programming_alignment(dag, mcns->cns->qseqs->buffer + blk->off, blk->len, g, bnodes, bedges, mem_buffer);
+		if(nbeg == 0){
+			fprintf(stderr, " -- something wrong in %s -- %s:%d --\n", __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			continue;
+		}
+		bdpgraph2dagcns(dag, g, bnodes, bedges, nbeg, linkstack);
+	}
+	merge_nodes_dagcns(dag);
+	gen_consensus_dagcns(dag, NULL);
+	if(cns_debug){
+		fprintf(stderr, "DAG%d\t%d\t", mcns->eid, (int)dag->cns->size);
+		print_seq_dagcns(dag, stderr);
+		fprintf(stderr, "\n");
+	}
+	clear_u1v(mcns->cns->cns);
+	append_u1v(mcns->cns->cns, (u1v*)dag->cns);
+	clear_string(mcns->cns->seq);
+	for(i=0;i<dag->cns->size;i++) add_char_string(mcns->cns->seq, bit_base_table[dag->cns->buffer[i]]);
+} else if(mcns->task == MCNS_TASK_OVERLAP){
+	if(mcns->seq2->size == 0){
+		mcns->ret = KSWX_NULL;
+		continue;
+	}
+	qb = 0; qe = mcns->seq1->size;
+	tb = 0; te = mcns->seq2->size;
+	if(qe > mcns->reglen) qb = qe - mcns->reglen;
+	if(te > mcns->reglen) te = mcns->reglen;
+	kswx_overlap_align_core(xs, cigars, qe - qb, mcns->seq1->buffer + qb, te - tb, mcns->seq2->buffer + tb, 1, mcns->M, mcns->X, mcns->I, mcns->D, mcns->E, mem_cache);
+	xs[0]->qb += qb;
+	xs[0]->qe += qb;
+	xs[0]->tb += tb;
+	xs[0]->te += tb;
+	mcns->ret = *xs[0];
+}
+thread_end_loop(mcns);
+free(xs[0]);
+free_u8list(mem_cache);
+free_u1v(mem_buffer);
+free_dagcns(dag);
+free_gegraph(g);
+free_bdpnodev(bnodes);
+free_bdpedgev(bedges);
+free_bdplinkv(linkstack);
+thread_end_func(mcns);
+
+int revise_joint_point(u32list *cigars, int *qe, int *te){
+	u4i i, op, ln, max;
+	int qq, q, tt, t;
+	q = t = 0;
+	qq = tt = 0;
+	max = 0;
+	for(i=1;i<=cigars->size;i++){
+		op = cigars->buffer[cigars->size - i] & 0xF;
+		ln = cigars->buffer[cigars->size - i] >> 4;
+		if(op == 0){
+			if(ln > max){
+				qq = q; tt = t;
+				max = ln;
+			}
+			q += ln;
+			t += ln;
+		} else if(op == 1){
+			q += ln;
+		} else {
+			t += ln;
+		}
+	}
+	if(cns_debug){
+		fprintf(stderr, "qe = %d -> %d\n", *qe, (*qe) - qq);
+		fprintf(stderr, "te = %d -> %d\n", *te, (*te) - tt);
+		fflush(stderr);
+	}
+	*qe -= qq;
+	*te -= tt;
+	return 1;
+}
+
+int revise_joint_point2(u32list *cigars, int *qe, int *te, int overhang){
+	u4i i, op, ln;
+	int qq, q, tt, t;
+	q = t = 0;
+	qq = tt = 0;
+	for(i=1;i<=cigars->size;i++){
+		op = cigars->buffer[cigars->size - i] & 0xF;
+		ln = cigars->buffer[cigars->size - i] >> 4;
+		switch(op){
+			case 1: q += ln; break;
+			case 2: t += ln; break;
+			default: qq = q; tt = t; q += ln; t += ln;
+		}
+		if(q >= overhang && t >= overhang){
+			if(cns_debug){
+				fprintf(stderr, "qe = %d -> %d\n", *qe, (*qe) - qq);
+				fprintf(stderr, "te = %d -> %d\n", *te, (*te) - tt);
+				fflush(stderr);
+			}
+			*qe -= qq;
+			*te -= tt;
+			return 1;
+		}
+	}
+	return 0;
+}
+
+uint32_t run(int reglen, int ksize, int Z, int W, int M, int X, int I, int D, int E, int H, int L, int XX, int OO, int EE, int cns_model, f4i pM, f4i pX, f4i pI, f4i pD, int candidate_mode, int corr_struct, uint32_t ncpu, FileReader *fr, FILE *out){
+	String *tag, *seq;
+	u1v *cseqs;
+	u4v *cxs, *cys, *tes, *qes;
+	uint32_t i, m, eid, beg, end;
+	int c, j, sl, b, e;
+	char *ss;
+	thread_preprocess(mcns);
+	tag = init_string(32);
+	seq = init_string(32);
+	cseqs = init_u1v(32);
+	cxs = init_u4v(32);
+	cys = init_u4v(32);
+	tes = init_u4v(32);
+	qes = init_u4v(32);
+	thread_beg_init(mcns, ncpu);
+	mcns->eid = 0;
+	mcns->cns = init_cns(ksize, Z, W, 0, X, I, D, E, H, L);
+	mcns->seq1 = init_u1v(32);
+	mcns->seq2 = init_u1v(32);
+	mcns->node1 = 0;
+	mcns->node2 = 0;
+	mcns->reglen = reglen;
+	mcns->W = 128;
+	mcns->M = M;
+	mcns->X = XX;
+	mcns->I = OO;
+	mcns->D = OO;
+	mcns->E = EE;
+	mcns->pM = pM;
+	mcns->pX = pX;
+	mcns->pI = pI;
+	mcns->pD = pD;
+	mcns->candidate_mode = candidate_mode;
+	mcns->corr_struct = corr_struct;
+	mcns->ret = KSWX_NULL;
+	mcns->cigars = init_u32list(16);
+	mcns->task = 0;
+	thread_end_init(mcns);
+	eid = 0;
+	thread_wait_one(mcns);
+	while(1){
+		c = readtable_filereader(fr);
+		if(c == -1 || fr->line->string[0] == 'E' || fr->line->string[0] == '>'){
+			thread_wake(mcns);
+			thread_wait_one(mcns);
+			if(mcns->task == cns_model && mcns->cns->seq->size){
+				if(cns_debug){
+					fprintf(stderr, "%s_%d_N%u_N%u\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%s\n", tag->string, mcns->eid, mcns->node1, mcns->node2, mcns->cns->qlen, mcns->cns->seq->size, mcns->cns->max_score, mcns->cns->alns[0], mcns->cns->alns[1], mcns->cns->alns[2], mcns->cns->alns[3], mcns->cns->seq->string);
+				}
+				cxs->buffer[mcns->eid] = cseqs->size;
+				for(j=0;j<mcns->cns->seq->size;j++) push_u1v(cseqs, base_bit_table[(int)mcns->cns->seq->string[j]]);
+				cys->buffer[mcns->eid] = cseqs->size;
+			}
+			reset_cns(mcns->cns);
+			clear_string(mcns->cns->seq);
+			mcns->task = cns_model;
+			mcns->eid = eid ++;
+			push_u4v(cxs, 0);
+			push_u4v(cys, 0);
+			if(fr->line->string[0] == 'E'){
+				mcns->node1 = atoll(get_col_str(fr, 2) + 1);
+				mcns->node2 = atoll(get_col_str(fr, 4) + 1);
+				continue;
+			}
+			if(tag->size){
+				thread_beg_iter(mcns);
+				thread_wait(mcns);
+				if(mcns->task == cns_model && mcns->cns->seq->size){
+					if(cns_debug){
+						//fprintf(stderr, "%s_%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%s\n", tag->string, mcns->eid, mcns->cns->qlen, mcns->cns->seq->size, mcns->cns->max_score, mcns->cns->alns[0], mcns->cns->alns[1], mcns->cns->alns[2], mcns->cns->alns[3], mcns->cns->seq->string);
+						fprintf(stderr, "%s_%d_N%u_N%u\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%s\n", tag->string, mcns->eid, mcns->node1, mcns->node2, mcns->cns->qlen, mcns->cns->seq->size, mcns->cns->max_score, mcns->cns->alns[0], mcns->cns->alns[1], mcns->cns->alns[2], mcns->cns->alns[3], mcns->cns->seq->string);
+						if(0){
+							u4i j, x;
+							for(j=x=0;j<mcns->cns->cigars->size;j++){
+								if(mcns->cns->cigars->buffer[j] == DBGCNS_PATH_I){
+									fputc('-', stderr);
+								} else {
+									fputc("ACGT"[mcns->cns->cns->buffer[x]], stderr); x ++;
+								}
+							}
+							fputc('\n', stderr);
+							for(j=0;j<mcns->cns->cigars->size;j++){
+								fputc("MXID"[mcns->cns->cigars->buffer[j]], stderr);
+							}
+							fputc('\n', stderr);
+						}
+					}
+					cxs->buffer[mcns->eid] = cseqs->size;
+					for(j=0;j<mcns->cns->seq->size;j++) push_u1v(cseqs, base_bit_table[(int)mcns->cns->seq->string[j]]);
+					cys->buffer[mcns->eid] = cseqs->size;
+				}
+				reset_cns(mcns->cns);
+				clear_string(mcns->cns->seq);
+				mcns->eid = 0;
+				mcns->task = 0;
+				mcns->node1 = 0;
+				mcns->node2 = 0;
+				thread_end_iter(mcns);
+				push_u4v(qes, 0);
+				push_u4v(tes, 0);
+				for(i=1;i<eid;i++){
+					thread_wait_one(mcns);
+					if(mcns->task == MCNS_TASK_OVERLAP && mcns->ret.aln > 0){
+						if(cns_debug){
+							fprintf(stderr, "#%s_%d\t%d\t%d\t%d", tag->string, mcns->eid - 1, (int)mcns->seq1->size, mcns->ret.qb, mcns->ret.qe);
+							fprintf(stderr, "\t%s_%d\t%d\t%d\t%d", tag->string, mcns->eid, (int)mcns->seq2->size, mcns->ret.tb, mcns->ret.te);
+							fprintf(stderr, "\t%d\t%d\t%d\t%d\t%d\n", mcns->ret.aln, mcns->ret.mat, mcns->ret.mis, mcns->ret.ins, mcns->ret.del);
+						}
+						{
+							b = mcns->ret.qe;
+							e = mcns->ret.te;
+							if(1){
+								revise_joint_point(mcns->cigars, &b, &e);
+							}
+							qes->buffer[mcns->eid] = b;
+							tes->buffer[mcns->eid] = e;
+						}
+					}
+					mcns->task = MCNS_TASK_OVERLAP;
+					mcns->eid = i;
+					clear_u1v(mcns->seq1); append_array_u1v(mcns->seq1, cseqs->buffer + cxs->buffer[i-1], cys->buffer[i-1] - cxs->buffer[i-1]);
+					clear_u1v(mcns->seq2); append_array_u1v(mcns->seq2, cseqs->buffer + cxs->buffer[i], cys->buffer[i] - cxs->buffer[i]);
+					mcns->ret = KSWX_NULL;
+					push_u4v(qes, mcns->seq1->size);
+					push_u4v(tes, 0);
+					thread_wake(mcns);
+				}
+				push_u4v(qes, cys->buffer[eid-1] - cxs->buffer[eid-1]);
+				push_u4v(tes, 0);
+				thread_beg_iter(mcns);
+				thread_wait(mcns);
+				if(mcns->task == MCNS_TASK_OVERLAP && mcns->ret.aln > 0){
+					if(cns_debug){
+						fprintf(stderr, "#%s_%d\t%d\t%d\t%d", tag->string, mcns->eid - 1, (int)mcns->seq1->size, mcns->ret.qb, mcns->ret.qe);
+						fprintf(stderr, "\t%s_%d\t%d\t%d\t%d", tag->string, mcns->eid, (int)mcns->seq2->size, mcns->ret.tb, mcns->ret.te);
+						fprintf(stderr, "\t%d\t%d\t%d\t%d\t%d\n", mcns->ret.aln, mcns->ret.mat, mcns->ret.mis, mcns->ret.ins, mcns->ret.del);
+					}
+					{
+						b = mcns->ret.qe;
+						e = mcns->ret.te;
+						if(1){
+							revise_joint_point(mcns->cigars, &b, &e);
+						}
+						qes->buffer[mcns->eid] = b;
+						tes->buffer[mcns->eid] = e;
+					}
+				}
+				mcns->ret = KSWX_NULL;
+				mcns->eid = 0;
+				mcns->task = 0;
+				thread_end_iter(mcns);
+				// generate contig seq
+				clear_string(seq);
+				for(i=0;i<eid;i++){
+					beg = cxs->buffer[i] + tes->buffer[i];
+					end = cxs->buffer[i] + qes->buffer[i + 1];
+					for(m=beg;m<end;m++){
+						push_string(seq, bit_base_table[cseqs->buffer[m]]);
+					}
+					if(cns_debug){
+						fprintf(stderr, "=%s_%d\t%d\t%d\n", tag->string, i, seq->size - (end - beg), seq->size);
+					}
+				}
+				fprintf(out, ">%s len=%d\n", tag->string, seq->size);
+				for(j=0;j<seq->size;j+=100){
+					sl = num_min(j + 100, seq->size);
+					char ch = seq->string[sl];
+					seq->string[sl] = 0;
+					fprintf(out, "%s\n", seq->string + j);
+					seq->string[sl] = ch;
+				}
+			}
+			if(c == -1) break;
+			clear_string(tag);
+			ss = get_col_str(fr, 0) + 1;
+			sl = get_col_len(fr, 0) - 1;
+			for(j=0;j<sl;j++){
+				if(ss[j] == ' ') break;
+				push_string(tag, ss[j]);
+			}
+			clear_string(seq);
+			eid = 0;
+			clear_u1v(cseqs);
+			clear_u4v(cxs);
+			clear_u4v(cys);
+			clear_u4v(tes);
+			clear_u4v(qes);
+			thread_wait_one(mcns);
+		} else if(fr->line->string[0] == 'S' || fr->line->string[0] == 's'){
+			ss = get_col_str(fr, 5);
+			sl = get_col_len(fr, 5);
+			add_seq_cns(mcns->cns, ss, sl, (fr->line->string[0] == 'S'));
+		}
+	}
+	thread_beg_close(mcns);
+	free_cns(mcns->cns);
+	free_u1v(mcns->seq1);
+	free_u1v(mcns->seq2);
+	free_u32list(mcns->cigars);
+	thread_end_close(mcns);
+	free_u1v(cseqs);
+	free_u4v(cxs);
+	free_u4v(cys);
+	free_u4v(tes);
+	free_u4v(qes);
+	free_string(tag);
+	free_string(seq);
+	return 0;
+}
+
+int usage(){
+	printf(
+	"WTDBG-CNS: Consensuser for wtdbg\n"
+	"Author: Jue Ruan <ruanjue@gmail.com>\n"
+	"Version: 1.1\n"
+	"Usage: wtdbg-cns [options]\n"
+	"Options:\n"
+	" -t <int>    Number of threads, [1]\n"
+	" -i <string> Input file(s) *.utg.cns from wtdbg, +, [STDIN]\n"
+	" -o <string> Output files, [STDOUT]\n"
+	" -f          Force overwrite\n"
+	" -j <int>    Expected length of node, or say the overlap length of two adject units in layout file, [1000] bp\n"
+	"-----------------BEG DBG options---------------------------------\n"
+	" -k <int>    Kmer size for long reads, [15]\n"
+	" -Z <int>    Z-cutoff, drop the lower  (score / <-X>), [4]\n"
+	" -W <int>    W-cutoff, drop the lagger (position), [48]\n"
+	"             In DAG correction, -W set the bandwidth of alignment\n"
+	" -H <int>    High coverage bonus, [1]\n"
+	" -L <int>    High coverage cutoff = avg_cov / <-L> [10]\n"
+	" -c <int>    Candidate strategy, 0: best-kmers, 1: median length, 2: first (include), 3: first (exclude), 4: longest, 5, shortest, [0]\n"
+	"             In DAG correction, force to use strategy 2\n"
+	"-----------------END DBG options---------------------------------\n"
+	" -M <int>    Match score, [2]\n"
+	" -X <int>    Mismatch score, [-7]\n"
+	" -I <int>    Insertion score, [-3]\n"
+	" -D <int>    Deletion score, [-4]\n"
+	" -E <int>    Gap extension score, [-2]\n"
+	" -m <int>    1: DBG correction; 2: DAG correction, [1]\n"
+	" -S <int>    whether to correct structure before error correction, [1]\n"
+	" -v          Verbose\n"
+	" -V          Print version information and then exit\n"
+	"\n");
+	return 1;
+}
+
+int main(int argc, char **argv){
+	FileReader *fr;
+	cplist *infs;
+	FILE *out;
+	char *outf;
+	int c, ncpu, overwrite, reglen, ksize, Z, W, C, M, X, I, D, E, H, L, XX, OO, EE;
+	int candidate_mode, cns_model, corr_struct;
+	f4i pM, pX, pI, pD;
+	ncpu = 1;
+	reglen = 1000;
+	ksize = 15;
+	Z = 4;
+	W = 48;
+	C = 1;
+	M = 2;
+	X = -7;
+	I = -3;
+	D = -4;
+	E = -2;
+	H = 1;
+	L = 10;
+	XX = -4;
+	OO = -2;
+	EE = -1;
+	candidate_mode = 0;
+	cns_model = 1;
+	corr_struct = 1;
+	pM = log(0.85);
+	pX = log(0.10);
+	pI = log(0.03);
+	pD = log(0.02);
+	infs = init_cplist(4);
+	outf = NULL;
+	overwrite = 0;
+	while((c = getopt(argc, argv, "hvVt:k:i:o:fj:Z:W:C:M:X:I:D:E:H:L:m:c:S:")) != -1){
+		switch(c){
+			case 'h': return usage();
+			case 't': ncpu = atoi(optarg); break;
+			case 'k': ksize = atoi(optarg); break;
+			case 'i': push_cplist(infs, optarg); break;
+			case 'o': outf = optarg; break;
+			case 'f': overwrite = 1; break;
+			case 'j': reglen = atoi(optarg); break;
+			case 'Z': Z = atoi(optarg); break;
+			case 'W': W = atoi(optarg); break;
+			case 'C': C = atoi(optarg); break;
+			case 'M': M = atoi(optarg); break;
+			case 'X': X = atoi(optarg); break;
+			case 'I': I = atoi(optarg); break;
+			case 'D': D = atoi(optarg); break;
+			case 'E': E = atoi(optarg); break;
+			case 'H': H = atoi(optarg); break;
+			case 'L': L = atoi(optarg); break;
+			case 'm': cns_model = atoi(optarg); break;
+			case 'c': candidate_mode = atoi(optarg); break;
+			case 'S': corr_struct = atoi(optarg); break;
+			case 'v': cns_debug ++; break;
+			case 'V': fprintf(stdout, "wtdbg-cns 1.1\n"); return 0;
+			default: return usage();
+		}
+	}
+	BEG_STAT_PROC_INFO(stderr, argc, argv);
+	if(cns_model != MCNS_TASK_DBGCNS && cns_model != MCNS_TASK_DAGCNS){
+		return usage();
+	}
+	if(ncpu <= 0 && _sig_proc_deamon) ncpu = _sig_proc_deamon->ncpu;
+	if(ncpu <= 0){
+		fprintf(stderr, " -- Invalid cpu number '%d' in %s -- %s:%d --\n", ncpu, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+		return 1;
+	}
+	if(outf && !overwrite && file_exists(outf)){
+		fprintf(stderr, "File exists! '%s'\n\n", outf);
+		return usage();
+	}
+	if(cns_debug > 1) DBGCNS_DEBUG = 1;
+	if(infs->size) fr = open_all_filereader(infs->size, infs->buffer, 1);
+	else fr = open_filereader(NULL, 1);
+	if(outf){
+		out = open_file_for_write(outf, NULL, 1);
+	} else out = stdout;
+	run(reglen, ksize, Z, W, M, X, I, D, E, H, L, XX, OO, EE, cns_model, pM, pX, pI, pD, candidate_mode, corr_struct, ncpu, fr, out);
+	close_filereader(fr);
+	if(outf) fclose(out);
+	free_cplist(infs);
+	END_STAT_PROC_INFO(stderr);
+	return 0;
+}
+

wtdbg2.pl

@@ -0,0 +1,116 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+use Getopt::Std;
+
+my %opts = (t=>4, m=>"2g");
+getopts('t:x:o:a:g:X:M:OPp:k:', \%opts);
+die (qq/Usage: wtdbg2.pl [options] <reads.fa>
+Options:
+  -o STR     output prefix [first input]
+  -t INT     number of threads [$opts{t}]
+  -x STR     preset: rs, ont, sq, ccs []
+  -g NUM     estimated genome size []
+  -X FLOAT   coverage cutoff [wtdbg2 default]
+  -k INT     k-mer length [wtdbg2 default]
+  -p INT     length of HPC k-mers [wtdbg2 default]
+  -a STR     additional wtdbg2 options []
+  -M STR     minimap2 preset, according to -x when set []
+  -m NUM     samtools sort block size [$opts{m}]
+  -P         skip minimap2-based polishing
+  -O         output without execution
+/) if (@ARGV == 0);
+
+$opts{asm} = gwhich("wtdbg2") || die;
+$opts{cns} = gwhich("wtpoa-cns") || die;
+unless (defined $opts{P}) {
+	$opts{smt} = gwhich("samtools") || die;
+	$opts{mm2} = gwhich("minimap2") || die;
+}
+
+my $prefix = defined($opts{o})? $opts{o} : $ARGV[0];
+my $smt_threads = $opts{t} < 4? 4 : $opts{t};
+
+my $asm_opt = "";
+$asm_opt .= " -x $opts{x}" if defined($opts{x});
+$asm_opt .= " -g $opts{g}" if defined($opts{g}) && $opts{g} =~ /^\d/;
+$asm_opt .= " -X $opts{X}" if defined($opts{X});
+$asm_opt .= " -k $opts{k}" if defined($opts{k}) && $opts{k} =~ /^\d+$/;
+$asm_opt .= " -p $opts{p}" if defined($opts{p}) && $opts{p} =~ /^\d+$/;
+$asm_opt .= " $opts{a}" if defined($opts{a});
+$asm_opt .= " -t $opts{t} -fo $prefix";
+$asm_opt .= " -i $ARGV[$_]" for (0 .. @ARGV-1);
+
+my %map_opts = (
+	rs=>'map-pb', rsII=>'map-pb', sq=>'map-pb', sequel=>'map-pb',
+	ont=>'map-ont', nanopore=>'map-ont',
+	ccs=>'asm20', corrected=>'asm20'
+);
+
+if (not defined $opts{M}) {
+	if (defined $opts{x}) {
+		$opts{M} = $map_opts{lc $opts{x}};
+		$opts{M} = 'map-ont' unless(defined $opts{M});
+	} else {
+		$opts{M} = 'map-ont';
+	}
+}
+
+my $mm2_input = join(' ', @ARGV);
+
+my @lines = ();
+push(@lines, qq/$opts{asm} $asm_opt/);
+push(@lines, qq/$opts{cns} -t $opts{t} -i $prefix.ctg.lay.gz -fo $prefix.raw.fa/);
+unless (defined $opts{P}) {
+	push(@lines, qq/$opts{mm2} -ax $opts{M} -t$opts{t} -r2k $prefix.raw.fa $mm2_input | $opts{smt} sort -m $opts{m} -\@$smt_threads -o $prefix.bam/);
+	push(@lines, qq/$opts{smt} view -F0x900 $prefix.bam | $opts{cns} -t $opts{t} -d $prefix.raw.fa -i - -fo $prefix.cns.fa/);
+}
+
+if (defined $opts{O}) {
+	print "$_\n" for (@lines);
+} else {
+	for (@lines) {
+		system($_) == 0 || die "The following command returns non-zero code:\n  $_\n";
+	}
+}
+
+sub which {
+	my $file = shift;
+	my $path = (@_)? shift : $ENV{PATH};
+	return if (!defined($path));
+	foreach my $x (split(":", $path)) {
+		$x =~ s/\/$//;
+		return "$x/$file" if (-x "$x/$file") && (-f "$x/$file");
+	}
+	return;
+}
+
+sub gwhich {
+    my $progname = shift;
+    my $addtional_path = shift if (@_);
+    my $dirname = &dirname($0);
+    my $tmp;
+
+    chomp($dirname);
+    if ($progname =~ /^\// && (-x $progname) && (-f $progname)) {
+        return $progname;
+    } elsif (defined($addtional_path) && ($tmp = &which($progname, $addtional_path))) {
+        return $tmp;
+    } elsif (defined($dirname) && (-x "$dirname/$progname") && (-f "$dirname/$progname")) {
+        return "$dirname/$progname";
+    } elsif ((-x "./$progname") && (-f "./$progname")) {
+        return "./$progname";
+    } elsif (($tmp = &which($progname))) {
+        return $tmp;
+    } else {
+        return;
+    }
+}
+
+sub dirname {
+	my $prog = shift;
+	return '.' unless ($prog =~ /\//);
+	$prog =~ s/\/[^\s\/]+$//g;
+	return $prog;
+}

wtpoa-cns.c

@@ -0,0 +1,280 @@
+/*
+ *
+ * Copyright (c) 2011, Jue Ruan <ruanjue@gmail.com>
+ *
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "wtpoa.h"
+
+#ifndef VERSION
+#define VERSION 0.0
+#endif
+#ifndef RELEASE
+#define RELEASE 19830203
+#endif
+
+int usage(){
+	printf(
+	"WTPOA-CNS: Consensuser for wtdbg using PO-MSA\n"
+	"Author: Jue Ruan <ruanjue@gmail.com>\n"
+	"Version: %s (%s)\n"
+	"Usage: wtpoa-cns [options]\n"
+	"Options:\n"
+	" -t <int>    Number of threads, [4]\n"
+	" -d <string> Reference sequences for SAM input, will invoke sorted-SAM input mode\n"
+	" -u <int>    XORed flags to handle SAM input. [0]\n"
+	"             0x1: Only process reference regions present in/between SAM alignments\n"
+	"             0x2: Don't fileter secondary/supplementary SAM records with flag (0x100 | 0x800)\n"
+	" -r          Force to use reference mode\n"
+	" -p <string> Similar with -d, but translate SAM into wtdbg layout file\n"
+	" -i <string> Input file(s) *.ctg.lay from wtdbg, +, [STDIN]\n"
+	"             Or sorted SAM files when having -d/-p\n"
+	" -o <string> Output files, [STDOUT]\n"
+	" -f          Force overwrite\n"
+	" -j <int>    Expected max length of node, or say the overlap length of two adjacent units in layout file, [1500] bp\n"
+	"             overlap will be default to 1500(or 150 for sam-sr) when having -d/-p, block size will be 2.5 * overlap\n"
+	" -b <int>    Bonus for tri-bases match, [0]\n"
+	" -M <int>    Match score, [2]\n"
+	" -X <int>    Mismatch score, [-5]\n"
+	" -I <int>    Insertion score, [-2]\n"
+	" -D <int>    Deletion score, [-4]\n"
+	" -H <float>  Homopolymer merge score used in dp-call-cns mode, [-3]\n"
+	" -B <expr>   Bandwidth in POA, [Wmin[,Wmax[,mat_rate]]], mat_rate = matched_bases/total_bases [64,1024,0.92]\n"
+	"             Program will double bandwidth from Wmin to Wmax when mat_rate is lower than setting\n"
+	" -W <int>    Window size in the middle of the first read for fast align remaining reads, [200]\n"
+	"             If $W is negative, will disable fast align, but use the abs($W) as Band align score cutoff\n"
+	" -w <int>    Min size of aligned size in window, [$W * 0.5]\n"
+	" -A          Abort TriPOA when any read cannot be fast aligned, then try POA\n"
+	" -S <int>    Shuffle mode, 0: don't shuffle reads, 1: by shared kmers, 2: subsampling. [1]\n"
+	" -R <int>    Realignment bandwidth, 0: disable, [16]\n"
+	" -c <int>    Consensus mode: 0, run-length; 1, dp-call-cns, [0]\n"
+	" -C <int>    Min count of bases to call a consensus base, [3]\n"
+	" -F <float>  Min frequency of non-gap bases to call a consensus base, [0.5]\n"
+	" -N <int>    Max number of reads in PO-MSA [20]\n"
+	"             Keep in mind that I am not going to generate high accurate consensus sequences here\n"
+	" -x <string> Presets, []\n"
+	"             sam-sr: polishs contigs from short reads mapping, accepts sorted SAM files\n"
+	"                     shorted for '-j 50 -W 0 -R 0 -b 1 -c 1 -N 50 -rS 2'\n"
+	" -q          Quiet\n"
+	" -v          Verbose\n"
+	" -V          Print version information and then exit\n"
+	"\n", TOSTR(VERSION), TOSTR(RELEASE));
+	return 1;
+}
+
+int main(int argc, char **argv){
+	POGPar par;
+	CTGCNS *cc;
+	ctg_cns_t *ctg;
+	SeqBank *refs;
+	FileReader *fr, *db;
+	cplist *infs, *dbfs;
+	FILE *out;
+	char *outf;
+	u4i i;
+	int reglen, shuffle, winlen, winmin, fail_skip;
+	int seqmax, wsize, print_lay, flags;
+	int c, ncpu, overwrite, quiet;
+	par = DEFAULT_POG_PAR;
+	ncpu = 4;
+	quiet = 0;
+	shuffle = 1;
+	seqmax = 20;
+	winlen = 200;
+	winmin = 0;
+	fail_skip = 1;
+	reglen = 0;
+	infs = init_cplist(4);
+	dbfs = init_cplist(4);
+	outf = NULL;
+	overwrite = 0;
+	print_lay = 0;
+	flags = 0;
+	while((c = getopt(argc, argv, "hqvVt:d:rp:u:i:o:fj:S:B:W:w:Ab:M:X:I:D:E:H:R:c:C:F:N:x:")) != -1){
+		switch(c){
+			case 'h': return usage();
+			case 't': ncpu = atoi(optarg); break;
+			case 'p': print_lay = 1;
+			case 'd': push_cplist(dbfs, optarg); break;
+			case 'u': flags = atoi(optarg); break;
+			case 'r': par.refmode = 1; break;
+			case 'i': push_cplist(infs, optarg); break;
+			case 'o': outf = optarg; break;
+			case 'f': overwrite = 1; break;
+			case 'j': reglen = atoi(optarg); break;
+			case 'S': shuffle = atoi(optarg); break;
+			case 'B':
+				{
+					char *ptr;
+					par.W = strtol(optarg, &ptr, 10);
+					if(ptr && ptr[0] == ','){
+						par.Wmax = strtol(ptr + 1, &ptr, 10);
+						if(ptr && ptr[0] == ','){
+							par.W_mat_rate = atof(ptr + 1);
+						}
+					}
+				}
+				break;
+			case 'W': winlen = atoi(optarg); break;
+			case 'w': winmin = atoi(optarg); break;
+			case 'A': fail_skip = 0; break;
+			case 'b': par.tribase = atoi(optarg); break;
+			case 'M': par.M = atoi(optarg); break;
+			case 'X': par.X = atoi(optarg); break;
+			case 'I': par.I = atoi(optarg); break;
+			case 'D': par.D = atoi(optarg); break;
+			case 'E': par.E = atoi(optarg); break;
+			case 'H': par.H = atof(optarg); break;
+			case 'R': par.rW = atoi(optarg); break;
+			case 'c': par.cnsmode = atoi(optarg); break;
+			case 'C': par.msa_min_cnt = atoi(optarg); break;
+			case 'F': par.msa_min_freq = atof(optarg); break;
+			case 'N': seqmax = atoi(optarg); break;
+			case 'x':
+				if(strcasecmp(optarg, "sam-sr") == 0){
+					winlen = 0;
+					reglen = 150;
+					par.rW = 0;
+					par.tribase = 1;
+					par.cnsmode = 1;
+					seqmax = 50;
+					par.refmode = 1;
+					shuffle = 2;
+				} else {
+					fprintf(stderr, "Unknown preset[%s]\n", optarg);
+					return 1;
+				}
+				break;
+			case 'q': quiet = 1; break;
+			case 'v': cns_debug ++; break;
+			case 'V': fprintf(stdout, "wtpoa-cns %s\n", TOSTR(VERSION)); return 0;
+			default: return usage();
+		}
+	}
+	if(quiet){
+		int devnull;
+		devnull = open("/dev/null", O_WRONLY);
+		dup2(devnull, STDERR_FILENO);
+	}
+	BEG_STAT_PROC_INFO(stderr, argc, argv);
+	//SET_PROC_LIMIT(10 * 1024 * 1024 * 1024ULL, 0); // TODO: remove it after debug
+	if(reglen == 0){
+		reglen = 1500;
+	}
+	wsize = 2.5 * reglen;
+	if(winlen < 0){
+		par.W_score = - winlen;
+		winlen = 0;
+	}
+	if(winmin <= 0){
+		winmin = winlen * 0.5;
+	}
+	if(seqmax <= 0 || seqmax >= POG_RDCNT_MAX){
+		seqmax = POG_RDCNT_MAX;
+	}
+	if(ncpu <= 0 && _sig_proc_deamon) ncpu = _sig_proc_deamon->ncpu;
+	if(ncpu <= 0){
+		fprintf(stderr, " -- Invalid cpu number '%d' in %s -- %s:%d --\n", ncpu, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+		return 1;
+	}
+	if(outf && !overwrite && file_exists(outf)){
+		fprintf(stderr, "File exists! '%s'\n\n", outf);
+		return usage();
+	}
+	if(infs->size) fr = open_all_filereader(infs->size, infs->buffer, 1);
+	else fr = open_filereader(NULL, 1);
+	if(outf){
+		out = open_file_for_write(outf, NULL, 1);
+	} else out = stdout;
+	if(shuffle == 2){
+		srand48(13);
+	}
+	if(dbfs->size == 0){
+		WTLAYBlock *wb;
+		wb = init_wtlayblock(fr);
+		cc = init_ctgcns(wb, iter_wtlayblock, info_wtlayblock, ncpu, shuffle, seqmax, winlen, winmin, fail_skip, reglen, &par);
+		cc->print_progress = 100;
+		if(print_lay){
+			print_lays_ctgcns(cc, out);
+		} else {
+			while((ctg = iter_ctgcns(cc))){
+				fprintf(out, ">%s len=%d\n", ctg->tag->string, (u4i)ctg->cns->size);
+				for(i=0;i<ctg->cns->size;i+=100){
+					if(i + 100 <= ctg->cns->size){
+						println_seq_basebank(ctg->cns, i, 100, out);
+					} else {
+						println_seq_basebank(ctg->cns, i, ctg->cns->size - i, out);
+					}
+				}
+				fflush(out);
+				if(cc->cycs->size){ // keep only one for reuse
+					free_ctg(ctg);
+				} else {
+					repay_ctgcns(cc, ctg);
+				}
+			}
+		}
+		free_ctgcns(cc);
+		free_wtlayblock(wb);
+	} else {
+		SAMBlock *sb;
+		BioSequence *seq;
+		if(2 * reglen >= wsize){
+			fprintf(stderr, " -- SAM Input mode: -w wsize(%d), -j reglen(%d), MUST has 2 * reglen >= wsize in %s -- %s:%d --\n", wsize, reglen, __FUNCTION__, __FILE__, __LINE__); fflush(stderr);
+			return 1;
+		}
+		refs = init_seqbank();
+		db = open_all_filereader(dbfs->size, dbfs->buffer, 1);
+		seq = init_biosequence();
+		while(readseq_filereader(db, seq)){
+			push_seqbank(refs, seq->tag->string, seq->tag->size, seq->seq->string, seq->seq->size);
+		}
+		free_biosequence(seq);
+		close_filereader(db);
+		sb = init_samblock(refs, fr, wsize, reglen * 2 / 3, flags);
+		cc = init_ctgcns(sb, iter_samblock, info_samblock, ncpu, shuffle, seqmax, winlen, winmin, fail_skip, reglen, &par);
+		cc->print_progress = 100;
+		if(print_lay){
+			print_lays_ctgcns(cc, out);
+		} else {
+			while((ctg = iter_ctgcns(cc))){
+				fprintf(out, ">%s len=%d\n", ctg->tag->string, (u4i)ctg->cns->size);
+				for(i=0;i<ctg->cns->size;i+=100){
+					if(i + 100 <= ctg->cns->size){
+						println_seq_basebank(ctg->cns, i, 100, out);
+					} else {
+						println_seq_basebank(ctg->cns, i, ctg->cns->size - i, out);
+					}
+				}
+				fflush(out);
+				if(cc->cycs->size){ // keep only one for reuse
+					free_ctg(ctg);
+				} else {
+					repay_ctgcns(cc, ctg);
+				}
+			}
+		}
+		free_ctgcns(cc);
+		free_samblock(sb);
+		free_seqbank(refs);
+	}
+	close_filereader(fr);
+	if(outf) fclose(out);
+	free_cplist(infs);
+	free_cplist(dbfs);
+	END_STAT_PROC_INFO(stderr);
+	return 0;
+}