finishm 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.document +5 -0
- data/.gitmodules +3 -0
- data/.rspec +1 -0
- data/Gemfile +31 -0
- data/LICENSE.txt +20 -0
- data/README.md +59 -0
- data/Rakefile +51 -0
- data/VERSION +1 -0
- data/bin/assembly_visualiser +106 -0
- data/bin/check_primer_combinations.rb +73 -0
- data/bin/contig_joiner.rb +244 -0
- data/bin/contigs_against_assembly.rb +153 -0
- data/bin/finishm +143 -0
- data/bin/finishm_assembler +55 -0
- data/bin/finishm_gap_closer.rb +241 -0
- data/bin/kmer_abundance_file_tool.rb +49 -0
- data/bin/kmer_pattern_to_assembly.rb +377 -0
- data/bin/kmer_profile_finder.rb +92 -0
- data/bin/kmers_count_parse.d +52 -0
- data/bin/kmers_count_tabulate.d +123 -0
- data/bin/kmers_count_tabulate.rb +84 -0
- data/bin/pcr_result_parser.rb +108 -0
- data/bin/primer_finder.rb +119 -0
- data/bin/read_selection_by_kmer.d +174 -0
- data/bin/scaffold_by_pattern.rb +119 -0
- data/bin/scaffold_connection_possibilities_to_knowns.rb +193 -0
- data/bin/scaffold_end_coverages.rb +69 -0
- data/bin/trail_validator.rb +84 -0
- data/ext/mkrf_conf.rb +56 -0
- data/ext/src/Makefile +140 -0
- data/ext/src/src/allocArray.c +305 -0
- data/ext/src/src/allocArray.h +86 -0
- data/ext/src/src/autoOpen.c +107 -0
- data/ext/src/src/autoOpen.h +18 -0
- data/ext/src/src/binarySequences.c +813 -0
- data/ext/src/src/binarySequences.h +125 -0
- data/ext/src/src/concatenatedGraph.c +233 -0
- data/ext/src/src/concatenatedGraph.h +30 -0
- data/ext/src/src/concatenatedPreGraph.c +262 -0
- data/ext/src/src/concatenatedPreGraph.h +29 -0
- data/ext/src/src/correctedGraph.c +2643 -0
- data/ext/src/src/correctedGraph.h +32 -0
- data/ext/src/src/dfib.c +509 -0
- data/ext/src/src/dfib.h +69 -0
- data/ext/src/src/dfibHeap.c +89 -0
- data/ext/src/src/dfibHeap.h +39 -0
- data/ext/src/src/dfibpriv.h +105 -0
- data/ext/src/src/fib.c +628 -0
- data/ext/src/src/fib.h +78 -0
- data/ext/src/src/fibHeap.c +79 -0
- data/ext/src/src/fibHeap.h +41 -0
- data/ext/src/src/fibpriv.h +110 -0
- data/ext/src/src/globals.h +154 -0
- data/ext/src/src/graph.c +3932 -0
- data/ext/src/src/graph.h +233 -0
- data/ext/src/src/graphReConstruction.c +1472 -0
- data/ext/src/src/graphReConstruction.h +30 -0
- data/ext/src/src/graphStats.c +2167 -0
- data/ext/src/src/graphStats.h +72 -0
- data/ext/src/src/graphStructures.h +52 -0
- data/ext/src/src/kmer.c +652 -0
- data/ext/src/src/kmer.h +73 -0
- data/ext/src/src/kmerOccurenceTable.c +236 -0
- data/ext/src/src/kmerOccurenceTable.h +44 -0
- data/ext/src/src/kseq.h +223 -0
- data/ext/src/src/locallyCorrectedGraph.c +557 -0
- data/ext/src/src/locallyCorrectedGraph.h +40 -0
- data/ext/src/src/passageMarker.c +677 -0
- data/ext/src/src/passageMarker.h +137 -0
- data/ext/src/src/preGraph.c +1717 -0
- data/ext/src/src/preGraph.h +106 -0
- data/ext/src/src/preGraphConstruction.c +990 -0
- data/ext/src/src/preGraphConstruction.h +26 -0
- data/ext/src/src/probe_node_finder.c +84 -0
- data/ext/src/src/probe_node_finder.h +6 -0
- data/ext/src/src/readCoherentGraph.c +557 -0
- data/ext/src/src/readCoherentGraph.h +30 -0
- data/ext/src/src/readSet.c +1734 -0
- data/ext/src/src/readSet.h +67 -0
- data/ext/src/src/readToNode.c +218 -0
- data/ext/src/src/readToNode.h +35 -0
- data/ext/src/src/recycleBin.c +199 -0
- data/ext/src/src/recycleBin.h +58 -0
- data/ext/src/src/roadMap.c +342 -0
- data/ext/src/src/roadMap.h +65 -0
- data/ext/src/src/run.c +318 -0
- data/ext/src/src/run.h +52 -0
- data/ext/src/src/run2.c +744 -0
- data/ext/src/src/runReadToNode.c +29 -0
- data/ext/src/src/scaffold.c +1876 -0
- data/ext/src/src/scaffold.h +64 -0
- data/ext/src/src/shortReadPairs.c +1243 -0
- data/ext/src/src/shortReadPairs.h +32 -0
- data/ext/src/src/splay.c +259 -0
- data/ext/src/src/splay.h +43 -0
- data/ext/src/src/splayTable.c +1315 -0
- data/ext/src/src/splayTable.h +31 -0
- data/ext/src/src/tightString.c +362 -0
- data/ext/src/src/tightString.h +82 -0
- data/ext/src/src/utility.c +199 -0
- data/ext/src/src/utility.h +98 -0
- data/ext/src/third-party/zlib-1.2.3/ChangeLog +855 -0
- data/ext/src/third-party/zlib-1.2.3/FAQ +339 -0
- data/ext/src/third-party/zlib-1.2.3/INDEX +51 -0
- data/ext/src/third-party/zlib-1.2.3/Makefile +154 -0
- data/ext/src/third-party/zlib-1.2.3/Makefile.in +154 -0
- data/ext/src/third-party/zlib-1.2.3/README +125 -0
- data/ext/src/third-party/zlib-1.2.3/adler32.c +149 -0
- data/ext/src/third-party/zlib-1.2.3/adler32.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/algorithm.txt +209 -0
- data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.pup +66 -0
- data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.sas +65 -0
- data/ext/src/third-party/zlib-1.2.3/as400/bndsrc +132 -0
- data/ext/src/third-party/zlib-1.2.3/as400/compile.clp +123 -0
- data/ext/src/third-party/zlib-1.2.3/as400/readme.txt +111 -0
- data/ext/src/third-party/zlib-1.2.3/as400/zlib.inc +331 -0
- data/ext/src/third-party/zlib-1.2.3/compress.c +79 -0
- data/ext/src/third-party/zlib-1.2.3/compress.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/configure +459 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/README.contrib +71 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/buffer_demo.adb +106 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/mtest.adb +156 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/read.adb +156 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/readme.txt +65 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/test.adb +463 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.adb +225 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.ads +114 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.adb +141 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.ads +450 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.adb +701 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.ads +328 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.gpr +20 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm586/README.586 +43 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm586/match.S +364 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm686/README.686 +34 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm686/match.S +329 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/Makefile +8 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/README +4 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.c +444 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.h +71 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.pk +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.txt +1 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLib.pas +557 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLibConst.pas +11 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/readme.txt +76 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/zlibd32.mak +93 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.build +33 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.chm +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.sln +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/AssemblyInfo.cs +58 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/ChecksumImpl.cs +202 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CircularBuffer.cs +83 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CodecBase.cs +198 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Deflater.cs +106 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.cs +288 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.csproj +141 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/GZipStream.cs +301 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Inflater.cs +105 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/UnitTests.cs +274 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/LICENSE_1_0.txt +23 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/readme.txt +58 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/README +1 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.c +608 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.h +37 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inffix9.h +107 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inflate9.h +47 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.c +323 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.h +55 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffas86.c +1157 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffast.S +1368 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/test.cpp +24 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.cpp +329 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.h +128 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream.h +307 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream_test.cpp +25 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/README +35 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/TODO +17 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/test.cc +50 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.cc +479 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.h +466 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masm686/match.asm +413 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/bld_ml64.bat +2 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.asm +513 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffas8664.c +186 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.asm +392 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/readme.txt +28 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/bld_ml32.bat +2 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.asm +972 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32c.c +62 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.asm +1083 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/mkasm.bat +3 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/readme.txt +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ChangeLogUnzip +67 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/Makefile +25 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/crypt.h +132 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.c +177 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.h +75 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.c +270 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.h +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/miniunz.c +585 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/minizip.c +420 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.c +281 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.h +31 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.c +1598 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.h +354 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.c +1219 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.h +235 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/example.pas +599 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/readme.txt +76 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibd32.mak +93 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibpas.pas +236 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/Makefile +8 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/README +63 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.c +837 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.h +31 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/zeros.raw +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.c +275 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.txt +10 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile +14 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile.msc +17 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/untgz.c +674 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/readme.txt +73 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/miniunz.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/minizip.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/testzlib.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlib.rc +32 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibstat.vcproj +246 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.def +92 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.sln +78 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.vcproj +445 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/miniunz.vcproj +566 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/minizip.vcproj +563 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlib.vcproj +948 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlibdll.vcproj +567 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlib.rc +32 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibstat.vcproj +870 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.def +92 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.sln +144 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.vcproj +1219 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.c +423 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.h +441 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.c +1736 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.h +331 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/example +0 -0
- data/ext/src/third-party/zlib-1.2.3/example.c +565 -0
- data/ext/src/third-party/zlib-1.2.3/examples/README.examples +42 -0
- data/ext/src/third-party/zlib-1.2.3/examples/fitblk.c +233 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gun.c +693 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzappend.c +500 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzjoin.c +448 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzlog.c +413 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzlog.h +58 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zlib_how.html +523 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zpipe.c +191 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zran.c +404 -0
- data/ext/src/third-party/zlib-1.2.3/gzio.c +1026 -0
- data/ext/src/third-party/zlib-1.2.3/gzio.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/infback.c +623 -0
- data/ext/src/third-party/zlib-1.2.3/infback.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.c +318 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.h +11 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inffixed.h +94 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.c +1368 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.h +115 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.c +329 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.h +55 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/libz.a +0 -0
- data/ext/src/third-party/zlib-1.2.3/make_vms.com +461 -0
- data/ext/src/third-party/zlib-1.2.3/minigzip +0 -0
- data/ext/src/third-party/zlib-1.2.3/minigzip.c +322 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.bor +109 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.dj2 +104 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.emx +69 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.msc +106 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.tc +94 -0
- data/ext/src/third-party/zlib-1.2.3/old/Makefile.riscos +151 -0
- data/ext/src/third-party/zlib-1.2.3/old/README +3 -0
- data/ext/src/third-party/zlib-1.2.3/old/descrip.mms +48 -0
- data/ext/src/third-party/zlib-1.2.3/old/os2/Makefile.os2 +136 -0
- data/ext/src/third-party/zlib-1.2.3/old/os2/zlib.def +51 -0
- data/ext/src/third-party/zlib-1.2.3/old/visual-basic.txt +160 -0
- data/ext/src/third-party/zlib-1.2.3/old/zlib.html +971 -0
- data/ext/src/third-party/zlib-1.2.3/projects/README.projects +41 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/README.txt +73 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/example.dsp +278 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/minigzip.dsp +278 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsp +609 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsw +59 -0
- data/ext/src/third-party/zlib-1.2.3/qnx/package.qpg +141 -0
- data/ext/src/third-party/zlib-1.2.3/trees.c +1219 -0
- data/ext/src/third-party/zlib-1.2.3/trees.h +128 -0
- data/ext/src/third-party/zlib-1.2.3/trees.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/uncompr.c +61 -0
- data/ext/src/third-party/zlib-1.2.3/uncompr.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/win32/DLL_FAQ.txt +397 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.bor +107 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.emx +69 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.gcc +141 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.msc +126 -0
- data/ext/src/third-party/zlib-1.2.3/win32/VisualC.txt +3 -0
- data/ext/src/third-party/zlib-1.2.3/win32/zlib.def +60 -0
- data/ext/src/third-party/zlib-1.2.3/win32/zlib1.rc +39 -0
- data/ext/src/third-party/zlib-1.2.3/zconf.h +332 -0
- data/ext/src/third-party/zlib-1.2.3/zconf.in.h +332 -0
- data/ext/src/third-party/zlib-1.2.3/zlib.3 +159 -0
- data/ext/src/third-party/zlib-1.2.3/zlib.h +1357 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.c +318 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.h +269 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.o +0 -0
- data/lib/assembly/a_b_visualiser.rb +169 -0
- data/lib/assembly/acyclic_connection_finder.rb +81 -0
- data/lib/assembly/all_orfs.rb +615 -0
- data/lib/assembly/bad_format_writer.rb +46 -0
- data/lib/assembly/bam_probe_read_selector.rb +48 -0
- data/lib/assembly/bubbly_assembler.rb +842 -0
- data/lib/assembly/c_probe_node_finder.rb +38 -0
- data/lib/assembly/connection_interpreter.rb +350 -0
- data/lib/assembly/contig_printer.rb +400 -0
- data/lib/assembly/coverage_based_graph_filter.rb +68 -0
- data/lib/assembly/depth_first_search.rb +63 -0
- data/lib/assembly/dijkstra.rb +216 -0
- data/lib/assembly/fluffer.rb +253 -0
- data/lib/assembly/graph_explorer.rb +85 -0
- data/lib/assembly/graph_generator.rb +315 -0
- data/lib/assembly/height_finder.rb +355 -0
- data/lib/assembly/hybrid_velvet_graph.rb +70 -0
- data/lib/assembly/input_genome.rb +182 -0
- data/lib/assembly/kmer_coverage_based_path_filter.rb +65 -0
- data/lib/assembly/node_finder.rb +171 -0
- data/lib/assembly/oriented_node_trail.rb +507 -0
- data/lib/assembly/paired_end_assembler.rb +53 -0
- data/lib/assembly/paired_end_neighbour_finder.rb +176 -0
- data/lib/assembly/probed_graph.rb +105 -0
- data/lib/assembly/read_input.rb +79 -0
- data/lib/assembly/read_to_node.rb +37 -0
- data/lib/assembly/scaffold_breaker.rb +126 -0
- data/lib/assembly/sequence_hasher.rb +71 -0
- data/lib/assembly/single_coherent_paths_between_nodes.rb +533 -0
- data/lib/assembly/single_coherent_wanderer.rb +261 -0
- data/lib/assembly/single_ended_assembler.rb +441 -0
- data/lib/assembly/velvet_c_binding.rb +54 -0
- data/lib/assembly/velvet_graph_sequence_extractor.rb +123 -0
- data/lib/external/VERSION +1 -0
- data/lib/finishm/assemble.rb +224 -0
- data/lib/finishm/explore.rb +217 -0
- data/lib/finishm/finisher.rb +303 -0
- data/lib/finishm/fluff.rb +122 -0
- data/lib/finishm/gapfiller.rb +325 -0
- data/lib/finishm/orfs_finder.rb +88 -0
- data/lib/finishm/path_counter.rb +90 -0
- data/lib/finishm/primers.rb +425 -0
- data/lib/finishm/primers_check.rb +176 -0
- data/lib/finishm/roundup.rb +344 -0
- data/lib/finishm/sequence.rb +142 -0
- data/lib/finishm/visualise.rb +430 -0
- data/lib/finishm/wander.rb +270 -0
- data/lib/kmer_abundance_pattern.rb +79 -0
- data/lib/kmer_multi_abundance_file.rb +48 -0
- data/lib/oligo_designer.rb +88 -0
- data/lib/priner.rb +66 -0
- data/spec/acyclic_connection_finder_spec.rb +551 -0
- data/spec/all_orfs_spec.rb +443 -0
- data/spec/assemble_spec.rb +186 -0
- data/spec/bubbly_assembler_spec.rb +707 -0
- data/spec/c_node_finder_spec.rb +58 -0
- data/spec/connection_interpreter_spec.rb +284 -0
- data/spec/contig_printer_spec.rb +291 -0
- data/spec/coverage_based_graph_filter_spec.rb +102 -0
- data/spec/data/6_3e4e5e6e.1vANME.bam +0 -0
- data/spec/data/6_3e4e5e6e.1vANME.bam.bai +0 -0
- data/spec/data/acyclic_connection_finder/1/probes.fa +5 -0
- data/spec/data/acyclic_connection_finder/1/random1.fa +2 -0
- data/spec/data/acyclic_connection_finder/1/random1.sammy.fa.gz +0 -0
- data/spec/data/acyclic_connection_finder/1/random2.fa +2 -0
- data/spec/data/acyclic_connection_finder/1/random2.sammy.fa.gz +0 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.fa +39 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.slightly_changed.fa +39 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/reads_combined.fa.gz +0 -0
- data/spec/data/assembly_visualiser/Contig_6_1_to_250.fa.kmers31 +220 -0
- data/spec/data/assembly_visualiser/Contig_7_1_to_250.fa.kmers31 +220 -0
- data/spec/data/assembly_visualiser/Graph +46 -0
- data/spec/data/assembly_visualiser/start_kmers1 +2 -0
- data/spec/data/bands.csv +1 -0
- data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq +0 -0
- data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq.names +544 -0
- data/spec/data/c_probe_node_finder/1/Graph2 +668 -0
- data/spec/data/c_probe_node_finder/1/LastGraph +668 -0
- data/spec/data/c_probe_node_finder/1/Log +756 -0
- data/spec/data/c_probe_node_finder/1/PreGraph +11 -0
- data/spec/data/c_probe_node_finder/1/Roadmaps +2009 -0
- data/spec/data/c_probe_node_finder/1/contigs.fa +29 -0
- data/spec/data/c_probe_node_finder/1/stats.txt +6 -0
- data/spec/data/contig_printer/1/HOWTO_RECREATE +17 -0
- data/spec/data/contig_printer/1/contigs.fa +4 -0
- data/spec/data/contig_printer/1/seq.fa +2408 -0
- data/spec/data/contig_printer/1/seq.fa.svg +153 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Graph2 +2953 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/LastGraph +2953 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Log +21 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/PreGraph +27 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Roadmaps +5182 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Sequences +3612 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/contigs.fa +36 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/stats.txt +14 -0
- data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam +0 -0
- data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam.bai +0 -0
- data/spec/data/contig_printer/1/seq.node12.fa +4 -0
- data/spec/data/contig_printer/1/seq1_1to550.fa +2 -0
- data/spec/data/contig_printer/1/seq2_1to550.fa +2 -0
- data/spec/data/contig_printer/1/seq2_1to550.fa.fai +1 -0
- data/spec/data/explore/1/2seqs.sammy.fa +12004 -0
- data/spec/data/explore/1/HOWTO_RECREATE.txt +6 -0
- data/spec/data/explore/1/a.fa +2 -0
- data/spec/data/explore/1/seq1_and_a.fa +3 -0
- data/spec/data/explore/1/seq2.fa +2 -0
- data/spec/data/fluff/1/2seqs.sammy.fa +12004 -0
- data/spec/data/fluff/1/HOWTO_RECREATE.txt +5 -0
- data/spec/data/fluff/1/seq1.fa +2 -0
- data/spec/data/fluff/1/seq2.fa +2 -0
- data/spec/data/gapfilling/1/reads.fa +171 -0
- data/spec/data/gapfilling/1/trail_with_Ns.fa +5 -0
- data/spec/data/gapfilling/1/velvetAssembly/Graph2 +130 -0
- data/spec/data/gapfilling/1/velvetAssembly/LastGraph +130 -0
- data/spec/data/gapfilling/1/velvetAssembly/Log +199 -0
- data/spec/data/gapfilling/1/velvetAssembly/PreGraph +7 -0
- data/spec/data/gapfilling/1/velvetAssembly/Roadmaps +239 -0
- data/spec/data/gapfilling/1/velvetAssembly/Sequences +281 -0
- data/spec/data/gapfilling/1/velvetAssembly/contigs.fa +12 -0
- data/spec/data/gapfilling/1/velvetAssembly/stats.txt +4 -0
- data/spec/data/gapfilling/2/HOWTO_recreate +17 -0
- data/spec/data/gapfilling/2/reference.fa +2 -0
- data/spec/data/gapfilling/2/reference_part1.fa +4 -0
- data/spec/data/gapfilling/2/reference_part2.fa +4 -0
- data/spec/data/gapfilling/2/sammy_reads.fa.gz +0 -0
- data/spec/data/gapfilling/2/with_gaps.fa +4 -0
- data/spec/data/gapfilling/3/HOWTO_recreate +4 -0
- data/spec/data/gapfilling/3/reads.fa.gz +0 -0
- data/spec/data/gapfilling/3/reference_part1.fa +4 -0
- data/spec/data/gapfilling/3/reference_part2.fa +4 -0
- data/spec/data/gapfilling/3/with_gaps.fa +4 -0
- data/spec/data/gapfilling/4/HOWTO_recreate +1 -0
- data/spec/data/gapfilling/4/reads.fa.gz +0 -0
- data/spec/data/gapfilling/5/HOWTO_RECREATE +7 -0
- data/spec/data/gapfilling/5/answer.fna +2 -0
- data/spec/data/gapfilling/5/gappy.fna +2 -0
- data/spec/data/gapfilling/5/reads.fa +17961 -0
- data/spec/data/gapfilling/5/velvet51_3.5/LastGraph +8337 -0
- data/spec/data/gapfilling/5/velvet51_3.5/Sequences +20921 -0
- data/spec/data/gapfilling/6/random1.fa +28 -0
- data/spec/data/gapfilling/6/random2.fa +28 -0
- data/spec/data/gapfilling/6/random_sequence_length_2000 +0 -0
- data/spec/data/gapfilling/6/reads.random1.fa.gz +0 -0
- data/spec/data/gapfilling/6/reads.random2.fa.gz +0 -0
- data/spec/data/gapfilling/6/to_gapfill.fa +22 -0
- data/spec/data/kmer_profile_to_assembly/multiple_abundance_file1.csv +2 -0
- data/spec/data/kmers_count1.csv +2 -0
- data/spec/data/kmers_count2.csv +3 -0
- data/spec/data/out +3 -0
- data/spec/data/positive_latching_pair.fa +2 -0
- data/spec/data/primers.csv +4 -0
- data/spec/data/read_selection_by_kmer/blacklist1.txt +1 -0
- data/spec/data/read_selection_by_kmer/input.fasta +6 -0
- data/spec/data/read_selection_by_kmer/whitelist1.txt +1 -0
- data/spec/data/read_selection_by_kmer/whitelist2.txt +2 -0
- data/spec/data/read_to_node/1_a_graph/HOWTO_RECREATE.txt +2 -0
- data/spec/data/read_to_node/1_a_graph/LastGraph +6695 -0
- data/spec/data/read_to_node/1_a_graph/ReadToNode.bin +0 -0
- data/spec/data/read_to_node/2_no_read256_or_259/HOWTO_RECREATE.txt +3 -0
- data/spec/data/read_to_node/2_no_read256_or_259/LastGraph +6693 -0
- data/spec/data/read_to_node/2_no_read256_or_259/ReadToNode.bin +0 -0
- data/spec/data/read_to_node/3_no_last_read/LastGraph +6694 -0
- data/spec/data/read_to_node/3_no_last_read/ReadToNode.bin +0 -0
- data/spec/data/t/details.txt +5 -0
- data/spec/data/t/details.txt.srt +5 -0
- data/spec/data/t/location.txt +3 -0
- data/spec/data/t/location.txt.srt +3 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/answer.fa +2 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/reads.fa.gz +0 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/scaffolds.fa +6 -0
- data/spec/data/tweak/2_second_genome/answer2.fa +2 -0
- data/spec/data/tweak/2_second_genome/reads.fa.gz +0 -0
- data/spec/data/tweak/3_variant/answer.fa +2 -0
- data/spec/data/tweak/3_variant/lesser_answer.fa +2 -0
- data/spec/data/tweak/3_variant/reads.fa.gz +0 -0
- data/spec/data/tweak/3_variant/with_gaps.fa +2 -0
- data/spec/data/velvet_test_trails/Assem/Graph +17 -0
- data/spec/data/velvet_test_trails/Assem/Graph2 +40 -0
- data/spec/data/velvet_test_trails/Assem/LastGraph +40 -0
- data/spec/data/velvet_test_trails/Assem/Log +35 -0
- data/spec/data/velvet_test_trails/Assem/PreGraph +9 -0
- data/spec/data/velvet_test_trails/Assem/Roadmaps +89 -0
- data/spec/data/velvet_test_trails/Assem/Sequences +50 -0
- data/spec/data/velvet_test_trails/Assem/a.svg +53 -0
- data/spec/data/velvet_test_trails/Assem/contigs.fa +15 -0
- data/spec/data/velvet_test_trails/Assem/stats.txt +5 -0
- data/spec/data/velvet_test_trails/node_fwds.fa +8 -0
- data/spec/data/velvet_test_trails/node_seqs.fa +9 -0
- data/spec/data/velvet_test_trails/nodes_fwd_rev.fa +16 -0
- data/spec/data/velvet_test_trails/read1.fa +2 -0
- data/spec/data/velvet_test_trails/reads.fa +50 -0
- data/spec/data/velvet_test_trails_reverse/Assem/LastGraph +17 -0
- data/spec/data/velvet_test_trails_reverse/Assem/a.svg +53 -0
- data/spec/data/velvet_test_trails_reverse/reads_reversed.fa +10 -0
- data/spec/data/visualise/1/LastGraph +6695 -0
- data/spec/data/visualise/2_paired_end/HOWTO_RECREATE.txt +10 -0
- data/spec/data/visualise/2_paired_end/rand1.fa +2 -0
- data/spec/data/visualise/2_paired_end/rand2.fa +2 -0
- data/spec/data/visualise/2_paired_end/with_gaps.fa +8 -0
- data/spec/data/visualise/2_paired_end/with_gaps.read_pairs.fa.gz +0 -0
- data/spec/data/wander/1/random1.fa +2 -0
- data/spec/data/wander/1/random1.sammy.fa +804 -0
- data/spec/depth_first_search_spec.rb +190 -0
- data/spec/dijkstra_spec.rb +143 -0
- data/spec/explore_spec.rb +29 -0
- data/spec/fluffer_spec.rb +155 -0
- data/spec/gapfiller_spec.rb +107 -0
- data/spec/graph_explorer_spec.rb +475 -0
- data/spec/graph_generator_spec.rb +99 -0
- data/spec/height_finder_spec.rb +306 -0
- data/spec/kmer_abundance_pattern_spec.rb +56 -0
- data/spec/kmer_coverage_based_path_filter_spec.rb +73 -0
- data/spec/kmer_profile_finder_spec.rb +38 -0
- data/spec/kmers_count_tabulate_spec.rb +120 -0
- data/spec/oriented_node_trail_spec.rb +221 -0
- data/spec/paired_end_neighbours_spec.rb +126 -0
- data/spec/paths_between_nodes_spec.rb +349 -0
- data/spec/priner_spec.rb +7 -0
- data/spec/read_input_spec.rb +23 -0
- data/spec/read_selection_by_kmer_spec.rb +166 -0
- data/spec/read_to_node_spec.rb +35 -0
- data/spec/roundup_spec.rb +366 -0
- data/spec/scaffold_breaker_spec.rb +144 -0
- data/spec/sequence_spec.rb +43 -0
- data/spec/single_coherent_paths_between_nodes_spec.rb +492 -0
- data/spec/single_coherent_wanderer_spec.rb +120 -0
- data/spec/single_ended_assembler_spec.rb +398 -0
- data/spec/spec_helper.rb +310 -0
- data/spec/velvet_graph_sequence_extractor_spec.rb +80 -0
- data/spec/visualise_spec.rb +105 -0
- data/spec/wander_spec.rb +119 -0
- data/spec/watch_for_changes.sh +16 -0
- data/validation/fasta_compare.rb +72 -0
- data/validation/gapfill_simulate_perfect.rb +108 -0
- metadata +899 -0
@@ -0,0 +1,72 @@
|
|
1
|
+
/*
|
2
|
+
Copyright 2007, 2008 Daniel Zerbino (zerbino@ebi.ac.uk)
|
3
|
+
|
4
|
+
This file is part of Velvet.
|
5
|
+
|
6
|
+
Velvet is free software; you can redistribute it and/or modify
|
7
|
+
it under the terms of the GNU General Public License as published by
|
8
|
+
the Free Software Foundation; either version 2 of the License, or
|
9
|
+
(at your option) any later version.
|
10
|
+
|
11
|
+
Velvet is distributed in the hope that it will be useful,
|
12
|
+
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
13
|
+
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
14
|
+
GNU General Public License for more details.
|
15
|
+
|
16
|
+
You should have received a copy of the GNU General Public License
|
17
|
+
along with Velvet; if not, write to the Free Software
|
18
|
+
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
19
|
+
|
20
|
+
*/
|
21
|
+
#ifndef _GRAPHSTATS_H_
|
22
|
+
#define _GRAPHSTATS_H_
|
23
|
+
|
24
|
+
|
25
|
+
void displayGeneralStatistics(Graph * graph, char *filename, ReadSet * reads);
|
26
|
+
|
27
|
+
void exportLongNodeSequences(char *filename, Graph * graph,
|
28
|
+
Coordinate minLength, ReadSet * reads, ShortLength * readLengths, IDnum minCov);
|
29
|
+
|
30
|
+
Coordinate readCoverage(Node * node);
|
31
|
+
|
32
|
+
Coordinate n50(Graph * graph);
|
33
|
+
|
34
|
+
double estimated_cov(Graph * graph, char * directory);
|
35
|
+
|
36
|
+
Coordinate maxLength(Graph * graph);
|
37
|
+
|
38
|
+
boolean *removeLowCoverageNodesAndDenounceDubiousReads(Graph * graph,
|
39
|
+
double minCov,
|
40
|
+
ReadSet * reads,
|
41
|
+
boolean export,
|
42
|
+
Coordinate minLength,
|
43
|
+
char *filename);
|
44
|
+
|
45
|
+
void removeLowLongCoverageNodesAndDenounceDubiousReads(Graph * graph,
|
46
|
+
double minCov,
|
47
|
+
ReadSet * reads,
|
48
|
+
boolean * dubious,
|
49
|
+
boolean export,
|
50
|
+
Coordinate minLength,
|
51
|
+
char *filename);
|
52
|
+
|
53
|
+
void removeLowCoverageReferenceNodes(Graph * graph, double minCov, double minLongCov, ReadSet * reads);
|
54
|
+
|
55
|
+
void exportAMOSContigs(char *filename, Graph * graph,
|
56
|
+
Coordinate cutoff_length, ReadSet * reads);
|
57
|
+
|
58
|
+
IDnum usedReads(Graph * graph, Coordinate minContigLength);
|
59
|
+
|
60
|
+
Coordinate totalAssemblyLength(Graph * graph);
|
61
|
+
|
62
|
+
void logFinalStats(Graph * graph, Coordinate minContigKmerLength, char *directory);
|
63
|
+
|
64
|
+
void exportUnusedReads(Graph* graph, ReadSet * reads, Coordinate minContigKmerLength, char* filename);
|
65
|
+
|
66
|
+
void exportLongNodeMappings(char *filename, Graph * graph, ReadSet * reads,
|
67
|
+
Coordinate minLength, SequencesReader *seqReadInfo);
|
68
|
+
|
69
|
+
void removeHighCoverageNodes(Graph * graph, double maxCov, boolean export, Coordinate minLength, char * filename);
|
70
|
+
|
71
|
+
void removeLowArcs(Graph * graph, double cutoff);
|
72
|
+
#endif
|
@@ -0,0 +1,52 @@
|
|
1
|
+
struct arc_st {
|
2
|
+
Arc *twinArc; // 64
|
3
|
+
Arc *next; // 64
|
4
|
+
Arc *previous; // 64
|
5
|
+
Arc *nextInLookupTable; // 64
|
6
|
+
Node *destination; // 64
|
7
|
+
IDnum multiplicity; // 32
|
8
|
+
} ATTRIBUTE_PACKED; // 352 Total
|
9
|
+
|
10
|
+
struct node_st {
|
11
|
+
Node *twinNode; // 64
|
12
|
+
Arc *arc; // 64
|
13
|
+
Descriptor *descriptor; // 64
|
14
|
+
PassageMarkerI marker; // 32
|
15
|
+
IDnum length; // 32
|
16
|
+
#ifndef SINGLE_COV_CAT
|
17
|
+
IDnum virtualCoverage[CATEGORIES]; // 32 * 2
|
18
|
+
IDnum originalVirtualCoverage[CATEGORIES]; // 32 * 2
|
19
|
+
#else
|
20
|
+
IDnum virtualCoverage; // 32 * 2
|
21
|
+
#endif
|
22
|
+
IDnum ID; // 32
|
23
|
+
IDnum arcCount; // 32
|
24
|
+
boolean status; // 1
|
25
|
+
boolean uniqueness; // 1
|
26
|
+
} ATTRIBUTE_PACKED; // 418 Total
|
27
|
+
|
28
|
+
struct shortReadMarker_st {
|
29
|
+
IDnum position;
|
30
|
+
IDnum readID;
|
31
|
+
ShortLength offset;
|
32
|
+
} ATTRIBUTE_PACKED;
|
33
|
+
|
34
|
+
struct gapMarker_st {
|
35
|
+
GapMarker *next;
|
36
|
+
IDnum position;
|
37
|
+
IDnum length;
|
38
|
+
} ATTRIBUTE_PACKED;
|
39
|
+
|
40
|
+
struct graph_st {
|
41
|
+
Node **nodes;
|
42
|
+
Arc **arcLookupTable;
|
43
|
+
ShortReadMarker **nodeReads;
|
44
|
+
IDnum *nodeReadCounts;
|
45
|
+
GapMarker **gapMarkers;
|
46
|
+
Coordinate insertLengths[CATEGORIES + 1];
|
47
|
+
double insertLengths_var[CATEGORIES + 1];
|
48
|
+
IDnum sequenceCount;
|
49
|
+
IDnum nodeCount;
|
50
|
+
int wordLength;
|
51
|
+
boolean double_stranded;
|
52
|
+
};
|
data/ext/src/src/kmer.c
ADDED
@@ -0,0 +1,652 @@
|
|
1
|
+
/*
|
2
|
+
Copyright 2007, 2008, 2009 Daniel Zerbino (zerbino@ebi.ac.uk)
|
3
|
+
|
4
|
+
This file is part of Velvet.
|
5
|
+
|
6
|
+
Velvet is free software; you can redistribute it and/or modify
|
7
|
+
it under the terms of the GNU General Public License as published by
|
8
|
+
the Free Software Foundation; either version 2 of the License, or
|
9
|
+
(at your option) any later version.
|
10
|
+
|
11
|
+
Velvet is distributed in the hope that it will be useful,
|
12
|
+
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
13
|
+
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
14
|
+
GNU General Public License for more details.
|
15
|
+
|
16
|
+
You should have received a copy of the GNU General Public License
|
17
|
+
along with Velvet; if not, write to the Free Software
|
18
|
+
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
19
|
+
|
20
|
+
*/
|
21
|
+
#include <stdlib.h>
|
22
|
+
#include <stdio.h>
|
23
|
+
|
24
|
+
#include "globals.h"
|
25
|
+
#include "kmer.h"
|
26
|
+
#include "utility.h"
|
27
|
+
|
28
|
+
static const uint64_t longLongLeftFilter = (uint64_t) 3 << 62;
|
29
|
+
static const uint32_t longLeftFilter = (uint32_t) 3 << 30;
|
30
|
+
static const uint16_t intLeftFilter = (uint16_t) 3 << 14;
|
31
|
+
static const uint8_t charLeftFilter = (uint8_t) 3 << 6;
|
32
|
+
|
33
|
+
#if KMER_LONGLONGS
|
34
|
+
static uint64_t longLongWordFilter = (uint64_t) ((int64_t) -1);
|
35
|
+
#endif
|
36
|
+
static uint32_t longWordFilter = (uint32_t) ((int32_t) -1);
|
37
|
+
static uint16_t intWordFilter = (uint16_t) ((int16_t) -1);
|
38
|
+
static uint8_t charWordFilter = (uint8_t) ((int8_t) -1);
|
39
|
+
|
40
|
+
#define UNDEFINED 0
|
41
|
+
#define CHARS 1
|
42
|
+
#define INTS 2
|
43
|
+
#define LONGS 3
|
44
|
+
#define LONGLONGS 4
|
45
|
+
static int kmerFilterIndex = UNDEFINED;
|
46
|
+
static int kmerFilterOffset = 0;
|
47
|
+
static int kmerFilterLength = 0;
|
48
|
+
#if KMER_LONGLONGS
|
49
|
+
static int longLongKmerFilterIndex = KMER_LONGLONGS - 1;
|
50
|
+
static uint64_t longLongKmerFilter = (uint64_t) ((int64_t) -1);
|
51
|
+
#endif
|
52
|
+
|
53
|
+
static uint64_t keyFilter = 0;
|
54
|
+
static int keyFilterIndex = UNDEFINED;
|
55
|
+
static int keyFilterOffset = 0;
|
56
|
+
static int keyFilterLength = 0;
|
57
|
+
#if KMER_LONGLONGS
|
58
|
+
static int longLongKeyFilterIndex = KMER_LONGLONGS - 1;
|
59
|
+
#endif
|
60
|
+
|
61
|
+
void resetWordFilter(int wordLength) {
|
62
|
+
int kmer_bit_size = wordLength * 2;
|
63
|
+
|
64
|
+
if (wordLength > MAXKMERLENGTH)
|
65
|
+
exitErrorf(EXIT_FAILURE, true, "Word length %i greater than max allowed value (%i).\nRecompile Velvet to deal with this word length.", wordLength, MAXKMERLENGTH);
|
66
|
+
|
67
|
+
#if KMER_LONGLONGS
|
68
|
+
int i;
|
69
|
+
for (i = 0; i < KMER_LONGLONGS; i++) {
|
70
|
+
if (kmer_bit_size > 64) {
|
71
|
+
kmer_bit_size -= 64;
|
72
|
+
continue;
|
73
|
+
} else if (kmer_bit_size == 64) {
|
74
|
+
longLongKmerFilterIndex = i;
|
75
|
+
longLongKmerFilter = longLongWordFilter;
|
76
|
+
kmerFilterIndex = LONGLONGS;
|
77
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
78
|
+
kmerFilterLength = kmer_bit_size;
|
79
|
+
longWordFilter = 0;
|
80
|
+
intWordFilter = 0;
|
81
|
+
charWordFilter = 0;
|
82
|
+
return;
|
83
|
+
} else {
|
84
|
+
longLongKmerFilterIndex = i;
|
85
|
+
longLongKmerFilter = (((uint64_t) 1) << kmer_bit_size) - 1;
|
86
|
+
kmerFilterIndex = LONGLONGS;
|
87
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
88
|
+
kmerFilterLength = kmer_bit_size;
|
89
|
+
longWordFilter = 0;
|
90
|
+
intWordFilter = 0;
|
91
|
+
charWordFilter = 0;
|
92
|
+
return;
|
93
|
+
}
|
94
|
+
}
|
95
|
+
#endif
|
96
|
+
#if KMER_LONGS
|
97
|
+
if (kmer_bit_size > 32)
|
98
|
+
kmer_bit_size -= 32;
|
99
|
+
else if (kmer_bit_size == 32) {
|
100
|
+
kmerFilterIndex = LONGS;
|
101
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
102
|
+
kmerFilterLength = kmer_bit_size;
|
103
|
+
intWordFilter = 0;
|
104
|
+
charWordFilter = 0;
|
105
|
+
return;
|
106
|
+
} else {
|
107
|
+
longWordFilter = (((uint32_t) 1) << kmer_bit_size) - 1;
|
108
|
+
kmerFilterIndex = LONGS;
|
109
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
110
|
+
kmerFilterLength = kmer_bit_size;
|
111
|
+
intWordFilter = 0;
|
112
|
+
charWordFilter = 0;
|
113
|
+
return;
|
114
|
+
}
|
115
|
+
#endif
|
116
|
+
#if KMER_INTS
|
117
|
+
if (kmer_bit_size > 16)
|
118
|
+
kmer_bit_size -= 16;
|
119
|
+
else if (kmer_bit_size == 16) {
|
120
|
+
kmerFilterIndex = INTS;
|
121
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
122
|
+
kmerFilterLength = kmer_bit_size;
|
123
|
+
charWordFilter = 0;
|
124
|
+
return;
|
125
|
+
} else {
|
126
|
+
intWordFilter = (((uint16_t) 1) << kmer_bit_size) - 1;
|
127
|
+
kmerFilterIndex = INTS;
|
128
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
129
|
+
kmerFilterLength = kmer_bit_size;
|
130
|
+
charWordFilter = 0;
|
131
|
+
return;
|
132
|
+
}
|
133
|
+
|
134
|
+
#endif
|
135
|
+
#if KMER_CHARS
|
136
|
+
if (kmer_bit_size < 8)
|
137
|
+
charWordFilter = (((uint8_t) 1) << kmer_bit_size) - 1;
|
138
|
+
|
139
|
+
kmerFilterIndex = CHARS;
|
140
|
+
kmerFilterOffset = kmer_bit_size - 2;
|
141
|
+
kmerFilterLength = kmer_bit_size;
|
142
|
+
#endif
|
143
|
+
|
144
|
+
}
|
145
|
+
|
146
|
+
void resetKeyFilter(int keyLength) {
|
147
|
+
int kmer_bit_size = keyLength;
|
148
|
+
|
149
|
+
if (keyLength > MAXKMERLENGTH)
|
150
|
+
exitErrorf(EXIT_FAILURE, true, "Key length %i greater than max allowed value (%i).\nRecompile Velvet to deal with this word length.", keyLength, MAXKMERLENGTH);
|
151
|
+
|
152
|
+
#if KMER_CHARS
|
153
|
+
if (kmerFilterIndex == CHARS) {
|
154
|
+
if (kmer_bit_size > kmerFilterLength) {
|
155
|
+
kmer_bit_size -= kmerFilterLength;
|
156
|
+
} else {
|
157
|
+
keyFilterIndex = CHARS;
|
158
|
+
keyFilterLength = kmer_bit_size;
|
159
|
+
keyFilterOffset = kmerFilterLength - keyFilterLength;
|
160
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
161
|
+
return;
|
162
|
+
}
|
163
|
+
}
|
164
|
+
#endif
|
165
|
+
#if KMER_INTS
|
166
|
+
if (kmerFilterIndex == INTS) {
|
167
|
+
if (kmer_bit_size > kmerFilterLength) {
|
168
|
+
kmer_bit_size -= kmerFilterLength;
|
169
|
+
} else {
|
170
|
+
keyFilterIndex = INTS;
|
171
|
+
keyFilterLength = kmer_bit_size;
|
172
|
+
keyFilterOffset = kmerFilterLength - keyFilterLength;
|
173
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
174
|
+
return;
|
175
|
+
}
|
176
|
+
} else if (intWordFilter) {
|
177
|
+
if (kmer_bit_size > 16)
|
178
|
+
kmer_bit_size -= 16;
|
179
|
+
else {
|
180
|
+
keyFilterIndex = INTS;
|
181
|
+
keyFilterLength = kmer_bit_size;
|
182
|
+
keyFilterOffset = 16 - keyFilterLength;
|
183
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
184
|
+
return;
|
185
|
+
}
|
186
|
+
}
|
187
|
+
#endif
|
188
|
+
#if KMER_LONGS
|
189
|
+
if (kmerFilterIndex == LONGS) {
|
190
|
+
if (kmer_bit_size > kmerFilterLength) {
|
191
|
+
kmer_bit_size -= kmerFilterLength;
|
192
|
+
} else {
|
193
|
+
keyFilterIndex = LONGS;
|
194
|
+
keyFilterLength = kmer_bit_size;
|
195
|
+
keyFilterOffset = kmerFilterLength - keyFilterLength;
|
196
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
197
|
+
return;
|
198
|
+
}
|
199
|
+
} else if (longWordFilter) {
|
200
|
+
if (kmer_bit_size > 32)
|
201
|
+
kmer_bit_size -= 32;
|
202
|
+
else {
|
203
|
+
keyFilterIndex = LONGS;
|
204
|
+
keyFilterLength = kmer_bit_size;
|
205
|
+
keyFilterOffset = 32 - keyFilterLength;
|
206
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
207
|
+
return;
|
208
|
+
}
|
209
|
+
}
|
210
|
+
#endif
|
211
|
+
#if KMER_LONGLONGS
|
212
|
+
if (kmerFilterIndex == LONGLONGS) {
|
213
|
+
if (kmer_bit_size > kmerFilterLength) {
|
214
|
+
kmer_bit_size -= kmerFilterLength;
|
215
|
+
} else {
|
216
|
+
keyFilterIndex = LONGLONGS;
|
217
|
+
longLongKeyFilterIndex = longLongKmerFilterIndex;
|
218
|
+
keyFilterLength = kmer_bit_size;
|
219
|
+
keyFilterOffset = kmerFilterLength - keyFilterLength;
|
220
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
221
|
+
return;
|
222
|
+
}
|
223
|
+
} else {
|
224
|
+
if (kmer_bit_size > 64)
|
225
|
+
abort();
|
226
|
+
else {
|
227
|
+
keyFilterIndex = LONGLONGS;
|
228
|
+
longLongKeyFilterIndex = longLongKmerFilterIndex;
|
229
|
+
keyFilterLength = kmer_bit_size;
|
230
|
+
keyFilterOffset = 64 - keyFilterLength;
|
231
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
232
|
+
return;
|
233
|
+
}
|
234
|
+
}
|
235
|
+
|
236
|
+
if (kmer_bit_size > 64)
|
237
|
+
abort();
|
238
|
+
else {
|
239
|
+
keyFilterIndex = LONGLONGS;
|
240
|
+
longLongKeyFilterIndex = longLongKmerFilterIndex - 1;
|
241
|
+
keyFilterLength = kmer_bit_size;
|
242
|
+
keyFilterOffset = 64 - keyFilterLength;
|
243
|
+
keyFilter = ((((uint64_t) 1) << keyFilterLength) - 1) << keyFilterOffset;
|
244
|
+
return;
|
245
|
+
}
|
246
|
+
#endif
|
247
|
+
|
248
|
+
}
|
249
|
+
|
250
|
+
static void shiftRight(Kmer * kmer) {
|
251
|
+
uint64_t leftBits = 0;
|
252
|
+
uint64_t rightBits;
|
253
|
+
|
254
|
+
#if KMER_CHARS
|
255
|
+
|
256
|
+
#if KMER_INTS | KMER_LONGS | KMER_LONGLONGS
|
257
|
+
rightBits = kmer->chars & 3;
|
258
|
+
#endif
|
259
|
+
|
260
|
+
kmer->chars >>= 2;
|
261
|
+
kmer->chars += (uint8_t) leftBits;
|
262
|
+
|
263
|
+
#if KMER_INTS | KMER_LONGS | KMER_LONGLONGS
|
264
|
+
leftBits = rightBits;
|
265
|
+
#endif
|
266
|
+
#endif
|
267
|
+
|
268
|
+
#if KMER_INTS
|
269
|
+
|
270
|
+
#if KMER_LONGS | KMER_LONGLONGS
|
271
|
+
rightBits = kmer->ints & 3;
|
272
|
+
#endif
|
273
|
+
|
274
|
+
leftBits <<= 14;
|
275
|
+
kmer->ints >>= 2;
|
276
|
+
kmer->ints += (uint16_t) leftBits;
|
277
|
+
|
278
|
+
#if KMER_LONGS | KMER_LONGLONGS
|
279
|
+
leftBits = rightBits;
|
280
|
+
#endif
|
281
|
+
#endif
|
282
|
+
|
283
|
+
#if KMER_LONGS
|
284
|
+
|
285
|
+
#if KMER_LONGLONGS
|
286
|
+
rightBits = kmer->longs & 3;
|
287
|
+
#endif
|
288
|
+
|
289
|
+
leftBits <<= 30;
|
290
|
+
kmer->longs >>= 2;
|
291
|
+
kmer->longs += (uint32_t) leftBits;
|
292
|
+
|
293
|
+
#if KMER_LONGLONGS
|
294
|
+
leftBits = rightBits;
|
295
|
+
#endif
|
296
|
+
#endif
|
297
|
+
|
298
|
+
#if KMER_LONGLONGS
|
299
|
+
int i;
|
300
|
+
for (i = KMER_LONGLONGS - 1; i >= 0; i--) {
|
301
|
+
rightBits = kmer->longlongs[i] & 3;
|
302
|
+
leftBits <<= 62;
|
303
|
+
kmer->longlongs[i] >>= 2;
|
304
|
+
kmer->longlongs[i] += leftBits;
|
305
|
+
leftBits = rightBits;
|
306
|
+
}
|
307
|
+
#endif
|
308
|
+
}
|
309
|
+
|
310
|
+
void copyKmers(Kmer* k1, Kmer* k2) {
|
311
|
+
|
312
|
+
#if KMER_LONGLONGS
|
313
|
+
int i;
|
314
|
+
for (i = 0; i < KMER_LONGLONGS; i++)
|
315
|
+
k1->longlongs[i] = k2->longlongs[i];
|
316
|
+
#endif
|
317
|
+
#if KMER_LONGS
|
318
|
+
k1->longs = k2->longs;
|
319
|
+
#endif
|
320
|
+
#if KMER_INTS
|
321
|
+
k1->ints = k2->ints;
|
322
|
+
#endif
|
323
|
+
#if KMER_CHARS
|
324
|
+
k1->chars = k2->chars;
|
325
|
+
#endif
|
326
|
+
}
|
327
|
+
|
328
|
+
int compareKmers(Kmer* k1, Kmer* k2) {
|
329
|
+
#if KMER_LONGLONGS
|
330
|
+
int i;
|
331
|
+
#endif
|
332
|
+
|
333
|
+
#if KMER_CHARS
|
334
|
+
if (k1->chars == k2->chars)
|
335
|
+
;
|
336
|
+
else if (k1->chars > k2->chars)
|
337
|
+
return 1;
|
338
|
+
else
|
339
|
+
return -1;
|
340
|
+
#endif
|
341
|
+
#if KMER_INTS
|
342
|
+
if (k1->ints == k2->ints)
|
343
|
+
;
|
344
|
+
else if (k1->ints > k2->ints)
|
345
|
+
return 1;
|
346
|
+
else
|
347
|
+
return -1;
|
348
|
+
#endif
|
349
|
+
#if KMER_LONGS
|
350
|
+
if (k1->longs == k2->longs)
|
351
|
+
;
|
352
|
+
else if (k1->longs > k2->longs)
|
353
|
+
return 1;
|
354
|
+
else
|
355
|
+
return -1;
|
356
|
+
#endif
|
357
|
+
#if KMER_LONGLONGS
|
358
|
+
for (i = KMER_LONGLONGS - 1; i >= 0; i--) {
|
359
|
+
if (k1->longlongs[i] == k2->longlongs[i])
|
360
|
+
continue;
|
361
|
+
else if (k1->longlongs[i] > k2->longlongs[i])
|
362
|
+
return 1;
|
363
|
+
else
|
364
|
+
return -1;
|
365
|
+
}
|
366
|
+
#endif
|
367
|
+
|
368
|
+
return 0;
|
369
|
+
}
|
370
|
+
|
371
|
+
void clearKmer(Kmer * kmer) {
|
372
|
+
#if KMER_LONGLONGS
|
373
|
+
int i;
|
374
|
+
for (i = 0; i < KMER_LONGLONGS; i++)
|
375
|
+
kmer->longlongs[i] = 0;
|
376
|
+
#endif
|
377
|
+
#if KMER_LONGS
|
378
|
+
kmer->longs = 0;
|
379
|
+
#endif
|
380
|
+
#if KMER_INTS
|
381
|
+
kmer->ints = 0;
|
382
|
+
#endif
|
383
|
+
#if KMER_CHARS
|
384
|
+
kmer->chars = 0;
|
385
|
+
#endif
|
386
|
+
}
|
387
|
+
|
388
|
+
void printKmer(Kmer * kmer) {
|
389
|
+
|
390
|
+
#if KMER_CHARS
|
391
|
+
velvetLog("%hx\t", kmer->chars);
|
392
|
+
#endif
|
393
|
+
#if KMER_INTS
|
394
|
+
velvetLog("%x\t", kmer->ints);
|
395
|
+
#endif
|
396
|
+
#if KMER_LONGS
|
397
|
+
velvetLog("%x\t", kmer->longs);
|
398
|
+
#endif
|
399
|
+
#if KMER_LONGLONGS
|
400
|
+
int i;
|
401
|
+
for (i = KMER_LONGLONGS - 1; i >= 0; i--)
|
402
|
+
velvetLog("%llx\t", (long long) kmer->longlongs[i]);
|
403
|
+
#endif
|
404
|
+
velvetLog("\n");
|
405
|
+
}
|
406
|
+
|
407
|
+
void pushNucleotide(Kmer * kmer, Nucleotide nucleotide) {
|
408
|
+
|
409
|
+
#if KMER_LONGLONGS
|
410
|
+
uint64_t * ptr;
|
411
|
+
int i;
|
412
|
+
#endif
|
413
|
+
#if KMER_LONGLONGS > 1 | KMER_LONGS | KMER_INTS | KMER_CHARS
|
414
|
+
uint64_t leftBits;
|
415
|
+
#endif
|
416
|
+
uint64_t rightBits = 0;
|
417
|
+
|
418
|
+
#if KMER_LONGLONGS
|
419
|
+
ptr = kmer->longlongs;
|
420
|
+
|
421
|
+
#if KMER_LONGLONGS > 1
|
422
|
+
for (i = 0; i < longLongKmerFilterIndex; i++) {
|
423
|
+
leftBits = (*ptr & longLongLeftFilter);
|
424
|
+
leftBits >>= 62;
|
425
|
+
*ptr <<= 2;
|
426
|
+
*ptr += rightBits;
|
427
|
+
*ptr &= longLongWordFilter;
|
428
|
+
rightBits = leftBits;
|
429
|
+
ptr++;
|
430
|
+
}
|
431
|
+
#endif
|
432
|
+
|
433
|
+
#if KMER_LONGS | KMER_INTS | KMER_CHARS
|
434
|
+
leftBits = (*ptr & longLongLeftFilter);
|
435
|
+
leftBits >>= 62;
|
436
|
+
#endif
|
437
|
+
|
438
|
+
*ptr <<= 2;
|
439
|
+
*ptr += rightBits;
|
440
|
+
*ptr &= longLongKmerFilter;
|
441
|
+
|
442
|
+
#if KMER_LONGS | KMER_INTS | KMER_CHARS
|
443
|
+
rightBits = leftBits;
|
444
|
+
#endif
|
445
|
+
#endif
|
446
|
+
|
447
|
+
#if KMER_LONGS
|
448
|
+
|
449
|
+
#if KMER_INTS | KMER_CHARS
|
450
|
+
leftBits = kmer->longs & longLeftFilter;
|
451
|
+
leftBits >>= 30;
|
452
|
+
#endif
|
453
|
+
kmer->longs <<= 2;
|
454
|
+
kmer->longs += rightBits;
|
455
|
+
kmer->longs &= longWordFilter;
|
456
|
+
|
457
|
+
#if KMER_INTS | KMER_CHARS
|
458
|
+
rightBits = leftBits;
|
459
|
+
#endif
|
460
|
+
|
461
|
+
#endif
|
462
|
+
|
463
|
+
#if KMER_INTS
|
464
|
+
|
465
|
+
#if KMER_CHARS
|
466
|
+
leftBits = kmer->ints & intLeftFilter;
|
467
|
+
leftBits >>= 14;
|
468
|
+
#endif
|
469
|
+
kmer->ints <<= 2;
|
470
|
+
kmer->ints += rightBits;
|
471
|
+
kmer->ints &= intWordFilter;
|
472
|
+
|
473
|
+
#if KMER_CHARS
|
474
|
+
rightBits = leftBits;
|
475
|
+
#endif
|
476
|
+
|
477
|
+
#endif
|
478
|
+
|
479
|
+
#if KMER_CHARS
|
480
|
+
kmer->chars <<= 2;
|
481
|
+
kmer->chars += rightBits;
|
482
|
+
kmer->chars &= charWordFilter;
|
483
|
+
#endif
|
484
|
+
|
485
|
+
#if KMER_LONGLONGS
|
486
|
+
kmer->longlongs[0] += nucleotide;
|
487
|
+
if (kmer->longlongs[0] >= nucleotide)
|
488
|
+
return;
|
489
|
+
|
490
|
+
for (i = 1; i < KMER_LONGLONGS; i++)
|
491
|
+
if (++kmer->longlongs[i])
|
492
|
+
return;
|
493
|
+
#if KMER_LONGS
|
494
|
+
if (++kmer->longs)
|
495
|
+
return;
|
496
|
+
#endif
|
497
|
+
#if KMER_INTS
|
498
|
+
if (++kmer->ints)
|
499
|
+
return;
|
500
|
+
#endif
|
501
|
+
#if KMER_CHARS
|
502
|
+
++kmer->chars;
|
503
|
+
#endif
|
504
|
+
|
505
|
+
#else
|
506
|
+
|
507
|
+
#if KMER_LONGS
|
508
|
+
kmer->longs += nucleotide;
|
509
|
+
if (kmer->longs >= nucleotide)
|
510
|
+
return;
|
511
|
+
#if KMER_INTS
|
512
|
+
if (++kmer->ints)
|
513
|
+
return;
|
514
|
+
#endif
|
515
|
+
#if KMER_CHARS
|
516
|
+
++kmer->chars;
|
517
|
+
#endif
|
518
|
+
|
519
|
+
#else
|
520
|
+
|
521
|
+
#if KMER_INTS
|
522
|
+
kmer->ints += nucleotide;
|
523
|
+
if (kmer->ints >= nucleotide)
|
524
|
+
return;
|
525
|
+
#if KMER_CHARS
|
526
|
+
++kmer->chars;
|
527
|
+
#endif
|
528
|
+
|
529
|
+
#else
|
530
|
+
|
531
|
+
#if KMER_CHARS
|
532
|
+
kmer->chars += nucleotide;
|
533
|
+
#endif
|
534
|
+
|
535
|
+
#endif
|
536
|
+
#endif
|
537
|
+
#endif
|
538
|
+
}
|
539
|
+
|
540
|
+
Nucleotide popNucleotide(Kmer * kmer) {
|
541
|
+
Nucleotide nucl;
|
542
|
+
|
543
|
+
#if KMER_LONGLONGS
|
544
|
+
nucl = kmer->longlongs[0] & 3;
|
545
|
+
#elif KMER_LONGS
|
546
|
+
nucl = kmer->longs & 3;
|
547
|
+
#elif KMER_INTS
|
548
|
+
nucl = kmer->ints & 3;
|
549
|
+
#elif KMER_CHARS
|
550
|
+
nucl = kmer->chars & 3;
|
551
|
+
#endif
|
552
|
+
|
553
|
+
shiftRight(kmer);
|
554
|
+
return nucl;
|
555
|
+
}
|
556
|
+
|
557
|
+
void reversePushNucleotide(Kmer * kmer, Nucleotide nucleotide) {
|
558
|
+
uint64_t templongLong = nucleotide;
|
559
|
+
|
560
|
+
shiftRight(kmer);
|
561
|
+
|
562
|
+
switch(kmerFilterIndex) {
|
563
|
+
case UNDEFINED:
|
564
|
+
abort();
|
565
|
+
#if KMER_LONGLONGS
|
566
|
+
case LONGLONGS:
|
567
|
+
kmer->longlongs[longLongKmerFilterIndex] += templongLong << kmerFilterOffset;
|
568
|
+
return;
|
569
|
+
#endif
|
570
|
+
#if KMER_LONGS
|
571
|
+
case LONGS:
|
572
|
+
kmer->longs += templongLong << kmerFilterOffset;
|
573
|
+
return;
|
574
|
+
#endif
|
575
|
+
#if KMER_INTS
|
576
|
+
case INTS:
|
577
|
+
kmer->ints += templongLong << kmerFilterOffset;
|
578
|
+
return;
|
579
|
+
#endif
|
580
|
+
#if KMER_CHARS
|
581
|
+
case CHARS:
|
582
|
+
kmer->chars += templongLong << kmerFilterOffset;
|
583
|
+
return;
|
584
|
+
#endif
|
585
|
+
}
|
586
|
+
|
587
|
+
exitErrorf(EXIT_FAILURE, true, "Anomaly in k-mer filering");
|
588
|
+
}
|
589
|
+
|
590
|
+
KmerKey getKmerKey(Kmer * kmer) {
|
591
|
+
KmerKey key = 0;
|
592
|
+
|
593
|
+
#if KMER_CHARS
|
594
|
+
if (keyFilterIndex == CHARS) {
|
595
|
+
key = kmer->chars & keyFilter;
|
596
|
+
key >>= keyFilterOffset;
|
597
|
+
return key;
|
598
|
+
}
|
599
|
+
#endif
|
600
|
+
|
601
|
+
#if KMER_INTS
|
602
|
+
if (keyFilterIndex == INTS) {
|
603
|
+
#if KMER_CHARS
|
604
|
+
key += kmer->chars;
|
605
|
+
#endif
|
606
|
+
key <<= keyFilterLength;
|
607
|
+
key += (kmer->ints & keyFilter) >> keyFilterOffset;
|
608
|
+
return key;
|
609
|
+
}
|
610
|
+
#endif
|
611
|
+
|
612
|
+
#if KMER_LONGS
|
613
|
+
if (keyFilterIndex == LONGS) {
|
614
|
+
#if KMER_CHARS
|
615
|
+
key += kmer->chars;
|
616
|
+
#endif
|
617
|
+
#if KMER_INTS
|
618
|
+
key <<= 16;
|
619
|
+
key += kmer->ints;
|
620
|
+
#endif
|
621
|
+
key <<= keyFilterLength;
|
622
|
+
key += (kmer->longs & keyFilter) >> keyFilterOffset;
|
623
|
+
return key;
|
624
|
+
}
|
625
|
+
#endif
|
626
|
+
|
627
|
+
#if KMER_LONGLONGS
|
628
|
+
if (keyFilterIndex == LONGLONGS) {
|
629
|
+
#if KMER_CHARS
|
630
|
+
key += kmer->chars;
|
631
|
+
#endif
|
632
|
+
#if KMER_INTS
|
633
|
+
key <<= 16;
|
634
|
+
key += kmer->ints;
|
635
|
+
#endif
|
636
|
+
#if KMER_LONGS
|
637
|
+
key <<= 32;
|
638
|
+
key += kmer->longs;
|
639
|
+
#endif
|
640
|
+
if (longLongKeyFilterIndex != longLongKmerFilterIndex) {
|
641
|
+
key <<= kmerFilterLength;
|
642
|
+
key += kmer->longlongs[longLongKmerFilterIndex];
|
643
|
+
}
|
644
|
+
key <<= keyFilterLength;
|
645
|
+
key += (kmer->longlongs[longLongKeyFilterIndex] & keyFilter) >> keyFilterOffset;
|
646
|
+
|
647
|
+
return key;
|
648
|
+
}
|
649
|
+
#endif
|
650
|
+
abort();
|
651
|
+
return 0;
|
652
|
+
}
|