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 +1,507 @@
|
|
1
|
+
|
2
|
+
module Bio
|
3
|
+
module Velvet
|
4
|
+
class Graph
|
5
|
+
# Return an Array of OrientedNode objects corresponding to all the nodes that are next
|
6
|
+
# in the graph.
|
7
|
+
def neighbours_of(node, first_side)
|
8
|
+
neighbour_nodes = nil
|
9
|
+
if first_side == OrientedNodeTrail::START_IS_FIRST
|
10
|
+
neighbour_nodes = neighbours_off_end(node)
|
11
|
+
else
|
12
|
+
neighbour_nodes = neighbours_into_start(node)
|
13
|
+
end
|
14
|
+
|
15
|
+
neighbours_with_orientation = []
|
16
|
+
neighbour_nodes.each do |neighbour|
|
17
|
+
arcs = get_arcs_by_node node, neighbour
|
18
|
+
|
19
|
+
# This if statement entered if two nodes are connected twice,
|
20
|
+
# in both directions. Remove one direction as it shouldn't be here
|
21
|
+
if arcs.length > 1
|
22
|
+
if first_side == OrientedNodeTrail::START_IS_FIRST
|
23
|
+
arcs = arcs.select do |arc|
|
24
|
+
(arc.begin_node_id == node.node_id and arc.begin_node_direction) or
|
25
|
+
(arc.end_node_id == node.node_id and !arc.end_node_direction)
|
26
|
+
end
|
27
|
+
else
|
28
|
+
arcs = arcs.select do |arc|
|
29
|
+
(arc.end_node_id == node.node_id and arc.end_node_direction) or
|
30
|
+
(arc.begin_node_id == node.node_id and !arc.begin_node_direction)
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
34
|
+
|
35
|
+
# Sometimes, but rarely, two nodes will be joined more than once, for whatever reason
|
36
|
+
arcs.each do |arc|
|
37
|
+
oriented = OrientedNodeTrail::OrientedNode.new
|
38
|
+
oriented.node = neighbour
|
39
|
+
if arc.begin_node_id == arc.end_node_id
|
40
|
+
# A node connecting to itself. Happens rarely.
|
41
|
+
if first_side == OrientedNodeTrail::START_IS_FIRST
|
42
|
+
if arc.begin_node_direction and arc.end_node_direction
|
43
|
+
oriented.first_side = OrientedNodeTrail::START_IS_FIRST
|
44
|
+
elsif arc.begin_node_direction and !arc.end_node_direction
|
45
|
+
oriented.first_side = OrientedNodeTrail::END_IS_FIRST
|
46
|
+
elsif !arc.begin_node_direction and arc.end_node_direction
|
47
|
+
raise "I don't think this is supposed to be possible. Programming error?"
|
48
|
+
elsif !arc.begin_node_direction and !arc.end_node_direction
|
49
|
+
oriented.first_side = OrientedNodeTrail::START_IS_FIRST
|
50
|
+
else
|
51
|
+
raise "programming error"
|
52
|
+
end
|
53
|
+
else
|
54
|
+
# coming from the end of the original node
|
55
|
+
if arc.begin_node_direction and arc.end_node_direction
|
56
|
+
oriented.first_side = OrientedNodeTrail::END_IS_FIRST
|
57
|
+
elsif arc.begin_node_direction and !arc.end_node_direction
|
58
|
+
raise "I don't think this is supposed to be possible. Programming error?"
|
59
|
+
elsif !arc.begin_node_direction and arc.end_node_direction
|
60
|
+
oriented.first_side = OrientedNodeTrail::START_IS_FIRST
|
61
|
+
elsif !arc.begin_node_direction and !arc.end_node_direction
|
62
|
+
oriented.first_side = OrientedNodeTrail::END_IS_FIRST
|
63
|
+
else
|
64
|
+
raise "programming error"
|
65
|
+
end
|
66
|
+
end
|
67
|
+
|
68
|
+
elsif arc.begin_node_id == neighbour.node_id
|
69
|
+
# connected to a different node, the 1st in the arc's pair
|
70
|
+
if arc.begin_node_direction
|
71
|
+
oriented.first_side = OrientedNodeTrail::END_IS_FIRST
|
72
|
+
else
|
73
|
+
oriented.first_side = OrientedNodeTrail::START_IS_FIRST
|
74
|
+
end
|
75
|
+
elsif arc.end_node_id == neighbour.node_id
|
76
|
+
# connected to a different node, the 2nd in the arc's pair
|
77
|
+
if arc.end_node_direction
|
78
|
+
oriented.first_side = OrientedNodeTrail::START_IS_FIRST
|
79
|
+
else
|
80
|
+
oriented.first_side = OrientedNodeTrail::END_IS_FIRST
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
neighbours_with_orientation.push oriented
|
85
|
+
end
|
86
|
+
end
|
87
|
+
|
88
|
+
return neighbours_with_orientation
|
89
|
+
end
|
90
|
+
|
91
|
+
# Like #neighbours_of, except takes a velvet node id rather than a node object
|
92
|
+
def neighbours_of_node_id(node_id, first_side)
|
93
|
+
neighbours_of(@nodes[node_id], first_side)
|
94
|
+
end
|
95
|
+
|
96
|
+
|
97
|
+
|
98
|
+
# An ordered list of nodes, each with an orientation along that trail
|
99
|
+
class OrientedNodeTrail
|
100
|
+
include Enumerable
|
101
|
+
include Bio::Velvet::Logging
|
102
|
+
|
103
|
+
attr_accessor :trail
|
104
|
+
|
105
|
+
START_IS_FIRST = :start_is_first
|
106
|
+
END_IS_FIRST = :end_is_first
|
107
|
+
|
108
|
+
class IllDefinedTrailDefinition < Exception; end
|
109
|
+
class InsufficientLengthException < Exception; end
|
110
|
+
|
111
|
+
# initialize a new path. If an array is given, each element should be a pair:
|
112
|
+
# first element of the pair is a node, and the second true/false or
|
113
|
+
# START_IS_FIRST/END_IS_FIRST
|
114
|
+
def initialize(node_pairs=[])
|
115
|
+
@trail = []
|
116
|
+
node_pairs.each do |pair|
|
117
|
+
node = pair[0]
|
118
|
+
dir = pair[1]
|
119
|
+
unless node.kind_of?(Bio::Velvet::Graph::Node) and [true, false, START_IS_FIRST, END_IS_FIRST].include?(dir)
|
120
|
+
raise "Bad initialisation of OrientedNodeTrail, with #{node_pairs.inspect}, particularly #{pair.inspect}"
|
121
|
+
end
|
122
|
+
onode = OrientedNode.new
|
123
|
+
onode.node = node
|
124
|
+
if dir==true
|
125
|
+
onode.first_side = START_IS_FIRST
|
126
|
+
elsif dir==false
|
127
|
+
onode.first_side = END_IS_FIRST
|
128
|
+
else
|
129
|
+
onode.first_side = dir
|
130
|
+
end
|
131
|
+
@trail.push onode
|
132
|
+
end
|
133
|
+
end
|
134
|
+
|
135
|
+
def self.create_from_shorthand(path_string, graph)
|
136
|
+
stones = path_string.split(',').collect{|s| s.strip}
|
137
|
+
return self.new if stones.length == 0
|
138
|
+
trail = []
|
139
|
+
stones.each do |stone|
|
140
|
+
onode = OrientedNode.new
|
141
|
+
if matches = stone.match(/^(\d+)([se])$/)
|
142
|
+
node = graph.nodes[matches[1].to_i]
|
143
|
+
raise IllDefinedTrailDefinition, "Unable to find node #{matches[1] } in the graph, cannot continue" if node.nil?
|
144
|
+
onode.node = node
|
145
|
+
|
146
|
+
if matches[2] == 's'
|
147
|
+
onode.first_side = START_IS_FIRST
|
148
|
+
else
|
149
|
+
onode.first_side = END_IS_FIRST
|
150
|
+
end
|
151
|
+
else
|
152
|
+
raise IllDefinedTrailDefinition, "Unable to underestand shorthand #{stone}"
|
153
|
+
end
|
154
|
+
trail.push onode
|
155
|
+
end
|
156
|
+
path = self.new
|
157
|
+
path.trail = trail
|
158
|
+
return path
|
159
|
+
end
|
160
|
+
|
161
|
+
# Given a string like '2,3,4' (super-shorthand form),
|
162
|
+
# return the OrientedNodeTrail that thise defines. Raises
|
163
|
+
# 'IllDefinedTrailDefinition Exception if there is any ambiguity.
|
164
|
+
def self.create_from_super_shorthand(path_string, graph)
|
165
|
+
stones = path_string.split(',').collect{|s| s.strip}
|
166
|
+
return self.new if stones.length == 0
|
167
|
+
if stones.length == 1
|
168
|
+
raise IllDefinedTrailDefinition, "Cannot know path orientation when only one node is given"
|
169
|
+
end
|
170
|
+
state = 'first'
|
171
|
+
trail = []
|
172
|
+
|
173
|
+
stones.each do |str|
|
174
|
+
if matches = str.match(/^([01-9]+)$/)
|
175
|
+
if state == 'first'
|
176
|
+
state = 'second'
|
177
|
+
elsif state == 'second'
|
178
|
+
# Determine the direction of the first two nodes
|
179
|
+
first, second = stones[0..1].collect do |str|
|
180
|
+
if matches = str.match(/^([01-9]+)$/)
|
181
|
+
node = graph.nodes[matches[1].to_i]
|
182
|
+
if node.nil?
|
183
|
+
raise IllDefinedTrailDefinition, "Node `#{matches[1] }' from #{path_string} does not appear to be a node ID in the graph"
|
184
|
+
end
|
185
|
+
OrientedNode.new(node, START_IS_FIRST)
|
186
|
+
else
|
187
|
+
raise IllDefinedTrailDefinition, "Unable to parse stepping stone along the path: `#{str}'. Entire path was `#{path_string}'."
|
188
|
+
end
|
189
|
+
end
|
190
|
+
neighbours_of_first_s = first.next_neighbours(graph)
|
191
|
+
|
192
|
+
rev_first = OrientedNode.new first.node, first.first_side
|
193
|
+
rev_first.first_side = END_IS_FIRST
|
194
|
+
neighbours_of_first_e = rev_first.next_neighbours(graph)
|
195
|
+
|
196
|
+
if neighbours_of_first_s.find{|n| n.node_id == second.node_id}
|
197
|
+
if neighbours_of_first_e.find{|n| n.node_id == second.node_id}
|
198
|
+
raise IllDefinedTrailDefinition, "Both start and end of first node connect to second node, I'm confused."
|
199
|
+
else
|
200
|
+
seconds = neighbours_of_first_s.select{|n| n.node_id == second.node_id}
|
201
|
+
if seconds.length > 1
|
202
|
+
raise IllDefinedTrailDefinition, "first node connects to both start and end of second node, I'm confused."
|
203
|
+
else
|
204
|
+
trail.push first
|
205
|
+
trail.push seconds[0]
|
206
|
+
end
|
207
|
+
end
|
208
|
+
elsif neighbours_of_first_e.find{|n| n.node_id == second.node_id}
|
209
|
+
seconds = neighbours_of_first_e.select{|n| n.node_id == second.node_id}
|
210
|
+
if seconds.length > 1
|
211
|
+
raise IllDefinedTrailDefinition, "first node connects to both start and end of second node, I'm confused."
|
212
|
+
else
|
213
|
+
trail.push rev_first
|
214
|
+
trail.push seconds[0]
|
215
|
+
end
|
216
|
+
else
|
217
|
+
raise IllDefinedTrailDefinition, "First and second nodes do not appear to be directly connected"
|
218
|
+
end
|
219
|
+
state = 'beyond'
|
220
|
+
|
221
|
+
else #we are at the third or later node in the path
|
222
|
+
last = trail[-1]
|
223
|
+
neighbours_of_last = last.next_neighbours(graph)
|
224
|
+
nexts = neighbours_of_last.select{|n| n.node_id == matches[1].to_i}
|
225
|
+
if nexts.length == 0
|
226
|
+
raise IllDefinedTrailDefinition, "Nodes #{last} and #{matches[1] } do not appear to be connected"
|
227
|
+
elsif nexts.length > 1
|
228
|
+
raise IllDefinedTrailDefinition, "Node #{last} connects to both the start and end of #{matches[1] }, I'm confused"
|
229
|
+
else
|
230
|
+
trail.push nexts[0]
|
231
|
+
last = nexts[0]
|
232
|
+
end
|
233
|
+
end
|
234
|
+
|
235
|
+
else #can't regex the text as shorthand stone or super-shorthand stone
|
236
|
+
raise "Unable to parse stepping stone along the path: `#{str}'. Entire path was `#{path_string}'."
|
237
|
+
end
|
238
|
+
end
|
239
|
+
|
240
|
+
to_return = OrientedNodeTrail.new
|
241
|
+
to_return.trail = trail
|
242
|
+
return to_return
|
243
|
+
end
|
244
|
+
|
245
|
+
# Add a node to the trail. start_or_end is either
|
246
|
+
# OrientedNodeTrail::START_IS_FIRST or OrientedNodeTrail::END_IS_FIRST
|
247
|
+
def add_node(node, start_or_end)
|
248
|
+
possible_orientations = [START_IS_FIRST, END_IS_FIRST]
|
249
|
+
unless possible_orientations.include?(start_or_end)
|
250
|
+
raise "Unexpected orientation in node trail. Need one of #{possible_orientations.inspect}, found #{start_or_end}"
|
251
|
+
end
|
252
|
+
oriented = OrientedNode.new
|
253
|
+
oriented.node = node
|
254
|
+
oriented.first_side = start_or_end
|
255
|
+
@trail.push oriented
|
256
|
+
end
|
257
|
+
|
258
|
+
def add_oriented_node(oriented_node)
|
259
|
+
@trail.push oriented_node
|
260
|
+
end
|
261
|
+
|
262
|
+
# Given an Array of [node_id, start_or_end] pairs
|
263
|
+
# add these to the trail
|
264
|
+
def add_setabled_nodes(setabled_nodes, graph)
|
265
|
+
setabled_nodes.each do |pair|
|
266
|
+
raise "programming error" if pair.length != 2
|
267
|
+
add_node graph.nodes[pair[0]], pair[1]
|
268
|
+
end
|
269
|
+
end
|
270
|
+
|
271
|
+
def each(&block)
|
272
|
+
@trail.each(&block)
|
273
|
+
end
|
274
|
+
|
275
|
+
def last
|
276
|
+
@trail[@trail.length-1]
|
277
|
+
end
|
278
|
+
|
279
|
+
def remove_last_node
|
280
|
+
@trail.pop
|
281
|
+
end
|
282
|
+
|
283
|
+
def delete_at(index)
|
284
|
+
@trail.delete_at(index)
|
285
|
+
end
|
286
|
+
|
287
|
+
def length
|
288
|
+
@trail.length
|
289
|
+
end
|
290
|
+
|
291
|
+
def [](index)
|
292
|
+
@trail[index]
|
293
|
+
end
|
294
|
+
|
295
|
+
# Return true if the path contains the oriented
|
296
|
+
# node
|
297
|
+
def include_oriented_node?(oriented_node)
|
298
|
+
@trail.include?(oriented_node)
|
299
|
+
end
|
300
|
+
|
301
|
+
# Return a list of OrientedNode objects, one for each neighbour
|
302
|
+
# of the last node in this path (in the correct direction)
|
303
|
+
def neighbours_of_last_node(graph)
|
304
|
+
graph.neighbours_of(last.node, last.first_side)
|
305
|
+
end
|
306
|
+
|
307
|
+
# Return the sequence of the entire trail, or an empty string if there is no
|
308
|
+
# nodes in the trail. For certain (small) configurations of (short) nodes, there may
|
309
|
+
# be insufficient information to uniquely determine the sequence of the trail.
|
310
|
+
# In that case an exception is thrown.
|
311
|
+
def sequence
|
312
|
+
return '' if @trail.empty?
|
313
|
+
fwd_nodes_sequence, twin_nodes_sequence = sequences_within_path
|
314
|
+
missing_length_from_each_side = @trail[0].node.parent_graph.hash_length-1
|
315
|
+
if twin_nodes_sequence.length < missing_length_from_each_side
|
316
|
+
raise InsufficientLengthException, "Not enough information to know the sequence of a node trail"
|
317
|
+
else
|
318
|
+
seq_length_required = @trail.collect{|n| n.node.length_alone}.reduce(:+) + missing_length_from_each_side - twin_nodes_sequence.length
|
319
|
+
log.debug "first part: #{twin_nodes_sequence}"
|
320
|
+
log.debug "second: #{fwd_nodes_sequence[-seq_length_required...fwd_nodes_sequence.length] }"
|
321
|
+
return revcom(twin_nodes_sequence)[0...(@trail[0].node.parent_graph.hash_length-1)]+fwd_nodes_sequence
|
322
|
+
# calculating this way should be the same, but is somehow buggy in velvet?
|
323
|
+
#return revcom(twin_nodes_sequence)+fwd_nodes_sequence[-seq_length_required...fwd_nodes_sequence.length]
|
324
|
+
end
|
325
|
+
end
|
326
|
+
|
327
|
+
def sequences_within_path
|
328
|
+
return '', '' if @trail.empty?
|
329
|
+
twin_nodes_sequence = ''
|
330
|
+
fwd_nodes_sequence = ''
|
331
|
+
@trail.each do |onode|
|
332
|
+
if onode.starts_at_start?
|
333
|
+
twin_nodes_sequence = onode.node.ends_of_kmers_of_twin_node + twin_nodes_sequence
|
334
|
+
fwd_nodes_sequence += onode.node.ends_of_kmers_of_node
|
335
|
+
else
|
336
|
+
twin_nodes_sequence = onode.node.ends_of_kmers_of_node + twin_nodes_sequence
|
337
|
+
fwd_nodes_sequence += onode.node.ends_of_kmers_of_twin_node
|
338
|
+
end
|
339
|
+
end
|
340
|
+
return fwd_nodes_sequence, twin_nodes_sequence
|
341
|
+
end
|
342
|
+
|
343
|
+
def copy
|
344
|
+
o = OrientedNodeTrail.new
|
345
|
+
o.trail = Array.new(@trail.collect{|onode| onode.copy})
|
346
|
+
return o
|
347
|
+
end
|
348
|
+
|
349
|
+
def to_s
|
350
|
+
"OrientedNodeTrail: #{object_id}: #{to_shorthand }"
|
351
|
+
end
|
352
|
+
|
353
|
+
def to_short_s
|
354
|
+
collect do |onode|
|
355
|
+
onode.node.node_id
|
356
|
+
end.join(',').to_s
|
357
|
+
end
|
358
|
+
|
359
|
+
def inspect
|
360
|
+
to_s
|
361
|
+
end
|
362
|
+
|
363
|
+
# Length of a contig made from this path
|
364
|
+
def length_in_bp
|
365
|
+
return 0 if @trail.empty?
|
366
|
+
return length_in_bp_within_path+@trail[0].node.parent_graph.hash_length-1
|
367
|
+
end
|
368
|
+
|
369
|
+
# Length of this trail if it is part of a larger path
|
370
|
+
def length_in_bp_within_path
|
371
|
+
return 0 if @trail.empty?
|
372
|
+
reduce(0) do |total, onode|
|
373
|
+
total + onode.node.length_alone
|
374
|
+
end
|
375
|
+
end
|
376
|
+
|
377
|
+
def to_shorthand
|
378
|
+
shorthand = @trail.collect do |onode|
|
379
|
+
[
|
380
|
+
onode.node.node_id,
|
381
|
+
onode.starts_at_start? ? 's' : 'e'
|
382
|
+
].join
|
383
|
+
end.join(',')
|
384
|
+
end
|
385
|
+
|
386
|
+
def reverse!
|
387
|
+
@trail.reverse!
|
388
|
+
@trail.each do |onode|
|
389
|
+
onode.reverse!
|
390
|
+
end
|
391
|
+
nil
|
392
|
+
end
|
393
|
+
|
394
|
+
def reverse
|
395
|
+
rev = copy
|
396
|
+
rev.reverse!
|
397
|
+
return rev
|
398
|
+
end
|
399
|
+
|
400
|
+
# The weighted average of coverages along the trail,
|
401
|
+
# (weighted by node length)
|
402
|
+
def coverage
|
403
|
+
total_length = 0
|
404
|
+
total_coverage = 0.0
|
405
|
+
each do |onode|
|
406
|
+
len = onode.node.length_alone
|
407
|
+
total_coverage += onode.node.coverage*len
|
408
|
+
total_length += len
|
409
|
+
end
|
410
|
+
return total_coverage / total_length
|
411
|
+
end
|
412
|
+
|
413
|
+
def ==(another)
|
414
|
+
return false if trail.length != another.trail.length
|
415
|
+
each_with_index do |onode, i|
|
416
|
+
return false unless onode == another[i]
|
417
|
+
end
|
418
|
+
return true
|
419
|
+
end
|
420
|
+
|
421
|
+
class OrientedNode
|
422
|
+
attr_accessor :node, :first_side
|
423
|
+
|
424
|
+
def initialize(node=nil, first_side=nil)
|
425
|
+
@node = node
|
426
|
+
if first_side == true
|
427
|
+
@first_side = OrientedNodeTrail::START_IS_FIRST
|
428
|
+
elsif first_side == false
|
429
|
+
@first_side = OrientedNodeTrail::END_IS_FIRST
|
430
|
+
else
|
431
|
+
@first_side = first_side
|
432
|
+
end
|
433
|
+
end
|
434
|
+
|
435
|
+
def starts_at_start?
|
436
|
+
@first_side == OrientedNodeTrail::START_IS_FIRST
|
437
|
+
end
|
438
|
+
|
439
|
+
def starts_at_end?
|
440
|
+
@first_side == OrientedNodeTrail::END_IS_FIRST
|
441
|
+
end
|
442
|
+
|
443
|
+
def to_s
|
444
|
+
"OrientedNode: node #{@node.node_id}, first_side: #{@first_side}"
|
445
|
+
end
|
446
|
+
|
447
|
+
def to_shorthand
|
448
|
+
if @first_side == OrientedNodeTrail::START_IS_FIRST
|
449
|
+
return "#{node_id}s"
|
450
|
+
else
|
451
|
+
return "#{node_id}e"
|
452
|
+
end
|
453
|
+
end
|
454
|
+
|
455
|
+
# Set#include? doesn't pick up when the same OrientedNode is picked
|
456
|
+
# up twice independently, I don't think. So convert to an array first
|
457
|
+
def to_settable
|
458
|
+
[@node.node_id, @first_side]
|
459
|
+
end
|
460
|
+
|
461
|
+
def hash
|
462
|
+
to_settable.hash
|
463
|
+
end
|
464
|
+
|
465
|
+
def node_id
|
466
|
+
@node.node_id
|
467
|
+
end
|
468
|
+
|
469
|
+
def ==(another)
|
470
|
+
@node == another.node and @first_side == another.first_side
|
471
|
+
end
|
472
|
+
|
473
|
+
def next_neighbours(graph)
|
474
|
+
graph.neighbours_of @node, @first_side
|
475
|
+
end
|
476
|
+
|
477
|
+
# switch @first_side of this node
|
478
|
+
def reverse!
|
479
|
+
if @first_side == OrientedNodeTrail::START_IS_FIRST
|
480
|
+
@first_side = OrientedNodeTrail::END_IS_FIRST
|
481
|
+
elsif @first_side == OrientedNodeTrail::END_IS_FIRST
|
482
|
+
@first_side = OrientedNodeTrail::START_IS_FIRST
|
483
|
+
else
|
484
|
+
raise "programming error"
|
485
|
+
end
|
486
|
+
end
|
487
|
+
|
488
|
+
# Return a new OrientedNode with the reverse direction
|
489
|
+
def reverse
|
490
|
+
rev = OrientedNode.new(@node, @first_side)
|
491
|
+
rev.reverse!
|
492
|
+
return rev
|
493
|
+
end
|
494
|
+
|
495
|
+
def copy
|
496
|
+
OrientedNode.new(@node, @first_side)
|
497
|
+
end
|
498
|
+
end
|
499
|
+
|
500
|
+
private
|
501
|
+
def revcom(seq)
|
502
|
+
Bio::Sequence::NA.new(seq).reverse_complement.to_s.upcase
|
503
|
+
end
|
504
|
+
end
|
505
|
+
end
|
506
|
+
end
|
507
|
+
end
|