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 +1,142 @@
|
|
1
|
+
class Bio::FinishM::Sequence
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
class PathSteppingStone
|
5
|
+
attr_accessor :node_id, :first_side
|
6
|
+
end
|
7
|
+
|
8
|
+
def add_options(optparse_object, options)
|
9
|
+
optparse_object.banner = "\nUsage: finishm sequence --assembly-??? --path PATH
|
10
|
+
|
11
|
+
Given a series of nodes and orientations, print the DNA sequence of the given path
|
12
|
+
\n\n"
|
13
|
+
|
14
|
+
options.merge!({
|
15
|
+
})
|
16
|
+
|
17
|
+
# Parse a string like '4s,2s,3e' into a programmatic version of a path
|
18
|
+
parse_path_string = lambda do |path_string|
|
19
|
+
path_string.collect do |str|
|
20
|
+
if matches = str.match(/^([01-9]+)([se])$/)
|
21
|
+
stone = PathSteppingStone.new
|
22
|
+
stone.node_id = matches[1].to_i
|
23
|
+
if matches[2] == 's'
|
24
|
+
stone.first_side = Bio::Velvet::Graph::OrientedNodeTrail::START_IS_FIRST
|
25
|
+
elsif matches[2] == 'e'
|
26
|
+
stone.first_side = Bio::Velvet::Graph::OrientedNodeTrail::END_IS_FIRST
|
27
|
+
else
|
28
|
+
raise "programming error"
|
29
|
+
end
|
30
|
+
stone
|
31
|
+
else
|
32
|
+
raise "Unable to parse stepping stone along the path: `#{arg}'. Entire path was `#{arg}'."
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
|
37
|
+
optparse_object.separator "\nOne of the following path defining arguments must be defined:\n\n"
|
38
|
+
optparse_object.on("--path-ids PATH", "A comma separated list of node IDs - the program attempts to determine the orientations automatically") do |arg|
|
39
|
+
options[:path_ids] = arg
|
40
|
+
end
|
41
|
+
optparse_object.on("--path PATH", Array, "A comma separated list of node IDs and orientations - explore from these probe IDs in the graph e.g. '4s,2s,3e' means start at the start of node 4, connecting to the beginning of node 2 and finally the end of probe 3.") do |arg|
|
42
|
+
options[:paths] = [parse_path_string.call(arg)]
|
43
|
+
end
|
44
|
+
optparse_object.on("--paths PATHS", "A colon separated list of comma separated lists of node IDs and orientations - e.g. '4s,2s,3e:532s,465s' means print 2 different paths") do |arg|
|
45
|
+
raise "Only one of --path and --paths can be specified" unless options[:paths].nil?
|
46
|
+
options[:paths] = []
|
47
|
+
arg.split(':').each do |split|
|
48
|
+
split.strip!
|
49
|
+
next if split == ''
|
50
|
+
options[:paths].push parse_path_string.call(split.split(','))
|
51
|
+
end
|
52
|
+
log.info "Read in #{options[:paths] } path definitions"
|
53
|
+
if log.debug?
|
54
|
+
|
55
|
+
end
|
56
|
+
end
|
57
|
+
|
58
|
+
optparse_object.separator "\nIf an assembly is to be done, there must be some definition of reads:\n\n" #TODO improve this help
|
59
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
60
|
+
|
61
|
+
optparse_object.separator "\nOptional graph-related arguments:\n\n"
|
62
|
+
Bio::FinishM::GraphGenerator.new.add_options optparse_object, options
|
63
|
+
end
|
64
|
+
|
65
|
+
def validate_options(options, argv)
|
66
|
+
#TODO: give a better description of the error that has occurred
|
67
|
+
#TODO: require reads options
|
68
|
+
if argv.length != 0
|
69
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
70
|
+
else
|
71
|
+
if options[:path_ids]
|
72
|
+
if options[:paths]
|
73
|
+
return "Multiple ways to define the path given, one at a time please"
|
74
|
+
end
|
75
|
+
else
|
76
|
+
if options[:paths].nil? or options[:paths].empty?
|
77
|
+
return "No path defined, so don't know how to procede through the graph"
|
78
|
+
end
|
79
|
+
end
|
80
|
+
|
81
|
+
|
82
|
+
# Need reads unless there is already an assembly
|
83
|
+
unless options[:previous_assembly] or options[:previously_serialized_parsed_graph_file]
|
84
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
85
|
+
else
|
86
|
+
return nil
|
87
|
+
end
|
88
|
+
end
|
89
|
+
end
|
90
|
+
|
91
|
+
def run(options, argv)
|
92
|
+
read_input = Bio::FinishM::ReadInput.new
|
93
|
+
read_input.parse_options options
|
94
|
+
|
95
|
+
# Generate the assembly graph
|
96
|
+
log.info "Reading in or generating the assembly graph"
|
97
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph([], read_input, options)
|
98
|
+
|
99
|
+
print_trail = lambda do |oriented_trail|
|
100
|
+
print '>'
|
101
|
+
puts oriented_trail.to_shorthand
|
102
|
+
puts oriented_trail.sequence
|
103
|
+
end
|
104
|
+
|
105
|
+
if options[:path_ids]
|
106
|
+
trail = Bio::Velvet::Graph::OrientedNodeTrail.create_from_super_shorthand(options[:path_ids], finishm_graph.graph)
|
107
|
+
print_trail.call trail
|
108
|
+
|
109
|
+
else
|
110
|
+
# Build the oriented node trail
|
111
|
+
log.info "Building the trail(s) from the nodes"
|
112
|
+
options[:paths].each do |path|
|
113
|
+
trail = Bio::Velvet::Graph::OrientedNodeTrail.new
|
114
|
+
path.each do |stone|
|
115
|
+
log.debug "Adding stone to the trail: #{stone.inspect}"
|
116
|
+
node = finishm_graph.graph.nodes[stone.node_id]
|
117
|
+
if node.nil?
|
118
|
+
raise "Unable to find node ID #{stone.node_id} in the graph, so cannot continue"
|
119
|
+
end
|
120
|
+
|
121
|
+
# check that the path actually connects in the graph, otherwise stop.
|
122
|
+
is_neighbour = false
|
123
|
+
unless trail.length == 0 #don't worry about the first stepping stone
|
124
|
+
trail.neighbours_of_last_node(finishm_graph.graph).each do |oneigh|
|
125
|
+
log.debug "Considering neighbour #{oneigh.inspect}"
|
126
|
+
is_neighbour = true if oneigh.node == node and oneigh.first_side == stone.first_side
|
127
|
+
end
|
128
|
+
unless is_neighbour
|
129
|
+
raise "In the graph, the node #{trail.last.to_s} does not connect with #{stone.inspect}"
|
130
|
+
end
|
131
|
+
end
|
132
|
+
|
133
|
+
# OK, all the checking done. Actually add it to the trail
|
134
|
+
trail.add_node node, stone.first_side
|
135
|
+
end
|
136
|
+
|
137
|
+
# Print the sequence
|
138
|
+
print_trail.call trail
|
139
|
+
end
|
140
|
+
end
|
141
|
+
end
|
142
|
+
end
|
@@ -0,0 +1,430 @@
|
|
1
|
+
class Bio::FinishM::Visualise
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
DEFAULT_OPTIONS = {
|
5
|
+
:min_adjoining_reads => 2,
|
6
|
+
:max_adjoining_node_coverage => 300,
|
7
|
+
:graph_search_leash_length => 20000,
|
8
|
+
:interesting_probes => nil,
|
9
|
+
:max_nodes => 50,
|
10
|
+
:contig_end_length => 200
|
11
|
+
}
|
12
|
+
|
13
|
+
def add_options(optparse_object, options)
|
14
|
+
options.merge! DEFAULT_OPTIONS
|
15
|
+
optparse_object.banner = "\nUsage: finishm visualise --assembly-??? <output_visualisation_file>
|
16
|
+
|
17
|
+
Visualise an assembly graph
|
18
|
+
\n\n"
|
19
|
+
|
20
|
+
optparse_object.separator "Output visualisation formats (one or more of these must be used)"
|
21
|
+
add_visualisation_options(optparse_object, options)
|
22
|
+
|
23
|
+
optparse_object.separator "Input genome information"
|
24
|
+
optparse_object.separator "\nIf an assembly is to be done, there must be some definition of reads:\n\n" #TODO improve this help
|
25
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
26
|
+
|
27
|
+
optparse_object.separator "\nOptional graph-exploration arguments:\n\n"
|
28
|
+
add_scaffold_options(optparse_object, options)
|
29
|
+
add_probe_options(optparse_object, options)
|
30
|
+
|
31
|
+
optparse_object.separator "\nOptional graph-related arguments:\n\n"
|
32
|
+
Bio::FinishM::GraphGenerator.new.add_options(optparse_object, options)
|
33
|
+
end
|
34
|
+
|
35
|
+
def validate_options(options, argv)
|
36
|
+
#TODO: give a better description of the error that has occurred
|
37
|
+
#TODO: require reads options
|
38
|
+
return validate_argv_length(argv) ||
|
39
|
+
validate_visualisation_options(options) ||
|
40
|
+
validate_probe_options(options) ||
|
41
|
+
validate_assembly_options(options)
|
42
|
+
end
|
43
|
+
|
44
|
+
def add_visualisation_options(optparse_object, options)
|
45
|
+
optparse_object.on("--assembly-svg PATH", "Output assembly as a SVG file [default: off]") do |arg|
|
46
|
+
options[:output_graph_svg] = arg
|
47
|
+
end
|
48
|
+
optparse_object.on("--assembly-png PATH", "Output assembly as a PNG file [default: off]") do |arg|
|
49
|
+
options[:output_graph_png] = arg
|
50
|
+
end
|
51
|
+
optparse_object.on("--assembly-dot PATH", "Output assembly as a DOT file [default: off]") do |arg|
|
52
|
+
options[:output_graph_dot] = arg
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
def validate_visualisation_options(options)
|
57
|
+
if options[:output_graph_png].nil? and options[:output_graph_svg].nil? and options[:output_graph_dot].nil?
|
58
|
+
return "No visualisation output format/file given, don't know how to visualise"
|
59
|
+
end
|
60
|
+
end
|
61
|
+
|
62
|
+
def add_scaffold_options(optparse_object, options)
|
63
|
+
optparse_object.on("--genomes FASTA_1[,FASTA_2...]", Array, "Fasta files of genomes used in the assembly. Required if --scaffolds is given [default: unused]") do |arg|
|
64
|
+
options[:assembly_files] = arg
|
65
|
+
end
|
66
|
+
optparse_object.on("--scaffolds SIDE_1[,SIDE_2...]", Array, "explore from these scaffold ends e.g 'contig1s' for the start of contig1, 'contig1e' for the end of contig1, and 'contig1,contig3e' for both sides of contig1 and the end of contig3 [default: unused]") do |arg|
|
67
|
+
options[:scaffold_sides] = arg.collect do |side|
|
68
|
+
if side.match(/[se]$/)
|
69
|
+
side
|
70
|
+
else
|
71
|
+
["#{side}s","#{side}e"]
|
72
|
+
end
|
73
|
+
end.flatten
|
74
|
+
end
|
75
|
+
optparse_object.on("--overhang NUM", Integer, "Start assembling this far from the ends of the contigs [default: #{options[:contig_end_length]}]") do |arg|
|
76
|
+
options[:contig_end_length] = arg.to_i
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
def validate_scaffold_options(options)
|
81
|
+
# If scaffolds are defined, then probe genomes must also be defined
|
82
|
+
if options[:scaffolds] and !options[:assembly_files]
|
83
|
+
return "If --scaffolds is defined, so then must --genomes"
|
84
|
+
end
|
85
|
+
end
|
86
|
+
|
87
|
+
def add_probe_options(optparse_object, options)
|
88
|
+
optparse_object.on("--probe-ids PROBE_IDS", Array, "explore from these probe IDs in the graph (comma separated). probe ID is the ID in the velvet Sequence file. See also --leash-length [default: don't start from a node, explore the entire graph]") do |arg|
|
89
|
+
options[:interesting_probes] = arg.collect do |read|
|
90
|
+
read_id = read.to_i
|
91
|
+
if read_id.to_s != read or read_id.nil? or read_id < 1
|
92
|
+
raise "Unable to parse probe ID #{read}, from #{arg}, cannot continue"
|
93
|
+
end
|
94
|
+
read_id
|
95
|
+
end
|
96
|
+
end
|
97
|
+
optparse_object.on("--probe-ids-file PROBE_IDS_FILE", String, "explore from the probe IDs given in the file (1 probe ID per line). See also --leash-length [default: don't start from a node, explore the entire graph]") do |arg|
|
98
|
+
raise "Cannot specify both --probe-ids and --probe-ids-file sorry" if options[:interesting_probes]
|
99
|
+
options[:interesting_probes] = []
|
100
|
+
log.info "Reading probe IDs from file: `#{arg}'"
|
101
|
+
File.foreach(arg) do |line|
|
102
|
+
line.strip!
|
103
|
+
next if line == '' or line.nil?
|
104
|
+
read_id = line.to_i
|
105
|
+
if read_id.to_s != line or read_id < 1 or read_id.nil?
|
106
|
+
raise "Unable to parse probe ID #{line}, from file #{arg}, cannot continue"
|
107
|
+
end
|
108
|
+
options[:interesting_probes].push read_id
|
109
|
+
end
|
110
|
+
log.info "Read #{options[:interesting_probes].length} probes in"
|
111
|
+
end
|
112
|
+
optparse_object.on("--probe-names-file PROBE_NAMES_FILE", String, "explore from the probe names (i.e. the first word in the fasta/fastq header) given in the file (1 probe name per line). See also --leash-length [default: don't start from a node, explore the entire graph]") do |arg|
|
113
|
+
raise "Cannot specify any two of --probe-names-file, --probe-ids and --probe-ids-file sorry" if options[:interesting_probes]
|
114
|
+
options[:interesting_probe_names] = []
|
115
|
+
log.info "Reading probe names from file: `#{arg}'"
|
116
|
+
File.foreach(arg) do |line|
|
117
|
+
line.strip!
|
118
|
+
next if line == '' or line.nil?
|
119
|
+
options[:interesting_probe_names].push line.split(/\s/)[0]
|
120
|
+
end
|
121
|
+
log.info "Read #{options[:interesting_probe_names].length} probes names in"
|
122
|
+
end
|
123
|
+
optparse_object.on("--probe-to-node-map FILE", String, "Output a tab separated file containing the read IDs and their respective node IDs [default: no output]") do |arg|
|
124
|
+
options[:probe_to_node_map] = arg
|
125
|
+
end
|
126
|
+
optparse_object.on("--node-ids NODE_IDS", Array, "explore from these nodes in the graph (comma separated). Node IDs are the nodes in the velvet graph. See also --leash-length [default: don't start from a node, explore the entire graph]") do |arg|
|
127
|
+
options[:interesting_nodes] = arg.collect do |read|
|
128
|
+
node_id = read.to_i
|
129
|
+
if node_id.to_s != read or node_id.nil? or node_id < 1
|
130
|
+
raise "Unable to parse node ID #{read}, from #{arg}, cannot continue"
|
131
|
+
end
|
132
|
+
node_id
|
133
|
+
end
|
134
|
+
end
|
135
|
+
optparse_object.on("--leash-length NUM", Integer, "Don't explore too far in the graph, only this far and not much more [default: unused unless --probe-ids or --nodes is specified, otherwise #{options[:graph_search_leash_length] }]") do |arg|
|
136
|
+
options[:graph_search_leash_length] = arg
|
137
|
+
end
|
138
|
+
optparse_object.on("--max-nodes NUM", Integer, "Maximum number of nodes to explore out from each probe node, or 0 for no maximum [default: #{options[:max_nodes] }]") do |arg|
|
139
|
+
if arg==0
|
140
|
+
options[:max_nodes] = nil
|
141
|
+
else
|
142
|
+
options[:max_nodes] = arg
|
143
|
+
end
|
144
|
+
end
|
145
|
+
end
|
146
|
+
|
147
|
+
def validate_probe_options(options)
|
148
|
+
if options[:interesting_probes] and options[:interesting_nodes]
|
149
|
+
return "Can only be interested in probes or nodes, not both, at least currently"
|
150
|
+
end
|
151
|
+
end
|
152
|
+
|
153
|
+
def validate_assembly_options(options)
|
154
|
+
# Need reads unless there is already an assembly
|
155
|
+
unless options[:previous_assembly] or options[:previously_serialized_parsed_graph_file]
|
156
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
157
|
+
end
|
158
|
+
end
|
159
|
+
|
160
|
+
def validate_argv_length(argv)
|
161
|
+
if argv.length != 0
|
162
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
163
|
+
end
|
164
|
+
end
|
165
|
+
|
166
|
+
def run(options, argv)
|
167
|
+
read_input = Bio::FinishM::ReadInput.new
|
168
|
+
read_input.parse_options options
|
169
|
+
|
170
|
+
# Generate the assembly graph
|
171
|
+
log.info "Reading in or generating the assembly graph"
|
172
|
+
|
173
|
+
if options[:interesting_probes] or options[:interesting_probe_names]
|
174
|
+
finishm_graph, interesting_node_ids = generate_graph_from_probes(read_input, options)
|
175
|
+
|
176
|
+
if (options[:interesting_probes] or options[:interesting_probe_names]) and options[:probe_to_node_map]
|
177
|
+
write_probe_to_node_map(options[:probe_to_node_map], finishm_graph, options[:interesting_probes])
|
178
|
+
end
|
179
|
+
elsif options[:interesting_nodes]
|
180
|
+
finishm_graph = generate_graph_from_nodes(read_input, options)
|
181
|
+
interesting_node_ids = options[:interesting_nodes]
|
182
|
+
elsif options[:assembly_files]
|
183
|
+
finishm_graph, interesting_node_ids, node_id_to_nickname = generate_graph_from_assembly(read_input, options)
|
184
|
+
options[:node_id_to_nickname] = node_id_to_nickname
|
185
|
+
else
|
186
|
+
# Visualising the entire graph
|
187
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph([], read_input, options)
|
188
|
+
end
|
189
|
+
|
190
|
+
|
191
|
+
if options[:graph_search_leash_length] and interesting_node_ids
|
192
|
+
#log.info "Finding nodes within the leash length of #{options[:graph_search_leash_length] }.."
|
193
|
+
nodes_within_leash, node_ids_at_leash = get_nodes_within_leash(finishm_graph, interesting_node_ids, options)
|
194
|
+
log.info "Found #{node_ids_at_leash.length} nodes at the end of the #{options[:graph_search_leash_length] }bp leash" if options[:graph_search_leash_length]
|
195
|
+
|
196
|
+
options.merge!({
|
197
|
+
:start_node_ids => interesting_node_ids,
|
198
|
+
:nodes => nodes_within_leash,
|
199
|
+
:end_node_ids => node_ids_at_leash,
|
200
|
+
|
201
|
+
})
|
202
|
+
else
|
203
|
+
options[:nodes] = finishm_graph.graph.nodes
|
204
|
+
end
|
205
|
+
|
206
|
+
# Determine paired-end connections
|
207
|
+
log.info "Determining paired-end node connections.."
|
208
|
+
paired_end_links = find_paired_end_linkages(finishm_graph, options[:nodes])
|
209
|
+
options[:paired_nodes_hash] = paired_end_links
|
210
|
+
|
211
|
+
create_graphviz_output(finishm_graph, options)
|
212
|
+
end
|
213
|
+
|
214
|
+
def create_graphviz_output(finishm_graph, options)
|
215
|
+
log.info "Converting assembly to a graphviz.."
|
216
|
+
gv = Bio::Assembly::ABVisualiser.new.graphviz(finishm_graph.graph, {
|
217
|
+
:start_node_ids => options[:start_node_ids],
|
218
|
+
:nodes => options[:nodes],
|
219
|
+
:end_node_ids => options[:end_node_ids],
|
220
|
+
:paired_nodes_hash => options[:paired_nodes_hash],
|
221
|
+
:node_id_to_nickname => options[:node_id_to_nickname]
|
222
|
+
})
|
223
|
+
|
224
|
+
# Convert gv object to something actually pictorial
|
225
|
+
if options[:output_graph_png]
|
226
|
+
log.info "Writing PNG #{options[:output_graph_png] }"
|
227
|
+
gv.output :png => options[:output_graph_png], :use => :neato
|
228
|
+
end
|
229
|
+
if options[:output_graph_svg]
|
230
|
+
log.info "Writing SVG #{options[:output_graph_svg] }"
|
231
|
+
gv.output :svg => options[:output_graph_svg], :use => :neato
|
232
|
+
end
|
233
|
+
if options[:output_graph_dot]
|
234
|
+
log.info "Writing DOT #{options[:output_graph_dot] }"
|
235
|
+
gv.output :dot => options[:output_graph_dot], :use => :neato
|
236
|
+
end
|
237
|
+
end
|
238
|
+
|
239
|
+
def generate_graph_from_probes(read_input, options)
|
240
|
+
# Looking based on probes
|
241
|
+
if options[:interesting_probe_names]
|
242
|
+
if options[:interesting_probe_names].length > 5
|
243
|
+
log.info "Targeting #{options[:interesting_probe_names].length} probes #{options[:interesting_probe_names][0..4].join(', ') }, ..."
|
244
|
+
else
|
245
|
+
log.info "Targeting #{options[:interesting_probe_names].length} probes #{options[:interesting_probe_names].inspect}"
|
246
|
+
end
|
247
|
+
options[:probe_read_names] = options[:interesting_probe_names]
|
248
|
+
else
|
249
|
+
if options[:interesting_probes].length > 5
|
250
|
+
log.info "Targeting #{options[:interesting_probes].length} probes #{options[:interesting_probes][0..4].join(', ') }, ..."
|
251
|
+
else
|
252
|
+
log.info "Targeting #{options[:interesting_probes].length} probes #{options[:interesting_probes].inspect}"
|
253
|
+
end
|
254
|
+
options[:probe_reads] = options[:interesting_probes]
|
255
|
+
end
|
256
|
+
|
257
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph([], read_input, options)
|
258
|
+
interesting_node_ids = finishm_graph.probe_nodes.reject{|n| n.nil?}.collect{|node| node.node_id}
|
259
|
+
|
260
|
+
return finishm_graph, interesting_node_ids
|
261
|
+
end
|
262
|
+
|
263
|
+
def generate_graph_from_nodes(read_input, options)
|
264
|
+
# Looking based on nodes
|
265
|
+
if options[:interesting_nodes].length > 5
|
266
|
+
log.info "Targeting #{options[:interesting_nodes].length} nodes #{options[:interesting_nodes][0..4].join(', ') }, ..."
|
267
|
+
else
|
268
|
+
log.info "Targeting #{options[:interesting_nodes].length} node(s) #{options[:interesting_nodes].inspect}"
|
269
|
+
end
|
270
|
+
|
271
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph([], read_input, options)
|
272
|
+
interesting_node_ids = options[:interesting_nodes]
|
273
|
+
|
274
|
+
return finishm_graph, interesting_node_ids
|
275
|
+
end
|
276
|
+
|
277
|
+
def generate_graph_from_assembly(read_input, options)
|
278
|
+
# Parse the genome fasta file in
|
279
|
+
genomes = Bio::FinishM::InputGenome.parse_genome_fasta_files(
|
280
|
+
options[:assembly_files],
|
281
|
+
options[:contig_end_length],
|
282
|
+
options
|
283
|
+
)
|
284
|
+
|
285
|
+
# Create hash of contig end name to probe index
|
286
|
+
contig_name_to_probe = {}
|
287
|
+
genomes.each do |genome|
|
288
|
+
genome.scaffolds.each_with_index do |swaff, scaffold_index|
|
289
|
+
probes = [
|
290
|
+
genome.first_probe(scaffold_index),
|
291
|
+
genome.last_probe(scaffold_index)
|
292
|
+
]
|
293
|
+
probes.each do |probe|
|
294
|
+
key = nil
|
295
|
+
if probe.side == :start
|
296
|
+
key = "#{probe.contig.scaffold.name}s"
|
297
|
+
elsif probe.side == :end
|
298
|
+
key = "#{probe.contig.scaffold.name}e"
|
299
|
+
else
|
300
|
+
raise "Programming error"
|
301
|
+
end
|
302
|
+
|
303
|
+
if contig_name_to_probe.key?(key)
|
304
|
+
log.error "Encountered multiple contigs with the same name, this might cause problems, so quitting #{key}"
|
305
|
+
end
|
306
|
+
contig_name_to_probe[key] = probe.index
|
307
|
+
end
|
308
|
+
end
|
309
|
+
end
|
310
|
+
|
311
|
+
# Gather a list of probe indexes that are of interest to the user
|
312
|
+
interesting_probe_ids = []
|
313
|
+
if options[:scaffold_sides]
|
314
|
+
# If looking at specified ends
|
315
|
+
nodes_to_start_from = options[:scaffold_sides].collect do |side|
|
316
|
+
if probe = contig_name_to_probe[side]
|
317
|
+
interesting_probe_ids << probe
|
318
|
+
else
|
319
|
+
raise "Unable to find scaffold side in given genome: #{side}"
|
320
|
+
end
|
321
|
+
end
|
322
|
+
log.info "Found #{interesting_probe_ids.length} scaffold sides in the assembly of interest"
|
323
|
+
else
|
324
|
+
# else looking at all the contig ends in all the genomes
|
325
|
+
interesting_probe_ids = contig_name_to_probe.values
|
326
|
+
log.info "Visualising all #{interesting_probe_ids.length} contig ends in all genomes"
|
327
|
+
end
|
328
|
+
|
329
|
+
# Generate the graph
|
330
|
+
probe_sequences = genomes.collect{|genome| genome.probe_sequences}.flatten
|
331
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
332
|
+
|
333
|
+
# Convert probe IDs into node IDs
|
334
|
+
interesting_node_ids = interesting_probe_ids.collect do |pid|
|
335
|
+
finishm_graph.probe_nodes[pid].node_id
|
336
|
+
end.uniq
|
337
|
+
|
338
|
+
# create a nickname hash, id of node to name. Include all nodes even if they weren't specified directly (they only get visualised if they are within leash length of another)
|
339
|
+
node_id_to_nickname = {}
|
340
|
+
contig_name_to_probe.each do |name, probe|
|
341
|
+
key = finishm_graph.probe_nodes[probe].node_id
|
342
|
+
if node_id_to_nickname.key?(key)
|
343
|
+
node_id_to_nickname[key] += " "+name
|
344
|
+
else
|
345
|
+
node_id_to_nickname[key] = name
|
346
|
+
end
|
347
|
+
end
|
348
|
+
|
349
|
+
return finishm_graph, interesting_node_ids, node_id_to_nickname
|
350
|
+
end
|
351
|
+
|
352
|
+
def get_nodes_within_leash(finishm_graph, node_ids, options={})
|
353
|
+
log.info "Finding nodes within the leash length of #{options[:graph_search_leash_length] } with maximum node count #{options[:max_nodes] }.."
|
354
|
+
dijkstra = Bio::AssemblyGraphAlgorithms::Dijkstra.new
|
355
|
+
|
356
|
+
@finder = Bio::FinishM::PairedEndNeighbourFinder.new(finishm_graph, 500) #TODO: this hard-coded 100 isn't great here
|
357
|
+
@finder.min_adjoining_reads = options[:min_adjoining_reads]
|
358
|
+
@finder.max_adjoining_node_coverage = options[:max_adjoining_node_coverage]
|
359
|
+
|
360
|
+
nodes_within_leash_hash = dijkstra.min_distances_from_many_nodes_in_both_directions(
|
361
|
+
finishm_graph.graph, node_ids.collect{|n| finishm_graph.graph.nodes[n]}, {
|
362
|
+
:ignore_directions => true,
|
363
|
+
:leash_length => options[:graph_search_leash_length],
|
364
|
+
:max_nodes => options[:max_nodes],
|
365
|
+
:neighbour_finder => @finder
|
366
|
+
})
|
367
|
+
nodes_within_leash = nodes_within_leash_hash.keys.collect{|k| finishm_graph.graph.nodes[k[0]]}
|
368
|
+
log.info "Found #{nodes_within_leash.collect{|o| o.node_id}.uniq.length} node(s) within the leash length"
|
369
|
+
|
370
|
+
# These nodes are at the end of the leash - a node is in here iff
|
371
|
+
# it has a neighbour that is not in the nodes_within_leash
|
372
|
+
node_ids_at_leash = Set.new
|
373
|
+
nodes_within_leash_hash.keys.each do |node_and_direction|
|
374
|
+
# Add it to the set if 1 or more nieghbours are not in the original set
|
375
|
+
node = finishm_graph.graph.nodes[node_and_direction[0]]
|
376
|
+
onode = Bio::Velvet::Graph::OrientedNodeTrail::OrientedNode.new node, node_and_direction[1]
|
377
|
+
onode.next_neighbours(finishm_graph.graph).each do |oneigh|
|
378
|
+
if !nodes_within_leash_hash.key?(oneigh.to_settable)
|
379
|
+
node_ids_at_leash << node_and_direction[0]
|
380
|
+
break #it only takes one to be listed
|
381
|
+
end
|
382
|
+
end
|
383
|
+
end
|
384
|
+
|
385
|
+
return nodes_within_leash.uniq, node_ids_at_leash.to_a.uniq
|
386
|
+
end
|
387
|
+
|
388
|
+
def find_paired_end_linkages(finishm_graph, node_array)
|
389
|
+
return {} if @finder.nil?
|
390
|
+
|
391
|
+
paired_end_links = {}
|
392
|
+
node_array.each do |node|
|
393
|
+
paired_end_links[node.node_id] = []
|
394
|
+
[Bio::Velvet::Graph::OrientedNodeTrail::START_IS_FIRST,
|
395
|
+
Bio::Velvet::Graph::OrientedNodeTrail::END_IS_FIRST].each do |direction|
|
396
|
+
onode = Bio::Velvet::Graph::OrientedNodeTrail::OrientedNode.new(node, direction)
|
397
|
+
|
398
|
+
paired_end_links[node.node_id].push @finder.neighbours(onode).collect{|n| n.node.node_id}.uniq
|
399
|
+
end
|
400
|
+
paired_end_links[node.node_id].flatten!
|
401
|
+
end
|
402
|
+
return paired_end_links
|
403
|
+
end
|
404
|
+
|
405
|
+
# Write to a file probe_to_node_map_file a map that shows the
|
406
|
+
# probe ID, which node that probe is on, and the name of the probe
|
407
|
+
def write_probe_to_node_map(probe_to_node_map_file, finishm_graph, names)
|
408
|
+
log.info "Writing probe-to-node map to #{x}.."
|
409
|
+
File.open(probe_to_node_map_file,'w') do |f|
|
410
|
+
f.puts %w(probe_number probe node direction).join("\t")
|
411
|
+
finishm_graph.probe_nodes.each_with_index do |node, i|
|
412
|
+
if node.nil?
|
413
|
+
f.puts [
|
414
|
+
i+1,
|
415
|
+
names[i],
|
416
|
+
'-',
|
417
|
+
'-',
|
418
|
+
].join("\t")
|
419
|
+
else
|
420
|
+
f.puts [
|
421
|
+
i+1,
|
422
|
+
names[i],
|
423
|
+
node.node_id,
|
424
|
+
finishm_graph.probe_node_directions[i] == true ? 'forward' : 'reverse',
|
425
|
+
].join("\t")
|
426
|
+
end
|
427
|
+
end
|
428
|
+
end
|
429
|
+
end
|
430
|
+
end
|