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
Binary file
|
@@ -0,0 +1,169 @@
|
|
1
|
+
require 'graphviz'
|
2
|
+
require 'set'
|
3
|
+
|
4
|
+
class Bio::Velvet::Graph::Node
|
5
|
+
def includes_kmers?(list_of_kmers)
|
6
|
+
list_of_kmers.each do |kmer|
|
7
|
+
return true if ends_of_kmers_of_node.include?(kmer) or ends_of_kmers_of_twin_node.include?(kmer)
|
8
|
+
end
|
9
|
+
return false
|
10
|
+
end
|
11
|
+
end
|
12
|
+
|
13
|
+
module Bio
|
14
|
+
module Assembly
|
15
|
+
class ABVisualiser
|
16
|
+
include Bio::FinishM::Logging
|
17
|
+
|
18
|
+
# Visualise a (velvet) graph, as a graphviz object
|
19
|
+
#
|
20
|
+
# Possible options:
|
21
|
+
# :start_kmers: list of kmers to denote the start node(s)
|
22
|
+
# :end_kmers: list of kmers to denote the end node(s)
|
23
|
+
# :start_node_id: ID of node to mark as a start
|
24
|
+
# :end_node_id:ID of node to mark as a end
|
25
|
+
# :start_node_ids: array of node IDs to mark as a start
|
26
|
+
# :end_node_ids:array of node IDs to mark as a end
|
27
|
+
# :coverage_cutoff: ignore nodes with less coverage than this cutoff
|
28
|
+
# :digraph: output as a digraph (default true, else output undirected graph)
|
29
|
+
# :nodes: an Enumerable of nodes to be visualised.
|
30
|
+
# :node_id_to_nickname: add these names to the node descriptions. Hash of integer node id to String.
|
31
|
+
# :paired_nodes_hash: a hash of node_id to Enumerable of node_ids where there is paired-end connections
|
32
|
+
def graphviz(graph, options={})
|
33
|
+
opts = {}
|
34
|
+
opts[:type] = :digraph unless options[:digraph] == false
|
35
|
+
opts[:overlap] = :scale
|
36
|
+
graphviz = GraphViz.new(:G, opts)
|
37
|
+
|
38
|
+
nodes_to_explore = Set.new(options[:nodes].to_a)
|
39
|
+
nodes_to_explore ||= Set.new(graph.nodes)
|
40
|
+
|
41
|
+
# Add all the nodes
|
42
|
+
blacklisted_node_ids = Set.new
|
43
|
+
log.debug "Converting nodes to GraphViz format"
|
44
|
+
nodes_to_explore.each do |node|
|
45
|
+
cov = node.coverage
|
46
|
+
if options[:coverage_cutoff] and cov < options[:coverage_cutoff] and !cov.nil?
|
47
|
+
blacklisted_node_ids.add node.node_id
|
48
|
+
else
|
49
|
+
cov_string = cov.nil? ? '' : cov.round
|
50
|
+
label = "n#{node.node_id}_length#{node.ends_of_kmers_of_node.length}_coverage#{cov_string}"
|
51
|
+
if options[:node_id_to_nickname] and options[:node_id_to_nickname].key?(node.node_id)
|
52
|
+
label += ' ' + options[:node_id_to_nickname][node.node_id]
|
53
|
+
end
|
54
|
+
mods = {
|
55
|
+
:label => label,
|
56
|
+
}
|
57
|
+
includes_start = false
|
58
|
+
includes_end = false
|
59
|
+
if options[:start_kmers]
|
60
|
+
includes_start = node.includes_kmers?(options[:start_kmers])
|
61
|
+
end
|
62
|
+
if options[:end_kmers]
|
63
|
+
includes_end = node.includes_kmers?(options[:end_kmers])
|
64
|
+
end
|
65
|
+
if options[:start_node_id]
|
66
|
+
includes_start = true if node.node_id == options[:start_node_id]
|
67
|
+
end
|
68
|
+
if options[:end_node_id]
|
69
|
+
includes_end = true if node.node_id == options[:end_node_id]
|
70
|
+
end
|
71
|
+
if options[:start_node_ids]
|
72
|
+
includes_start = true if options[:start_node_ids].include? node.node_id
|
73
|
+
end
|
74
|
+
if options[:end_node_ids]
|
75
|
+
includes_end = true if options[:end_node_ids].include? node.node_id
|
76
|
+
end
|
77
|
+
|
78
|
+
if includes_start and includes_end
|
79
|
+
log.warn "Start and end kmers detected in the same node!"
|
80
|
+
elsif includes_start
|
81
|
+
mods[:color] = "red"
|
82
|
+
elsif includes_end
|
83
|
+
mods[:color] = "green"
|
84
|
+
end
|
85
|
+
|
86
|
+
graphviz.add_nodes node.node_id.to_s, mods
|
87
|
+
end
|
88
|
+
end
|
89
|
+
|
90
|
+
# Add all the edges
|
91
|
+
arcs_of_interest = graph.arcs
|
92
|
+
if options[:nodes]
|
93
|
+
arcs_of_interest = Set.new
|
94
|
+
nodes_to_explore.each do |node|
|
95
|
+
graph.arcs.get_arcs_by_node_id(node.node_id).each do |arc|
|
96
|
+
arcs_of_interest << arc
|
97
|
+
end
|
98
|
+
end
|
99
|
+
end
|
100
|
+
|
101
|
+
log.info "Converting #{arcs_of_interest.length} arcs to GraphViz format"
|
102
|
+
arcs_of_interest.each do |arc|
|
103
|
+
# Add unless the node has been blacklisted
|
104
|
+
unless blacklisted_node_ids.include? arc.begin_node_id or
|
105
|
+
blacklisted_node_ids.include? arc.end_node_id or
|
106
|
+
!nodes_to_explore.include?(graph.nodes[arc.begin_node_id]) or
|
107
|
+
!nodes_to_explore.include?(graph.nodes[arc.end_node_id])
|
108
|
+
|
109
|
+
# Direction of the arrows, to denote connection to beginning of node (connects to start = in-arrow-head to node on output graph)
|
110
|
+
if arc.connects_end_to_beginning?(arc.begin_node_id, arc.end_node_id)
|
111
|
+
graphviz.add_edges arc.begin_node_id.to_s, arc.end_node_id.to_s
|
112
|
+
elsif arc.connects_end_to_end?(arc.begin_node_id, arc.end_node_id)
|
113
|
+
graphviz.add_edges arc.begin_node_id.to_s, arc.end_node_id.to_s, {:dir => "none"}
|
114
|
+
elsif arc.connects_beginning_to_beginning?(arc.begin_node_id, arc.end_node_id)
|
115
|
+
graphviz.add_edges arc.begin_node_id.to_s, arc.end_node_id.to_s, {:dir => "both"}
|
116
|
+
elsif arc.connects_beginning_to_end?(arc.begin_node_id, arc.end_node_id)
|
117
|
+
graphviz.add_edges arc.end_node_id.to_s, arc.begin_node_id.to_s
|
118
|
+
end
|
119
|
+
end
|
120
|
+
end
|
121
|
+
|
122
|
+
# Add paired_nodes_hash pairs
|
123
|
+
unless options[:paired_nodes_hash].nil?
|
124
|
+
# Create a list of arc node pairs for len calculation
|
125
|
+
arc_pairs = arcs_of_interest.collect do |arc|
|
126
|
+
[arc.begin_node_id, arc.end_node_id].sort
|
127
|
+
end
|
128
|
+
directly_connected_node_pairs = Set.new(arc_pairs)
|
129
|
+
|
130
|
+
# Keep track of pairs so multiple arcs are not drawn e.g. node1 => node2 and node2=>node1
|
131
|
+
pairs_added = Set.new
|
132
|
+
log.info "Adding paired-end linkages to GraphViz format.."
|
133
|
+
options[:paired_nodes_hash].each do |node1_id, connected_node_ids|
|
134
|
+
connected_node_ids.each do |node2_id|
|
135
|
+
next if node1_id == node2_id #skip within-node connections
|
136
|
+
sorted = [node1_id, node2_id].sort #sort so only a single connection is shown
|
137
|
+
unless pairs_added.include?(sorted) or
|
138
|
+
!nodes_to_explore.include?(graph.nodes[node1_id]) or
|
139
|
+
!nodes_to_explore.include?(graph.nodes[node2_id]) or
|
140
|
+
directly_connected_node_pairs.include?([node1_id, node2_id].sort)
|
141
|
+
|
142
|
+
graphviz.add_edges sorted[0].to_s, sorted[1].to_s, {:color => "grey", :dir => "none", :style => 'dashed'}
|
143
|
+
pairs_added << sorted
|
144
|
+
end
|
145
|
+
end
|
146
|
+
end
|
147
|
+
end
|
148
|
+
|
149
|
+
return graphviz
|
150
|
+
end
|
151
|
+
|
152
|
+
|
153
|
+
class SimplifiedGraph
|
154
|
+
def self.create_from_velvet_graph(graph)
|
155
|
+
nodes_incorporated = 0
|
156
|
+
|
157
|
+
# While there is more of the graph to incorporate
|
158
|
+
while nodes_incorporated < graph.nodes.length
|
159
|
+
raise "not implemented"
|
160
|
+
end
|
161
|
+
end
|
162
|
+
|
163
|
+
# A class representing a linear string of nodes without forks
|
164
|
+
class Path
|
165
|
+
end
|
166
|
+
end
|
167
|
+
end
|
168
|
+
end
|
169
|
+
end
|
@@ -0,0 +1,81 @@
|
|
1
|
+
require 'ds'
|
2
|
+
require 'set'
|
3
|
+
|
4
|
+
module Bio
|
5
|
+
module AssemblyGraphAlgorithms
|
6
|
+
|
7
|
+
# Represents a set of trails, and whether or not circularity has been detected,
|
8
|
+
# and whether too many paths have been detected.
|
9
|
+
class TrailSet
|
10
|
+
attr_accessor :trails
|
11
|
+
attr_accessor :circular_paths_detected
|
12
|
+
attr_accessor :max_path_limit_exceeded
|
13
|
+
include Enumerable
|
14
|
+
|
15
|
+
def initialize
|
16
|
+
@circular_paths_detected = false
|
17
|
+
@max_path_limit_exceeded = false
|
18
|
+
end
|
19
|
+
|
20
|
+
def each
|
21
|
+
unless @trails.nil?
|
22
|
+
@trails.each{|t| yield t}
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|
26
|
+
|
27
|
+
class AcyclicConnectionFinder
|
28
|
+
include Bio::FinishM::Logging
|
29
|
+
|
30
|
+
# Find trails between two oriented nodes, both facing the same way along the path.
|
31
|
+
#
|
32
|
+
# Options:
|
33
|
+
# * :recoherence_kmer: use a longer kmer to help de-bubble and de-cicularise (default don't use this)
|
34
|
+
# * :sequences: Bio::Velvet::Sequence object holding sequences of nodes within leash length
|
35
|
+
def find_trails_between_nodes(graph, initial_oriented_node, terminal_oriented_node, leash_length, options={})
|
36
|
+
|
37
|
+
#TODO: this is now implemented in the finishm_graph object - just get it from there
|
38
|
+
initial_path = Bio::Velvet::Graph::OrientedNodeTrail.new
|
39
|
+
initial_path.add_oriented_node initial_oriented_node
|
40
|
+
|
41
|
+
finder = Bio::AssemblyGraphAlgorithms::SingleCoherentPathsBetweenNodesFinder.new
|
42
|
+
return finder.find_all_connections_between_two_nodes(
|
43
|
+
graph, initial_path, terminal_oriented_node, leash_length, options[:recoherence_kmer], options[:sequences], options
|
44
|
+
)
|
45
|
+
end
|
46
|
+
|
47
|
+
# Algorithms like SingleCoherentWanderer#wander give an overly short
|
48
|
+
# base pair distance between two probes, because the length of the node
|
49
|
+
# containing the probe at either end is not included in the calculation.
|
50
|
+
#
|
51
|
+
# Return the calibrated distance i.e. the true base pair distance between
|
52
|
+
# the start of each node pair. Returned is the given distance plus the
|
53
|
+
# distance between the start of each probe and the end of the containing
|
54
|
+
# node.
|
55
|
+
def calibrate_distance_accounting_for_probes(finishm_graph, probe1_index, probe2_index, distance)
|
56
|
+
read1 = finishm_graph.probe_node_reads[probe1_index]
|
57
|
+
read2 = finishm_graph.probe_node_reads[probe2_index]
|
58
|
+
probe_node1 = finishm_graph.probe_nodes[probe1_index]
|
59
|
+
probe_node2 = finishm_graph.probe_nodes[probe2_index]
|
60
|
+
|
61
|
+
# If the start and end nodes are the same, that's a special case:
|
62
|
+
if finishm_graph.probe_nodes[probe1_index].node_id == finishm_graph.probe_nodes[probe2_index].node_id
|
63
|
+
if (read1.direction == true and read2.direction == false) or
|
64
|
+
(read1.direction == false and read2.direction == true)
|
65
|
+
return probe_node1.length - read1.offset_from_start_of_node - read2.offset_from_start_of_node - finishm_graph.graph.hash_length
|
66
|
+
else
|
67
|
+
raise "Programming error: to connect within a single contig two probes must have opposite directions: found #{read1.direction} and #{read2.direction}"
|
68
|
+
end
|
69
|
+
else
|
70
|
+
# Usual case - start and end nodes are different nodes
|
71
|
+
to_return = distance
|
72
|
+
# add the first probe side
|
73
|
+
to_return += probe_node1.length-read1.offset_from_start_of_node-finishm_graph.graph.hash_length
|
74
|
+
# add the second probe side
|
75
|
+
to_return += probe_node2.length-read1.offset_from_start_of_node-finishm_graph.graph.hash_length
|
76
|
+
return to_return
|
77
|
+
end
|
78
|
+
end
|
79
|
+
end
|
80
|
+
end
|
81
|
+
end
|
@@ -0,0 +1,615 @@
|
|
1
|
+
require 'ds'
|
2
|
+
require 'set'
|
3
|
+
|
4
|
+
module Bio
|
5
|
+
module AssemblyGraphAlgorithms
|
6
|
+
class AllOrfsFinder
|
7
|
+
include Bio::FinishM::Logging
|
8
|
+
|
9
|
+
CODON_LENGTH = 3
|
10
|
+
START_CODONS = ['ATG']
|
11
|
+
STOP_CODONS = ['TAG', 'TAA', 'TGA']
|
12
|
+
|
13
|
+
# Search for open reading frames in a graph, in all the paths begining at a set of
|
14
|
+
# nodes through a graph (or a subset defined by range)
|
15
|
+
def find_orfs_in_graph(graph, initial_paths, minimum_orf_length=nil,
|
16
|
+
range=nil, max_gapfill_paths=nil, max_cycles=nil)
|
17
|
+
|
18
|
+
problems = find_all_problems(graph,
|
19
|
+
initial_paths,
|
20
|
+
:range => range
|
21
|
+
)
|
22
|
+
|
23
|
+
find_orfs_from_problems(problems, {
|
24
|
+
:min_orf_length => minimum_orf_length,
|
25
|
+
:max_gapfill_paths => max_gapfill_paths,
|
26
|
+
:max_cycles => max_cycles,
|
27
|
+
})
|
28
|
+
end
|
29
|
+
|
30
|
+
|
31
|
+
def find_all_problems(graph, initial_paths, options={})
|
32
|
+
problems = SingleCoherentPathsBetweenNodesFinder::ProblemSet.new
|
33
|
+
prob_finder = AllProblemTrailsFinder.new(graph, initial_paths)
|
34
|
+
|
35
|
+
while current_path = prob_finder.pop
|
36
|
+
log.debug "considering #{current_path}" if log.debug?
|
37
|
+
set_key = path_to_settable(current_path)
|
38
|
+
|
39
|
+
if problems.has_key? set_key
|
40
|
+
log.debug "Already seen this problem" if log.debug?
|
41
|
+
prob = problems[set_key]
|
42
|
+
prob.known_paths.push current_path
|
43
|
+
next
|
44
|
+
end
|
45
|
+
|
46
|
+
log.debug "New dynamic problem being solved" if log.debug?
|
47
|
+
# new problem being solved here
|
48
|
+
problem = SingleCoherentPathsBetweenNodesFinder::DynamicProgrammingProblem.new
|
49
|
+
problem.known_paths.push current_path.copy
|
50
|
+
problems[set_key] = problem
|
51
|
+
|
52
|
+
neighbours = current_path.neighbours_of_last_node(graph)
|
53
|
+
if options[:range]
|
54
|
+
neighbours.select!{|onode| options[:range].include? onode.node_id}
|
55
|
+
end
|
56
|
+
if neighbours.empty?
|
57
|
+
log.debug "last is terminal" if log.debug?
|
58
|
+
|
59
|
+
problems.terminal_node_keys ||= Set.new
|
60
|
+
problems.terminal_node_keys << set_key
|
61
|
+
next
|
62
|
+
end
|
63
|
+
|
64
|
+
# explore the forward neighbours
|
65
|
+
prob_finder.push_next_neighbours current_path
|
66
|
+
log.debug "Priority queue size: #{prob_finder.size}" if log.debug?
|
67
|
+
end
|
68
|
+
|
69
|
+
return problems
|
70
|
+
end
|
71
|
+
|
72
|
+
def path_to_settable(path)
|
73
|
+
return SingleCoherentPathsBetweenNodesFinder.new.path_to_settable(path, path.last.node.length_alone + CODON_LENGTH - 1)
|
74
|
+
end
|
75
|
+
|
76
|
+
|
77
|
+
def find_orfs_from_problems(problems, options={})
|
78
|
+
max_num_paths = options[:max_gapfill_paths]
|
79
|
+
max_num_paths ||= 2196
|
80
|
+
max_cycles = options[:max_cycles] || 1
|
81
|
+
min_orf_length = options[:minimum_orf_length] || 0
|
82
|
+
|
83
|
+
counter = SingleCoherentPathsBetweenNodesFinder::CycleCounter.new(max_cycles)
|
84
|
+
decide_stack = lambda do |to_push|
|
85
|
+
part_nodes = [to_push[0].trail, to_push[1].otrail ? to_push[1].otrail.trail : []]
|
86
|
+
if max_cycles < counter.path_cycle_count(part_nodes.flatten)
|
87
|
+
log.debug "Pushing #{part_nodes.collect{|part| part.collect{|onode| onode.node.node_id}.join(',')}.join(' and ') } to secondary stack" if log.debug?
|
88
|
+
return true
|
89
|
+
else
|
90
|
+
log.debug "Pushing #{part_nodes.collect{|part| part.collect{|onode| onode.node.node_id}.join(',')}.join(' and ') } to main stack" if log.debug?
|
91
|
+
return false
|
92
|
+
end
|
93
|
+
end
|
94
|
+
|
95
|
+
stack = SingleCoherentPathsBetweenNodesFinder::DualStack.new &decide_stack
|
96
|
+
to_return = Bio::AssemblyGraphAlgorithms::TrailSet.new
|
97
|
+
|
98
|
+
# if there is no solutions to the overall problem then there is no solution at all
|
99
|
+
if problems.terminal_node_keys.nil? or problems.terminal_node_keys.empty?
|
100
|
+
to_return.trails = []
|
101
|
+
return to_return
|
102
|
+
end
|
103
|
+
|
104
|
+
# push all "ending in the final node" solutions to the stack
|
105
|
+
problems.terminal_node_keys.each do |key|
|
106
|
+
overall_solution = problems[key]
|
107
|
+
first_part = overall_solution.known_paths[0].copy
|
108
|
+
second_part = ORFsTracingTrail.new
|
109
|
+
second_part.otrail = Bio::Velvet::Graph::OrientedNodeTrail.new
|
110
|
+
stack.push [first_part, second_part]
|
111
|
+
end
|
112
|
+
|
113
|
+
all_paths_hash = {}
|
114
|
+
while path_parts = stack.pop
|
115
|
+
first_part = path_parts[0]
|
116
|
+
second_part = path_parts[1]
|
117
|
+
log.debug "#{first_part.to_shorthand} and #{second_part.otrail.to_shorthand}" if log.debug?
|
118
|
+
|
119
|
+
# Look for codons
|
120
|
+
log.debug "Searching for codons in first node of second part" if log.debug?
|
121
|
+
fwd_result, twin_result = search_for_codons(second_part.otrail) # search from start of second part
|
122
|
+
|
123
|
+
# Forward direction
|
124
|
+
if not fwd_result.stop_markers.empty? or not fwd_result.start_markers.empty?
|
125
|
+
[fwd_result.stop_markers, fwd_result.start_markers].each do |markers|
|
126
|
+
markers.each do |marker|
|
127
|
+
marker.position_in_trail = marker.position_in_node
|
128
|
+
end
|
129
|
+
end
|
130
|
+
current_fwd_stops = []
|
131
|
+
current_fwd_starts = []
|
132
|
+
if second_part.fwd_orfs_result
|
133
|
+
current_fwd_stops.concat second_part.fwd_orfs_result.initial_stop_markers
|
134
|
+
current_fwd_starts.concat second_part.fwd_orfs_result.initial_start_markers
|
135
|
+
current_fwd_starts.concat second_part.fwd_orfs_result.final_start_markers
|
136
|
+
end
|
137
|
+
current_fwd_stops.concat fwd_result.stop_markers
|
138
|
+
current_fwd_starts.concat fwd_result.start_markers
|
139
|
+
log.debug "Attempt to pair start codons at #{current_fwd_starts.collect{|m| m.position_in_trail}.join(',')} with stop codons at #{current_fwd_stops.collect{|m| m.position_in_trail}.join(',')}" if log.debug?
|
140
|
+
fwd_orfs_result = orfs_from_start_stop_markers(current_fwd_starts, current_fwd_stops, min_orf_length)
|
141
|
+
log.debug "Found pairs #{fwd_orfs_result.start_stop_pairs.collect{|pair| pair.collect{|m| m.position_in_trail}.join(',')}.join('],[')}" if log.debug?
|
142
|
+
|
143
|
+
# collect previous start-stop pairs
|
144
|
+
if second_part.fwd_orfs_result
|
145
|
+
fwd_orfs_result.start_stop_pairs.concat second_part.fwd_orfs_result.start_stop_pairs
|
146
|
+
end
|
147
|
+
second_part.fwd_orfs_result = fwd_orfs_result
|
148
|
+
log.debug "Remaining forward stops: #{second_part.fwd_orfs_result.initial_stop_markers.collect{|m| m.position_in_trail}.join(',')}" if log.debug?
|
149
|
+
end
|
150
|
+
|
151
|
+
# Reverse direction
|
152
|
+
if not twin_result.stop_markers.empty? or not twin_result.start_markers.empty?
|
153
|
+
# twin stop positons are relative to start of first path twin node
|
154
|
+
# add length of rest of path to get position relative to start of last path twin node
|
155
|
+
length_of_rest_of_path = second_part.otrail.length_in_bp_within_path - second_part.otrail[0].node.length_alone
|
156
|
+
[twin_result.stop_markers, twin_result.start_markers].each do |markers|
|
157
|
+
markers.each do |marker|
|
158
|
+
marker.position_in_trail = marker.position_in_node + length_of_rest_of_path
|
159
|
+
end
|
160
|
+
end
|
161
|
+
current_twin_stops = []
|
162
|
+
current_twin_starts = []
|
163
|
+
if second_part.twin_orfs_result
|
164
|
+
current_twin_stops.concat second_part.twin_orfs_result.initial_stop_markers
|
165
|
+
current_twin_starts.concat second_part.twin_orfs_result.initial_start_markers
|
166
|
+
current_twin_starts.concat second_part.twin_orfs_result.final_start_markers
|
167
|
+
end
|
168
|
+
current_twin_stops.concat twin_result.stop_markers
|
169
|
+
current_twin_starts.concat twin_result.start_markers
|
170
|
+
log.debug "Attempt to pair stop codons in reverse direction at #{current_twin_stops.collect{|m| m.position_in_trail}.join(',')} with starts at #{current_twin_starts.collect{|m| m.position_in_trail}.join(',')}" if log.debug?
|
171
|
+
twin_orfs_result = orfs_from_start_stop_markers(current_twin_starts, current_twin_stops, min_orf_length)
|
172
|
+
log.debug "Found pairs #{twin_orfs_result.start_stop_pairs.collect{|pair| pair.collect{|m| m.position_in_trail}.join(',')}.join('],[')}" if log.debug?
|
173
|
+
|
174
|
+
# collect previous start-stop pairs
|
175
|
+
if second_part.twin_orfs_result
|
176
|
+
twin_orfs_result.start_stop_pairs.concat second_part.twin_orfs_result.start_stop_pairs
|
177
|
+
end
|
178
|
+
second_part.twin_orfs_result = twin_orfs_result
|
179
|
+
log.debug "Remaining twin starts: #{second_part.twin_orfs_result.final_start_markers.collect{|m| m.position_in_trail}.join(',')}" if log.debug?
|
180
|
+
end
|
181
|
+
|
182
|
+
if first_part.length == 0
|
183
|
+
# If we've tracked all the way to the beginning, then there's no need to track further
|
184
|
+
|
185
|
+
key = second_part.otrail.trail.hash
|
186
|
+
all_paths_hash[key] ||= second_part
|
187
|
+
next
|
188
|
+
end
|
189
|
+
|
190
|
+
last = first_part.last
|
191
|
+
if second_part.otrail.trail.include? last
|
192
|
+
log.debug "Cycle at node #{last.node_id} detected in previous path #{second_part.collect{|onode| onode.node_id}.join(',')}." if log.debug?
|
193
|
+
to_return.circular_paths_detected = true
|
194
|
+
if max_cycles == 0 or max_cycles < counter.path_cycle_count([last, second_part.otrail.trail].flatten)
|
195
|
+
log.debug "Not finishing cyclic path with too many repeated cycles." if log.debug?
|
196
|
+
next
|
197
|
+
end
|
198
|
+
end
|
199
|
+
|
200
|
+
paths_to_last = problems[path_to_settable(first_part)].known_paths
|
201
|
+
paths_to_last.each do |path|
|
202
|
+
new_second_part = ORFsTracingTrail.new
|
203
|
+
new_second_part.otrail = second_part.otrail.copy
|
204
|
+
new_second_part.otrail.trail.unshift last
|
205
|
+
|
206
|
+
if second_part.fwd_orfs_result
|
207
|
+
# offset positions in forward direction
|
208
|
+
offset = last.node.length_alone
|
209
|
+
copy_and_offset_marker = lambda do |marker|
|
210
|
+
m = marker.copy
|
211
|
+
m.position_in_trail += offset
|
212
|
+
m
|
213
|
+
end
|
214
|
+
|
215
|
+
new_fwd_orfs_result = ORFsResult.new
|
216
|
+
new_fwd_orfs_result.start_stop_pairs = second_part.fwd_orfs_result.start_stop_pairs.collect do |pairs|
|
217
|
+
pairs.collect ©_and_offset_marker
|
218
|
+
end
|
219
|
+
new_fwd_orfs_result.initial_start_markers = second_part.fwd_orfs_result.initial_start_markers.collect ©_and_offset_marker
|
220
|
+
new_fwd_orfs_result.initial_stop_markers = second_part.fwd_orfs_result.initial_stop_markers.collect ©_and_offset_marker
|
221
|
+
new_fwd_orfs_result.final_start_markers = second_part.fwd_orfs_result.final_start_markers.collect ©_and_offset_marker
|
222
|
+
new_second_part.fwd_orfs_result = new_fwd_orfs_result
|
223
|
+
end
|
224
|
+
|
225
|
+
if second_part.twin_orfs_result
|
226
|
+
new_twin_orfs_result = ORFsResult.new
|
227
|
+
new_twin_orfs_result.start_stop_pairs = second_part.twin_orfs_result.start_stop_pairs.collect do |pairs|
|
228
|
+
pairs.collect{|marker| marker.copy}
|
229
|
+
end
|
230
|
+
new_twin_orfs_result.initial_stop_markers = second_part.twin_orfs_result.initial_stop_markers.collect{|marker| marker.copy}
|
231
|
+
new_twin_orfs_result.final_start_markers = second_part.twin_orfs_result.final_start_markers.collect{|marker| marker.copy}
|
232
|
+
new_second_part.twin_orfs_result = new_twin_orfs_result
|
233
|
+
end
|
234
|
+
|
235
|
+
new_first_part = path.copy
|
236
|
+
new_first_part.remove_last_node
|
237
|
+
|
238
|
+
stack.push [new_first_part, new_second_part]
|
239
|
+
end
|
240
|
+
|
241
|
+
# max_num_paths parachute
|
242
|
+
# The parachute can kill the search once the main stack exceeds max_gapfill_paths,
|
243
|
+
# since all paths on it are valid.
|
244
|
+
if !max_num_paths.nil? and (stack.sizes[0] + all_paths_hash.length) > max_num_paths
|
245
|
+
log.info "Exceeded the maximum number of allowable paths in this gapfill" if log.info?
|
246
|
+
to_return.max_path_limit_exceeded = true
|
247
|
+
all_paths_hash = {}
|
248
|
+
break
|
249
|
+
end
|
250
|
+
end
|
251
|
+
|
252
|
+
to_return.trails = all_paths_hash.values
|
253
|
+
return to_return
|
254
|
+
end
|
255
|
+
|
256
|
+
# Returns:
|
257
|
+
# SearchResult relative to start of first node
|
258
|
+
# SearchResult relative to start of first twin node
|
259
|
+
def search_for_codons(otrail)
|
260
|
+
return SearchResult.new, SearchResult.new if otrail.trail.empty?
|
261
|
+
onode = otrail[0]
|
262
|
+
|
263
|
+
make_marker = lambda do |position|
|
264
|
+
marker = Marker.new
|
265
|
+
marker.position_in_node = position
|
266
|
+
marker.node = onode.node
|
267
|
+
marker
|
268
|
+
end
|
269
|
+
|
270
|
+
#log.debug "Looking for codons #{words.to_a}" if log.debug?
|
271
|
+
words = Set.new(START_CODONS).merge(STOP_CODONS)
|
272
|
+
|
273
|
+
# search within first / last node
|
274
|
+
fwd_nodes_sequence, twin_nodes_sequence = get_sequences onode
|
275
|
+
#log.debug "Looking in #{fwd_nodes_sequence}" if log.debug?
|
276
|
+
fwd_within_first = word_search(fwd_nodes_sequence, words, CODON_LENGTH)
|
277
|
+
#log.debug "Found codons #{fwd_within_first.keys.join(',')} at positions #{fwd_within_first.values.join(',')} in #{fwd_nodes_sequence}" if log.debug?
|
278
|
+
#log.debug "Looking in #{twin_nodes_sequence}" if log.debug?
|
279
|
+
twin_within_first = word_search(twin_nodes_sequence, words, CODON_LENGTH)
|
280
|
+
#log.debug "Found codons #{twin_within_first.keys.join(',')} in twin node at positions #{twin_within_first.values.join(',')} in #{fwd_nodes_sequence}" if log.debug?
|
281
|
+
|
282
|
+
# extend search along trail
|
283
|
+
fwd_overlap_sequence, twin_overlap_sequence = get_overlap_sequences(otrail, CODON_LENGTH)
|
284
|
+
#log.debug "Looking in #{fwd_overlap_sequence}" if log.debug?
|
285
|
+
fwd_in_overlap = word_search(fwd_overlap_sequence, words, CODON_LENGTH)
|
286
|
+
#log.debug "Found codons #{fwd_in_overlap.keys.join(',')} in twin node at positions #{fwd_in_overlap.values.join(',')} in #{fwd_overlap_sequence}" if log.debug?
|
287
|
+
#log.debug "Looking for stops in #{twin_overlap_sequence}" if log.debug?
|
288
|
+
twin_in_overlap = word_search(twin_overlap_sequence, words, CODON_LENGTH)
|
289
|
+
#log.debug "Found codons #{twin_in_overlap.keys.join(',')} in twin node at positions #{twin_in_overlap.values.join(',')} in #{twin_overlap_sequence}" if log.debug?
|
290
|
+
|
291
|
+
# offset positions in overlap to be relative to start of node / twin node
|
292
|
+
offset = onode.node.length_alone
|
293
|
+
fwd_in_overlap.each{|word, inds| fwd_in_overlap[word] = inds.collect{|pos| pos + offset}}
|
294
|
+
twin_in_overlap.each{|word, inds| twin_in_overlap[word] = inds.collect{|pos| pos + 1 - CODON_LENGTH}}
|
295
|
+
#log.debug "Codons in overlap positions relative to start of first node #{fwd_in_overlap.values.join(',')}" if log.debug?
|
296
|
+
#log.debug "Codons in overlap positions relative to start of first twin node #{twin_in_overlap.values.join(',')}" if log.debug?
|
297
|
+
|
298
|
+
# assemble result
|
299
|
+
fwd_result = SearchResult.new
|
300
|
+
twin_result = SearchResult.new
|
301
|
+
|
302
|
+
push_mark_to_list = lambda do |list, word, positions|
|
303
|
+
if positions.has_key? word
|
304
|
+
list.push positions[word].collect{|pos| make_marker.call pos}
|
305
|
+
end
|
306
|
+
end
|
307
|
+
|
308
|
+
fwd_positions = fwd_within_first.merge fwd_in_overlap
|
309
|
+
twin_positions = twin_within_first.merge twin_in_overlap
|
310
|
+
START_CODONS.each do |word|
|
311
|
+
# fwd starts
|
312
|
+
push_mark_to_list.call(fwd_result.start_markers, word, fwd_positions)
|
313
|
+
# twin starts
|
314
|
+
push_mark_to_list.call(twin_result.start_markers, word, twin_positions)
|
315
|
+
end
|
316
|
+
fwd_result.start_markers.flatten!
|
317
|
+
twin_result.start_markers.flatten!
|
318
|
+
#log.debug "Positions of start codons #{fwd_result.start_markers.join(',')}" if log.debug?
|
319
|
+
#log.debug "Positions of start codons in twin node #{twin_result.start_markers.join(',')}" if log.debug?
|
320
|
+
|
321
|
+
STOP_CODONS.each do |word|
|
322
|
+
#fwd stops
|
323
|
+
push_mark_to_list.call(fwd_result.stop_markers, word, fwd_positions)
|
324
|
+
# twin stops
|
325
|
+
push_mark_to_list.call(twin_result.stop_markers, word, twin_positions)
|
326
|
+
end
|
327
|
+
fwd_result.stop_markers.flatten!
|
328
|
+
twin_result.stop_markers.flatten!
|
329
|
+
#log.debug "Positions of stop codons #{fwd_result.stop_markers.join(',')}" if log.debug?
|
330
|
+
#log.debug "Positions of stop codons in twin node #{twin_result.stop_markers.join(',')}" if log.debug?
|
331
|
+
|
332
|
+
|
333
|
+
return fwd_result, twin_result
|
334
|
+
end
|
335
|
+
|
336
|
+
def get_overlap_sequences(otrail, size, from_end=false)
|
337
|
+
return '' if otrail.trail.empty?
|
338
|
+
trail = otrail.trail
|
339
|
+
if from_end # reverse as new sequence is taken from front of trail
|
340
|
+
trail = trail.reverse
|
341
|
+
end
|
342
|
+
twin_nodes_sequence = ''
|
343
|
+
fwd_nodes_sequence = ''
|
344
|
+
|
345
|
+
index = 0
|
346
|
+
onode = trail[index]
|
347
|
+
|
348
|
+
start_length = onode.node.length_alone
|
349
|
+
extension_length = -start_length
|
350
|
+
|
351
|
+
while extension_length < (size - 1) and index < trail.length
|
352
|
+
#log.debug "Extended #{extension_length} / #{size} bps and #{index+1} / #{otrail.length} nodes" if log.debug?
|
353
|
+
extend_fwd_nodes_sequence, extend_twin_nodes_sequence = get_sequences(onode)
|
354
|
+
if from_end
|
355
|
+
twin_nodes_sequence += extend_twin_nodes_sequence
|
356
|
+
fwd_nodes_sequence = extend_fwd_nodes_sequence + fwd_nodes_sequence
|
357
|
+
else
|
358
|
+
twin_nodes_sequence = extend_twin_nodes_sequence + twin_nodes_sequence
|
359
|
+
fwd_nodes_sequence += extend_fwd_nodes_sequence
|
360
|
+
end
|
361
|
+
|
362
|
+
extension_length += onode.node.length_alone
|
363
|
+
index += 1
|
364
|
+
onode = trail[index]
|
365
|
+
end
|
366
|
+
|
367
|
+
#log.debug "Found forward and twin sequences #{fwd_nodes_sequence} and #{twin_nodes_sequence} before trimming" if log.debug?
|
368
|
+
|
369
|
+
trim_start = start_length - (size - 1)
|
370
|
+
trim_start = 0 if trim_start < 0
|
371
|
+
trim_end = extension_length - (size - 1)
|
372
|
+
trim_end = 0 if trim_end < 0
|
373
|
+
#log.debug "Trimming first #{trim_start} and last #{trim_end} positions for output" if log.debug?
|
374
|
+
if from_end
|
375
|
+
return fwd_nodes_sequence[trim_end..-(trim_start+1)], twin_nodes_sequence[trim_start..-(trim_end+1)]
|
376
|
+
else
|
377
|
+
return fwd_nodes_sequence[trim_start..-(trim_end+1)], twin_nodes_sequence[trim_end..-(trim_start+1)]
|
378
|
+
end
|
379
|
+
end
|
380
|
+
|
381
|
+
def get_sequences(onode)
|
382
|
+
if onode.starts_at_start?
|
383
|
+
twin_nodes_sequence = onode.node.ends_of_kmers_of_twin_node
|
384
|
+
fwd_nodes_sequence = onode.node.ends_of_kmers_of_node
|
385
|
+
else
|
386
|
+
twin_nodes_sequence = onode.node.ends_of_kmers_of_node
|
387
|
+
fwd_nodes_sequence = onode.node.ends_of_kmers_of_twin_node
|
388
|
+
end
|
389
|
+
return fwd_nodes_sequence, twin_nodes_sequence
|
390
|
+
end
|
391
|
+
|
392
|
+
def word_search(sequence, words, size)
|
393
|
+
position = size
|
394
|
+
inds = {}
|
395
|
+
|
396
|
+
while position <= sequence.length
|
397
|
+
word = sequence[position-size...position]
|
398
|
+
if words.include? word
|
399
|
+
inds[word] ||=[]
|
400
|
+
inds[word].push position
|
401
|
+
end
|
402
|
+
position += 1
|
403
|
+
end
|
404
|
+
|
405
|
+
return inds
|
406
|
+
end
|
407
|
+
|
408
|
+
|
409
|
+
# Given an Array of start positions and stop positions, return
|
410
|
+
# start,stop base position pairs (not inclusive of the stop codon's bases)
|
411
|
+
# that are ORFs with a given minimum orf length (length measured in nucleotides).
|
412
|
+
# The returned object is an instance of ORFsResult.
|
413
|
+
def orfs_from_start_stop_markers(start_markers, stop_markers, minimum_orf_length)
|
414
|
+
# Split up the start and stop positions into 3 frames
|
415
|
+
frame_starts = [[],[],[]]
|
416
|
+
frame_stops = [[],[],[]]
|
417
|
+
start_markers.each do |marker|
|
418
|
+
frame_starts[marker.position_in_trail % 3].push marker
|
419
|
+
end
|
420
|
+
stop_markers.each do |marker|
|
421
|
+
frame_stops[marker.position_in_trail % 3].push marker
|
422
|
+
end
|
423
|
+
|
424
|
+
# For each frame
|
425
|
+
to_return = ORFsResult.new
|
426
|
+
(0..2).each do |frame|
|
427
|
+
frame_pairs = []
|
428
|
+
|
429
|
+
# Sort arrays in descending order because Array#pop removes from the end of the array
|
430
|
+
starts = frame_starts[frame].sort{|a,b| b.position_in_trail<=>a.position_in_trail}
|
431
|
+
stops = frame_stops[frame].sort{|a,b| b.position_in_trail<=>a.position_in_trail}
|
432
|
+
|
433
|
+
current_start = starts.pop
|
434
|
+
current_stop = stops.pop
|
435
|
+
if current_stop
|
436
|
+
# Record first stop codon
|
437
|
+
to_return.initial_stop_markers.push current_stop
|
438
|
+
end
|
439
|
+
if current_start and (current_stop.nil? or current_start.position_in_trail < current_stop.position_in_trail)
|
440
|
+
# Record first start codon before any stop codons
|
441
|
+
to_return.initial_start_markers.push current_start
|
442
|
+
end
|
443
|
+
|
444
|
+
while current_start and current_stop
|
445
|
+
# Move to next start after current stop
|
446
|
+
while current_start and current_start.position_in_trail < current_stop.position_in_trail
|
447
|
+
current_start = starts.pop
|
448
|
+
end
|
449
|
+
|
450
|
+
if current_start
|
451
|
+
# Move to next stop after current start
|
452
|
+
while current_stop and current_stop.position_in_trail < current_start.position_in_trail
|
453
|
+
current_stop = stops.pop
|
454
|
+
end
|
455
|
+
end
|
456
|
+
|
457
|
+
if current_start and current_stop
|
458
|
+
# This stop codon stops the current reading frame.
|
459
|
+
if current_stop.position_in_trail - current_start.position_in_trail >= minimum_orf_length
|
460
|
+
# Found a legit ORF
|
461
|
+
to_return.start_stop_pairs.push [current_start, current_stop]
|
462
|
+
end
|
463
|
+
# Whether or not last ORF was long enough, search for the next start codon
|
464
|
+
next
|
465
|
+
else
|
466
|
+
if current_start
|
467
|
+
to_return.final_start_markers.push current_start
|
468
|
+
end
|
469
|
+
break
|
470
|
+
end
|
471
|
+
end
|
472
|
+
end
|
473
|
+
|
474
|
+
return to_return
|
475
|
+
end
|
476
|
+
|
477
|
+
def orf_sequences_from_trails(trails)
|
478
|
+
to_return = {}
|
479
|
+
trails.each do |trail|
|
480
|
+
fwd_sequence, twin_sequence = trail.otrail.sequences_within_path
|
481
|
+
# forward / twin directions
|
482
|
+
[
|
483
|
+
[fwd_sequence, trail.fwd_orfs_result],
|
484
|
+
[twin_sequence, trail.twin_orfs_result]
|
485
|
+
].each do |sequence_and_result|
|
486
|
+
sequence, result = sequence_and_result
|
487
|
+
if result
|
488
|
+
result.start_stop_pairs.each do |pair|
|
489
|
+
start_position = pair[0].position_in_trail - 3
|
490
|
+
end_position = pair[1].position_in_trail
|
491
|
+
|
492
|
+
# orf name
|
493
|
+
last_node = nil
|
494
|
+
onodes = trail.otrail.trail.drop_while do |onode|
|
495
|
+
onode.node != pair[0].node
|
496
|
+
end.take_while do |onode|
|
497
|
+
next false if last_node == pair[1].node
|
498
|
+
last_node = onode.node
|
499
|
+
true
|
500
|
+
end
|
501
|
+
name = "(#{onodes[0].to_shorthand}:#{pair[0].position_in_node}),#{onodes[1...-1].collect{|onode| onode.to_shorthand}.join(',')},(#{onodes[-1].to_shorthand}:#{pair[1].position_in_node})"
|
502
|
+
|
503
|
+
to_return[name] ||= sequence[start_position...end_position]
|
504
|
+
end
|
505
|
+
result.initial_stop_markers.each do |marker|
|
506
|
+
end_position = marker.position_in_trail
|
507
|
+
|
508
|
+
# orf_name
|
509
|
+
last_node = nil
|
510
|
+
onodes = trail.otrail.trail.take_while do |onode|
|
511
|
+
next false if last_node == marker.node
|
512
|
+
last_node = onode.node
|
513
|
+
true
|
514
|
+
end
|
515
|
+
name = "#{onodes[0...-1].collect{|onode| onode.to_shorthand}.join(',')},(#{onodes[-1].to_shorthand}:#{marker.position_in_node})"
|
516
|
+
|
517
|
+
to_return[name] ||= sequence[0...end_position]
|
518
|
+
end
|
519
|
+
result.final_start_markers.each do |marker|
|
520
|
+
start_position = marker.position_in_trail - 3
|
521
|
+
|
522
|
+
# orf_name
|
523
|
+
onodes = trail.otrail.trail.drop_while{|onode| onode.node != marker.node}
|
524
|
+
name = "(#{onodes[0].to_shorthand}:#{marker.position_in_node}),#{onodes[1..-1].collect{|onode| onode.to_shorthand}.join(',')}"
|
525
|
+
end
|
526
|
+
end
|
527
|
+
if result.nil? or (result.start_stop_pairs.empty? and result.final_start_markers.empty? and result.initial_stop_markers.empty?)
|
528
|
+
name = "#{trail.otrail.to_shorthand}"
|
529
|
+
|
530
|
+
to_return[name] ||= sequence
|
531
|
+
end
|
532
|
+
end
|
533
|
+
end
|
534
|
+
|
535
|
+
return to_return
|
536
|
+
end
|
537
|
+
|
538
|
+
# positions of last base of codons
|
539
|
+
class Marker
|
540
|
+
attr_accessor :position_in_trail, :position_in_node, :node
|
541
|
+
|
542
|
+
def copy
|
543
|
+
copy = Marker.new
|
544
|
+
copy.position_in_trail = @position_in_trail
|
545
|
+
copy.position_in_node = @position_in_node
|
546
|
+
copy.node = @node
|
547
|
+
return copy
|
548
|
+
end
|
549
|
+
end
|
550
|
+
|
551
|
+
class SearchResult
|
552
|
+
attr_accessor :start_markers, :stop_markers
|
553
|
+
|
554
|
+
def initialize
|
555
|
+
@start_markers = []
|
556
|
+
@stop_markers = []
|
557
|
+
end
|
558
|
+
end
|
559
|
+
|
560
|
+
class ORFsTracingTrail
|
561
|
+
attr_accessor :otrail, :fwd_orfs_result, :twin_orfs_result
|
562
|
+
include Enumerable
|
563
|
+
|
564
|
+
def each(&block)
|
565
|
+
unless @otrail.nil?
|
566
|
+
@otrail.each(&block)
|
567
|
+
end
|
568
|
+
end
|
569
|
+
end
|
570
|
+
|
571
|
+
class ORFsResult
|
572
|
+
attr_accessor :start_stop_pairs, :final_start_markers, :initial_start_markers, :initial_stop_markers
|
573
|
+
|
574
|
+
def initialize
|
575
|
+
@start_stop_pairs = []
|
576
|
+
@initial_start_markers = []
|
577
|
+
@final_start_markers = []
|
578
|
+
@initial_stop_markers = []
|
579
|
+
end
|
580
|
+
end
|
581
|
+
|
582
|
+
class AllProblemTrailsFinder
|
583
|
+
include Bio::FinishM::Logging
|
584
|
+
|
585
|
+
def initialize(graph, initial_paths)
|
586
|
+
@stack = DS::Stack.new
|
587
|
+
initial_paths.each do |path|
|
588
|
+
@stack.push path
|
589
|
+
end
|
590
|
+
@graph = graph
|
591
|
+
end
|
592
|
+
|
593
|
+
def pop
|
594
|
+
@stack.pop
|
595
|
+
end
|
596
|
+
|
597
|
+
def size
|
598
|
+
@stack.size
|
599
|
+
end
|
600
|
+
|
601
|
+
def push_next_neighbours(current_path)
|
602
|
+
next_nodes = current_path.neighbours_of_last_node(@graph)
|
603
|
+
log.debug "Pushing #{next_nodes.length} new neighbours of #{current_path.last}" if log.debug?
|
604
|
+
#TODO: not neccessary to copy all paths, can just continue one of them
|
605
|
+
next_nodes.each do |n|
|
606
|
+
log.debug "Pushing neighbour to stack: #{n}" if log.debug?
|
607
|
+
path = current_path.copy
|
608
|
+
path.add_oriented_node n
|
609
|
+
@stack.push path
|
610
|
+
end
|
611
|
+
end
|
612
|
+
end
|
613
|
+
end
|
614
|
+
end
|
615
|
+
end
|