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 +1,126 @@
|
|
1
|
+
require 'bio'
|
2
|
+
|
3
|
+
class Bio::FinishM::ScaffoldBreaker
|
4
|
+
include Bio::FinishM::Logging
|
5
|
+
|
6
|
+
class UnscaffoldedContig
|
7
|
+
attr_accessor :scaffold_position_start, :scaffold_position_end
|
8
|
+
|
9
|
+
# The Scaffold to which this contig once belonged
|
10
|
+
attr_accessor :scaffold
|
11
|
+
|
12
|
+
# The actual nucleotide sequence of this contig, from scaffold start position to
|
13
|
+
# end (not revcom)
|
14
|
+
attr_accessor :sequence
|
15
|
+
|
16
|
+
def length
|
17
|
+
@scaffold_position_end - @scaffold_position_start +1
|
18
|
+
end
|
19
|
+
|
20
|
+
def name
|
21
|
+
contig_number = scaffold.contigs.find_index(self)+1
|
22
|
+
if contig_number.nil?
|
23
|
+
raise "A contig finds itself unexpectedly not in the scaffold it is supposed to belong to"
|
24
|
+
end
|
25
|
+
return "#{scaffold.name}_#{contig_number}of#{scaffold.contigs.length}_#{scaffold_position_start}to#{scaffold_position_end}"
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
class Scaffold
|
30
|
+
# unscaffolded contigs from this scaffold, as an array in sorted order.
|
31
|
+
attr_accessor :contigs
|
32
|
+
|
33
|
+
# Name of sequence found in the fasta file
|
34
|
+
attr_accessor :name
|
35
|
+
|
36
|
+
# Return an array of Gap objects
|
37
|
+
def gaps
|
38
|
+
gaps = []
|
39
|
+
last_contig = nil
|
40
|
+
@contigs.each_with_index do |contig, i|
|
41
|
+
if i!=0
|
42
|
+
gap = Bio::FinishM::ScaffoldBreaker::Gap.new
|
43
|
+
gap.scaffold = self
|
44
|
+
gap.start = last_contig.scaffold_position_end + 1
|
45
|
+
gap.stop = contig.scaffold_position_start - 1
|
46
|
+
gap.number = i-1
|
47
|
+
gaps.push gap
|
48
|
+
end
|
49
|
+
last_contig = contig
|
50
|
+
end
|
51
|
+
return gaps
|
52
|
+
end
|
53
|
+
|
54
|
+
def sequence
|
55
|
+
to_return = []
|
56
|
+
last_contig = nil
|
57
|
+
@contigs.each_with_index do |contig, i|
|
58
|
+
if i==0
|
59
|
+
to_return.push contig.sequence
|
60
|
+
else
|
61
|
+
gap_start = last_contig.scaffold_position_end + 1
|
62
|
+
gap_stop = contig.scaffold_position_start - 1
|
63
|
+
to_return.push 'N'*(gap_stop-gap_start+1)
|
64
|
+
to_return.push contig.sequence
|
65
|
+
end
|
66
|
+
last_contig = contig
|
67
|
+
end
|
68
|
+
return to_return.join
|
69
|
+
end
|
70
|
+
|
71
|
+
# Which contig number is this, in the scaffold?
|
72
|
+
def contig_number(contig)
|
73
|
+
@contigs.each_with_index do |current_contig, i|
|
74
|
+
return i if contig==current_contig
|
75
|
+
end
|
76
|
+
raise "Contig not found in scaffold"
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
class Gap
|
81
|
+
attr_accessor :scaffold, :start, :stop, :number
|
82
|
+
|
83
|
+
def coords
|
84
|
+
@scaffold.name+':'+(@start+1).to_s+'-'+(@stop).to_s
|
85
|
+
end
|
86
|
+
|
87
|
+
#i.e. the number of N characters that would represent this gap
|
88
|
+
def length
|
89
|
+
@stop-@start+1
|
90
|
+
end
|
91
|
+
end
|
92
|
+
|
93
|
+
# Given a path to a scaffold fasta file, read in the scaffolds, and break them apart
|
94
|
+
# into constituent contigs. Then return an array of Scaffold objects containing the
|
95
|
+
# contig information therein.
|
96
|
+
def break_scaffolds(contigs_filename)
|
97
|
+
scaffolds = []
|
98
|
+
Bio::FlatFile.foreach(Bio::FastaFormat, contigs_filename) do |seq|
|
99
|
+
scaffold = Scaffold.new
|
100
|
+
scaffold.name = seq.definition
|
101
|
+
|
102
|
+
unless seq.seq.match(/^[ATGCN]+$/i)
|
103
|
+
example = seq.seq.match(/([^ATGCN])/i)[1]
|
104
|
+
log.warn "Found unexpected characters in the sequence #{seq.definition} e.g. #{example}, continuing optimistically, but not quite sure what will happen.. good luck"
|
105
|
+
end
|
106
|
+
|
107
|
+
if seq.seq.match(/^N+$/i)
|
108
|
+
raise "Found a scaffold that contains all N characters, ignoring this (perhaps your input is mangled?): #{scaffold.name}"
|
109
|
+
end
|
110
|
+
|
111
|
+
# Find all Ns in the current sequence
|
112
|
+
seq.seq.scan(/([^N]+)/i) do
|
113
|
+
contig = UnscaffoldedContig.new
|
114
|
+
contig.scaffold = scaffold
|
115
|
+
contig.scaffold_position_start = $~.offset(0)[0]+1#Convert to 1-based indices in line with bioruby
|
116
|
+
contig.scaffold_position_end = $~.offset(0)[1]
|
117
|
+
contig.sequence = $~.to_s
|
118
|
+
scaffold.contigs ||= []
|
119
|
+
scaffold.contigs.push contig
|
120
|
+
end
|
121
|
+
scaffolds.push scaffold
|
122
|
+
end
|
123
|
+
log.info "Detected #{scaffolds.length} scaffolds, containing #{scaffolds.collect{|s| s.contigs.length}.reduce(:+)} different contigs"
|
124
|
+
return scaffolds
|
125
|
+
end
|
126
|
+
end
|
@@ -0,0 +1,71 @@
|
|
1
|
+
require 'ds'
|
2
|
+
require 'set'
|
3
|
+
|
4
|
+
class Bio::AssemblyGraphAlgorithms::SequenceHasher
|
5
|
+
include Bio::FinishM::Logging
|
6
|
+
|
7
|
+
#
|
8
|
+
def extend_overlap(graph, oriented_onode, overlap, options={})
|
9
|
+
trails = []
|
10
|
+
|
11
|
+
current_path = DistancedOrientedNodeTrail.new
|
12
|
+
current_path.add_oriented_node oriented_onode
|
13
|
+
current_path.distance = 0
|
14
|
+
|
15
|
+
stack = DS::Stack.new
|
16
|
+
stack.push current_path
|
17
|
+
|
18
|
+
# While there is more on the stack
|
19
|
+
while current_path = stack.pop
|
20
|
+
|
21
|
+
current_distance = current_path.distance
|
22
|
+
|
23
|
+
if current_distance >= overlap
|
24
|
+
# Found all the sequence we need
|
25
|
+
trails.push current_path
|
26
|
+
next
|
27
|
+
end
|
28
|
+
|
29
|
+
# Find neighbouring nodes
|
30
|
+
neighbours = nil
|
31
|
+
if options[:neighbour_finder]
|
32
|
+
neighbours = options[:neighbour_finder].neighbours(oriented_onode)
|
33
|
+
else
|
34
|
+
neighbours = oriented_node.next_neighbours(graph)
|
35
|
+
end
|
36
|
+
|
37
|
+
neighbours.each do |onode|
|
38
|
+
new_distance = current_distance
|
39
|
+
if options[:neighbour_finder]
|
40
|
+
if onode.distance
|
41
|
+
new_distance += onode.distance
|
42
|
+
else
|
43
|
+
new_distance += 0
|
44
|
+
end
|
45
|
+
end
|
46
|
+
new_distance += onode.node.length_alone
|
47
|
+
|
48
|
+
new_path = current_path.copy
|
49
|
+
new_path.add_oriented_node onode
|
50
|
+
new_path.distance = new_distance
|
51
|
+
stack.push new_path
|
52
|
+
end
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
class DistancedOrientedNodeTrail < Bio::Velvet::Graph::OrientedNodeTrail
|
57
|
+
attr_accessor :distance
|
58
|
+
|
59
|
+
def copy
|
60
|
+
o = DistancedOrientedNodeTrail.new
|
61
|
+
o.trail = Array.new(@trail.collect{|onode| onode.copy})
|
62
|
+
o.distance = @distance
|
63
|
+
return o
|
64
|
+
end
|
65
|
+
|
66
|
+
def to_s
|
67
|
+
"DistancedOrientedTrail: #{object_id}: #{to_shorthand} distance=#{@distance}"
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
end
|
@@ -0,0 +1,533 @@
|
|
1
|
+
require 'ds'
|
2
|
+
require 'set'
|
3
|
+
|
4
|
+
class Bio::AssemblyGraphAlgorithms::SingleCoherentPathsBetweenNodesFinder
|
5
|
+
include Bio::FinishM::Logging
|
6
|
+
|
7
|
+
SINGLE_BASE_REVCOM = {
|
8
|
+
'A'=>'T',
|
9
|
+
'T'=>'A',
|
10
|
+
'G'=>'C',
|
11
|
+
'C'=>'G',
|
12
|
+
}
|
13
|
+
|
14
|
+
# Find all paths between the initial and terminal node in the graph.
|
15
|
+
# Don't search in the graph when the distance in base pairs exceeds the leash length.
|
16
|
+
# Recohere reads (singled ended only) in an attempt to remove bubbles.
|
17
|
+
#
|
18
|
+
# Options:
|
19
|
+
# * max_gapfill_paths: the maxmimum number of paths to return. If this maximum is exceeded, an empty solution set is returned
|
20
|
+
def find_all_connections_between_two_nodes(graph, initial_path, terminal_oriented_node,
|
21
|
+
leash_length, recoherence_kmer, sequence_hash, options={})
|
22
|
+
|
23
|
+
problems = find_all_problems(graph, initial_path, terminal_oriented_node, leash_length, recoherence_kmer, sequence_hash, options)
|
24
|
+
|
25
|
+
paths = find_paths_from_problems(problems, recoherence_kmer, options)
|
26
|
+
return paths
|
27
|
+
end
|
28
|
+
|
29
|
+
# Options:
|
30
|
+
#
|
31
|
+
# :max_explore_nodes: only explore this many nodes, not further.
|
32
|
+
def find_all_problems(graph, initial_path, terminal_node, leash_length, recoherence_kmer, sequence_hash, options={})
|
33
|
+
# setup dynamic programming cache
|
34
|
+
problems = ProblemSet.new
|
35
|
+
|
36
|
+
# setup stack to keep track of initial nodes
|
37
|
+
finder = ProblemTrailFinder.new(graph, initial_path)
|
38
|
+
|
39
|
+
#current_oriented_node_trail = Bio::Velvet::Graph::OrientedNodeTrail.new
|
40
|
+
#last_number_of_problems_observed_checkpoint = 0
|
41
|
+
|
42
|
+
while current_path = finder.dequeue
|
43
|
+
path_length = current_path.length_in_bp
|
44
|
+
log.debug "considering #{current_path}, path length: #{path_length}" if log.debug?
|
45
|
+
|
46
|
+
# Have we solved this before? If so, add this path to that solved problem.
|
47
|
+
set_key = path_to_settable current_path, recoherence_kmer
|
48
|
+
log.debug "Set key is #{set_key}" if log.debug?
|
49
|
+
|
50
|
+
# Unless the path validates, forget it.
|
51
|
+
if recoherence_kmer.nil?
|
52
|
+
# Continue, assume that it validates if there is no recoherence_kmer
|
53
|
+
elsif !validate_last_node_of_path_by_recoherence(current_path, recoherence_kmer, sequence_hash)
|
54
|
+
log.debug "Path did not validate, skipping" if log.debug?
|
55
|
+
next
|
56
|
+
elsif log.debug?
|
57
|
+
log.debug "Path validates"
|
58
|
+
end
|
59
|
+
|
60
|
+
if current_path.last == terminal_node
|
61
|
+
log.debug "last is terminal" if log.debug?
|
62
|
+
problems[set_key] ||= DynamicProgrammingProblem.new
|
63
|
+
problems[set_key].known_paths ||= []
|
64
|
+
problems[set_key].known_paths.push current_path
|
65
|
+
|
66
|
+
problems.terminal_node_keys ||= Set.new
|
67
|
+
problems.terminal_node_keys << set_key
|
68
|
+
|
69
|
+
elsif problems[set_key]
|
70
|
+
log.debug "Already seen this problem" if log.debug?
|
71
|
+
prob = problems[set_key]
|
72
|
+
prob.known_paths.push current_path
|
73
|
+
|
74
|
+
# If a lesser min distance is found, then we need to start exploring from the
|
75
|
+
# current place again
|
76
|
+
if path_length < prob.min_distance
|
77
|
+
log.debug "Found a node with min_distance greater than path length.." if log.debug?
|
78
|
+
prob.min_distance = path_length
|
79
|
+
finder.push_next_neighbours current_path
|
80
|
+
end
|
81
|
+
elsif !leash_length.nil? and path_length > leash_length
|
82
|
+
# we are past the leash length, give up
|
83
|
+
log.debug "Past leash length, giving up" if log.debug?
|
84
|
+
else
|
85
|
+
log.debug "New dynamic problem being solved" if log.debug?
|
86
|
+
# new problem being solved here
|
87
|
+
problem = DynamicProgrammingProblem.new
|
88
|
+
problem.min_distance = path_length
|
89
|
+
problem.known_paths.push current_path.copy
|
90
|
+
problems[set_key] = problem
|
91
|
+
|
92
|
+
num_done = problems.length
|
93
|
+
if num_done > 0 and num_done % 512 == 0
|
94
|
+
log.info "So far worked with #{num_done} head node sets, up to distance #{path_length}" if log.info?
|
95
|
+
end
|
96
|
+
if options[:max_explore_nodes] and num_done > options[:max_explore_nodes]
|
97
|
+
log.warn "Explored too many nodes (#{num_done}), giving up.."
|
98
|
+
problems = ProblemSet.new
|
99
|
+
break
|
100
|
+
end
|
101
|
+
|
102
|
+
# explore the forward neighbours
|
103
|
+
finder.push_next_neighbours current_path
|
104
|
+
end
|
105
|
+
log.debug "Priority queue size: #{finder.length}" if log.debug?
|
106
|
+
end
|
107
|
+
|
108
|
+
return problems
|
109
|
+
end
|
110
|
+
|
111
|
+
def path_to_settable(path, recoherence_kmer)
|
112
|
+
log.debug "Making settable a path: #{path}" if log.debug?
|
113
|
+
return array_trail_to_settable(path.trail, recoherence_kmer)
|
114
|
+
end
|
115
|
+
|
116
|
+
def array_trail_to_settable(trail, recoherence_kmer)
|
117
|
+
return trail.last.to_settable if recoherence_kmer.nil?
|
118
|
+
|
119
|
+
cumulative_length = 0
|
120
|
+
i = trail.length - 1
|
121
|
+
while i >= 0 and cumulative_length < recoherence_kmer
|
122
|
+
cumulative_length += trail[i].node.length_alone
|
123
|
+
i -= 1
|
124
|
+
end
|
125
|
+
i += 1
|
126
|
+
# 'Return' an array made up of the settables
|
127
|
+
to_return = trail[i..-1].collect{|t| t.to_settable}.flatten
|
128
|
+
log.debug "'Returning' settable version of path: #{to_return}" if log.debug?
|
129
|
+
to_return
|
130
|
+
end
|
131
|
+
|
132
|
+
# Given an OrientedNodeTrail, and an expected number of
|
133
|
+
def validate_last_node_of_path_by_recoherence(path, recoherence_kmer, sequence_hash, min_concurring_reads=1)
|
134
|
+
#not possible to fail on a 1 or 2 node path, by debruijn graph definition.
|
135
|
+
#TODO: that ain't true! If one of the two nodes is sufficiently long, reads may not agree.
|
136
|
+
return true if path.length < 3
|
137
|
+
|
138
|
+
# Walk backwards along the path from the 2nd last node,
|
139
|
+
# collecting nodes until the length in bp of the nodes is > recoherence_kmer
|
140
|
+
collected_nodes = []
|
141
|
+
length_of_nodes = lambda do |nodes|
|
142
|
+
if nodes.empty?
|
143
|
+
0
|
144
|
+
else
|
145
|
+
hash_offset = nodes[0].node.parent_graph.hash_length-1
|
146
|
+
nodes.reduce(hash_offset) do |sum, node|
|
147
|
+
sum += node.node.length_alone
|
148
|
+
end
|
149
|
+
end
|
150
|
+
end
|
151
|
+
i = path.length-2
|
152
|
+
while i >= 0
|
153
|
+
collected_nodes.push path.trail[i]
|
154
|
+
i -= 1
|
155
|
+
# break if the recoherence_kmer doesn't cover
|
156
|
+
break if length_of_nodes.call(collected_nodes) + 1 >= recoherence_kmer
|
157
|
+
end
|
158
|
+
log.debug "validate: Collected nodes: #{collected_nodes}" if log.debug?
|
159
|
+
if collected_nodes.length < 2
|
160
|
+
log.debug "Only #{collected_nodes.length+1} nodes being tested for validation, so returning validated" if log.debug?
|
161
|
+
return true
|
162
|
+
end
|
163
|
+
|
164
|
+
# There should be at least 1 read that spans the collected nodes and the last node
|
165
|
+
# The trail validates if the above statement is true.
|
166
|
+
#TODO: there's a possible 'bug' here in that there's guarantee that the read is overlays the
|
167
|
+
# nodes in a consecutive and gapless manner. But I suspect that is unlikely to be a problem in practice.
|
168
|
+
final_node = path.trail[-1].node
|
169
|
+
possible_reads = final_node.short_reads.collect{|nr| nr.read_id}
|
170
|
+
log.debug "validate starting from #{final_node.node_id}: Initial short reads: #{possible_reads.join(',') }" if log.debug?
|
171
|
+
collected_nodes.each do |node|
|
172
|
+
log.debug "Validating node #{node}" if log.debug?
|
173
|
+
current_set = Set.new node.node.short_reads.collect{|nr| nr.read_id}
|
174
|
+
possible_reads.select! do |r|
|
175
|
+
current_set.include? r
|
176
|
+
end
|
177
|
+
if possible_reads.length < min_concurring_reads
|
178
|
+
log.debug "First line validation failed, now detecting sub-kmer sequence overlap" if log.debug?
|
179
|
+
trail_to_validate = path.trail[i+1..-1]
|
180
|
+
return sub_kmer_sequence_overlap?(trail_to_validate, sequence_hash, min_concurring_reads)
|
181
|
+
end
|
182
|
+
end
|
183
|
+
log.debug "Found #{possible_reads.length} reads that concurred with validation e.g. #{possible_reads[0]}" if log.debug?
|
184
|
+
return true
|
185
|
+
end
|
186
|
+
|
187
|
+
# Is there overlap across the given nodes, even if the overlap
|
188
|
+
# does not include an entire kmer?
|
189
|
+
# nodes: an OrientedNodeTrail. To validate, there must be at least 1 read that spans all of these nodes
|
190
|
+
# sequence_hash: Bio::Velvet::Sequence object with the sequences from the reads in the nodes
|
191
|
+
def sub_kmer_sequence_overlap?(nodes, sequence_hash, min_concurring_reads=1)
|
192
|
+
raise if nodes.length < 3 #should not get here - this is taken care of above
|
193
|
+
log.debug "validating by sub-kmer sequence overlap with min #{min_concurring_reads}: #{nodes}" if log.debug?
|
194
|
+
|
195
|
+
# Only reads that are in the second last node are possible, by de-bruijn graph definition.
|
196
|
+
candidate_noded_reads = nodes[-2].node.short_reads
|
197
|
+
middle_nodes_length = nodes[1..-2].reduce(0){|sum, n| sum += n.node.length}+
|
198
|
+
+nodes[0].node.parent_graph.hash_length-1
|
199
|
+
log.debug "Found middle nodes length #{middle_nodes_length}" if log.debug?
|
200
|
+
|
201
|
+
num_confirming_reads = 0
|
202
|
+
|
203
|
+
candidate_noded_reads.each do |read|
|
204
|
+
# Ignore reads that don't come in at the start of the node
|
205
|
+
log.debug "Considering read #{read.inspect}" if log.debug?
|
206
|
+
if read.offset_from_start_of_node != 0
|
207
|
+
log.debug "Read doesn't start at beginning of node, skipping" if log.debug?
|
208
|
+
next
|
209
|
+
else
|
210
|
+
seq = sequence_hash[read.read_id]
|
211
|
+
raise "No sequence stored for #{read.read_id}, programming fail." if seq.nil?
|
212
|
+
|
213
|
+
if read.start_coord == 0
|
214
|
+
log.debug "start_coord Insufficient length of read" if log.debug?
|
215
|
+
next
|
216
|
+
elsif seq.length-read.start_coord-middle_nodes_length < 1
|
217
|
+
log.debug "other_side Insufficient length of read" if log.debug?
|
218
|
+
next
|
219
|
+
end
|
220
|
+
|
221
|
+
# Now ensure that the sequence matches correctly
|
222
|
+
# left base, the base from the first node
|
223
|
+
first_node = nodes[0].node
|
224
|
+
left_base = !(read.direction ^ nodes[-2].starts_at_start?) ?
|
225
|
+
SINGLE_BASE_REVCOM[seq[read.start_coord-1]] :
|
226
|
+
seq[read.start_coord+middle_nodes_length]
|
227
|
+
left_comparison_base = nodes[0].starts_at_start? ?
|
228
|
+
first_node.ends_of_kmers_of_twin_node[0] :
|
229
|
+
first_node.ends_of_kmers_of_node[0]
|
230
|
+
if left_base != left_comparison_base
|
231
|
+
log.debug "left comparison base mismatch, this is not a validating read" if log.debug?
|
232
|
+
next
|
233
|
+
end
|
234
|
+
|
235
|
+
# right base, overlapping the last node
|
236
|
+
last_node = nodes[-1].node
|
237
|
+
right_base = !(read.direction ^ nodes[-2].starts_at_start?) ?
|
238
|
+
seq[read.start_coord+middle_nodes_length] :
|
239
|
+
SINGLE_BASE_REVCOM[seq[read.start_coord-1]]
|
240
|
+
right_comparison_base = nodes[-1].starts_at_start? ?
|
241
|
+
last_node.ends_of_kmers_of_node[0] :
|
242
|
+
last_node.ends_of_kmers_of_twin_node[0]
|
243
|
+
if right_base != right_comparison_base
|
244
|
+
log.debug "right comparison base mismatch, this is not a validating read" if log.debug?
|
245
|
+
next
|
246
|
+
end
|
247
|
+
|
248
|
+
log.debug "Read validates path"
|
249
|
+
num_confirming_reads += 1
|
250
|
+
if num_confirming_reads >= min_concurring_reads
|
251
|
+
return true #gauntlet passed, this is enough confirmatory reads, and so the path is validated.
|
252
|
+
end
|
253
|
+
end
|
254
|
+
end
|
255
|
+
return false #no candidate reads pass
|
256
|
+
end
|
257
|
+
|
258
|
+
|
259
|
+
# Separate stacks for valid paths and paths which exceed the maximum allowed
|
260
|
+
# cycle count.
|
261
|
+
# Each backtrack spawns a set of new paths, which are cycle counted. If any cycle
|
262
|
+
# is repeated more than max_cycles, the new path is pushed to the max_cycle_stack,
|
263
|
+
# otherwise the path is pushed to the main stack. Main stack paths are prioritised.
|
264
|
+
# The max_cycle_stack paths must be tracked until cycle repeats in second_part exceed
|
265
|
+
# max_cycles, as they can spawn valid paths with backtracking.
|
266
|
+
def find_paths_from_problems(problems, recoherence_kmer, options={})
|
267
|
+
max_num_paths = options[:max_gapfill_paths]
|
268
|
+
max_num_paths ||= 2196
|
269
|
+
max_cycles = options[:max_cycles] || 1
|
270
|
+
|
271
|
+
counter = CycleCounter.new(max_cycles)
|
272
|
+
decide_stack = lambda do |to_push|
|
273
|
+
if max_cycles < counter.path_cycle_count(to_push.flatten)
|
274
|
+
log.debug "Pushing #{to_push.collect{|part| part.collect{|onode| onode.node.node_id}.join(',')}.join(' and ') } to secondary stack" if log.debug?
|
275
|
+
return true
|
276
|
+
else
|
277
|
+
log.debug "Pushing #{to_push.collect{|part| part.collect{|onode| onode.node.node_id}.join(',')}.join(' and ') } to main stack" if log.debug?
|
278
|
+
return false
|
279
|
+
end
|
280
|
+
end
|
281
|
+
|
282
|
+
stack = DualStack.new &decide_stack
|
283
|
+
to_return = Bio::AssemblyGraphAlgorithms::TrailSet.new
|
284
|
+
|
285
|
+
# if there is no solutions to the overall problem then there is no solution at all
|
286
|
+
if problems.terminal_node_keys.nil? or problems.terminal_node_keys.empty?
|
287
|
+
to_return.trails = []
|
288
|
+
return to_return
|
289
|
+
end
|
290
|
+
|
291
|
+
# push all solutions to the "ending in the final node" solutions to the stack
|
292
|
+
problems.terminal_node_keys.each do |key|
|
293
|
+
overall_solution = problems[key]
|
294
|
+
first_part = overall_solution.known_paths[0].to_a
|
295
|
+
stack.push [first_part, []]
|
296
|
+
end
|
297
|
+
|
298
|
+
all_paths_hash = {}
|
299
|
+
while path_parts = stack.pop
|
300
|
+
log.debug path_parts.collect{|half| half.collect{|onode| onode.node.node_id}.join(',')}.join(' and ') if log.debug?
|
301
|
+
first_part = path_parts[0]
|
302
|
+
second_part = path_parts[1]
|
303
|
+
|
304
|
+
if first_part.length == 0
|
305
|
+
# If we've tracked all the way to the beginning,
|
306
|
+
# then there's no need to track further
|
307
|
+
|
308
|
+
# add this solution if required
|
309
|
+
# I've had some trouble getting the Ruby Set to work here, but this is effectively the same thing.
|
310
|
+
log.debug "Found solution: #{second_part.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
311
|
+
key = second_part.hash
|
312
|
+
all_paths_hash[key] ||= second_part
|
313
|
+
else
|
314
|
+
last = first_part.last
|
315
|
+
|
316
|
+
if second_part.include? last
|
317
|
+
log.debug "Cycle at node #{last.node_id} detected in previous path #{second_part.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
318
|
+
to_return.circular_paths_detected = true
|
319
|
+
if max_cycles == 0 or max_cycles < counter.path_cycle_count([last, second_part].flatten)
|
320
|
+
log.debug "Not finishing cyclic path with too many repeated cycles." if log.debug?
|
321
|
+
next
|
322
|
+
end
|
323
|
+
end
|
324
|
+
|
325
|
+
paths_to_last = problems[array_trail_to_settable(first_part, recoherence_kmer)].known_paths
|
326
|
+
paths_to_last.each do |path|
|
327
|
+
stack.push [path[0...(path.length-1)], [last,second_part].flatten]
|
328
|
+
end
|
329
|
+
end
|
330
|
+
|
331
|
+
# max_num_paths parachute
|
332
|
+
# The parachute can kill the search once the main stack exceeds max_gapfill_paths,
|
333
|
+
# since all paths on it are valid.
|
334
|
+
if !max_num_paths.nil? and (stack.sizes[0] + all_paths_hash.length) > max_num_paths
|
335
|
+
log.info "Exceeded the maximum number of allowable paths in this gapfill" if log.info?
|
336
|
+
to_return.max_path_limit_exceeded = true
|
337
|
+
all_paths_hash = {}
|
338
|
+
break
|
339
|
+
end
|
340
|
+
end
|
341
|
+
|
342
|
+
to_return.trails = all_paths_hash.values
|
343
|
+
return to_return
|
344
|
+
end
|
345
|
+
|
346
|
+
class DualStack
|
347
|
+
def initialize(&block)
|
348
|
+
@checker = block
|
349
|
+
@stack = DS::Stack.new
|
350
|
+
@dual_stack = DS::Stack.new
|
351
|
+
end
|
352
|
+
|
353
|
+
def push to_push
|
354
|
+
if @checker.call to_push
|
355
|
+
@dual_stack.push to_push
|
356
|
+
else
|
357
|
+
@stack.push to_push
|
358
|
+
end
|
359
|
+
end
|
360
|
+
|
361
|
+
def pop
|
362
|
+
@stack.pop || @dual_stack.pop
|
363
|
+
end
|
364
|
+
|
365
|
+
def sizes
|
366
|
+
return @stack.size, @dual_stack.size
|
367
|
+
end
|
368
|
+
end
|
369
|
+
|
370
|
+
# Count occurrences of cycles in paths through an assembly graph. Works by building a hash of paths and
|
371
|
+
# the frequency of the modal cycle in that path (up to the cut-off max_cycles). For an unknown path, looks
|
372
|
+
# for a subset of the path in hash by removing nodes from start (or end if :forward option is set), and
|
373
|
+
# then extends the subset by iteratively re-adding a single node and adding to the hash of paths the larger
|
374
|
+
# of the subset count or the frequency for the modal cycle beginning with the re-added node.
|
375
|
+
class CycleCounter
|
376
|
+
include Bio::FinishM::Logging
|
377
|
+
|
378
|
+
def initialize(max_cycles, options = {})
|
379
|
+
@max_cycles = max_cycles
|
380
|
+
@path_cache = Hash.new # Cache max_cycles for previously seen paths
|
381
|
+
@forward = options[:forward] || false # By default builds hash assuming backtracking from end of path. This flag will reverse path direction and build hash assuming moving forwards.
|
382
|
+
end
|
383
|
+
|
384
|
+
|
385
|
+
# Iterate through unique nodes of path and find maximal cycle counts
|
386
|
+
def path_cycle_count(path)
|
387
|
+
log.debug "Finding cycles in path #{path.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
388
|
+
first_part = []
|
389
|
+
second_part = path
|
390
|
+
keys = []
|
391
|
+
count = nil
|
392
|
+
reached_max_cycles = false
|
393
|
+
|
394
|
+
second_part = second_part.reverse if @forward
|
395
|
+
|
396
|
+
# Iterate along path and look for the remaining path in cache. Remember the iterated
|
397
|
+
# path and the remaining path. Stop if a cache count is found, else use zero.
|
398
|
+
while count.nil? and !second_part.empty?
|
399
|
+
key = second_part.collect{|onode| onode.to_settable}.flatten
|
400
|
+
|
401
|
+
# Check if path value is cached
|
402
|
+
if @path_cache.has_key? key
|
403
|
+
count = @path_cache[key]
|
404
|
+
#log.debug "Found cached count #{count} for path #{second_part.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
405
|
+
break
|
406
|
+
else
|
407
|
+
first_part = [first_part, second_part.first].flatten
|
408
|
+
second_part = second_part[1..-1]
|
409
|
+
end
|
410
|
+
end
|
411
|
+
|
412
|
+
if second_part.empty?
|
413
|
+
#log.debug "Reached end of path without finding cached count." if log.debug?
|
414
|
+
count = 0
|
415
|
+
end
|
416
|
+
|
417
|
+
# The max cycle count for a path is the largest of:
|
418
|
+
# I. Cycle count for initial node of path in remaining path (without initial
|
419
|
+
# node), or
|
420
|
+
# II. Max cycle count of remaining path.
|
421
|
+
|
422
|
+
# We then iterate back through the iterated path. If count does not exceed
|
423
|
+
# max_cycles, we count cycles for each node in the remaining path, then
|
424
|
+
# backtrack by moving the node to the remaining path set. We record the count
|
425
|
+
# for each remaining path
|
426
|
+
while !first_part.empty?
|
427
|
+
|
428
|
+
node = first_part.last
|
429
|
+
if !reached_max_cycles
|
430
|
+
#log.debug "Next node is #{node.node.node_id}." if log.debug?
|
431
|
+
node_count = path_cycle_count_for_node(node, second_part, @max_cycles)
|
432
|
+
count = [count, node_count].max
|
433
|
+
reached_max_cycles = count > @max_cycles
|
434
|
+
end
|
435
|
+
|
436
|
+
second_part = [node, second_part].flatten
|
437
|
+
first_part = first_part[0...-1]
|
438
|
+
|
439
|
+
key = second_part.collect{|onode| onode.to_settable}.flatten
|
440
|
+
@path_cache[key] = count
|
441
|
+
#log.debug "Caching cycle count #{count} for path #{second_part.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
442
|
+
end
|
443
|
+
if reached_max_cycles and log.debug?
|
444
|
+
log.debug "Most repeated cycle in path occured #{count} or more times."
|
445
|
+
elsif log.debug?
|
446
|
+
log.debug "Most repeated cycle in path occured #{count} times."
|
447
|
+
end
|
448
|
+
return count
|
449
|
+
end
|
450
|
+
|
451
|
+
# For an initial node, find and count unique 'simple' cycles in a path that begin at the initial
|
452
|
+
# node, up to a max_cycles. Return count for the maximally repeated cycle if less than max_cycles,
|
453
|
+
# or max_cycles.
|
454
|
+
def path_cycle_count_for_node(node, path, max_cycles=1)
|
455
|
+
#log.debug "Finding all simple cycles for node #{node.node_id} in path #{path.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
456
|
+
remaining = path
|
457
|
+
cycles = Hash.new
|
458
|
+
|
459
|
+
remaining = remaining.reverse if @forward
|
460
|
+
|
461
|
+
while remaining.include?(node)
|
462
|
+
position = remaining.index(node)
|
463
|
+
cycle = remaining[0..position]
|
464
|
+
remaining = remaining[(position+1)..-1]
|
465
|
+
#log.debug "Found cycle: #{cycle.collect{|onode| onode.node.node_id}.join(',')}." if log.debug?
|
466
|
+
|
467
|
+
set_key = cycle.collect{|onode| onode.to_settable}.flatten
|
468
|
+
cycles[set_key] ||= 0
|
469
|
+
cycles[set_key] += 1
|
470
|
+
#log.debug "Found repeat #{cycles[set_key]}." if log.debug?
|
471
|
+
|
472
|
+
if cycles[set_key] > max_cycles
|
473
|
+
#log.debug "Max cycles #{max_cycles} exceeded." if log.debug?
|
474
|
+
return cycles[set_key]
|
475
|
+
end
|
476
|
+
end
|
477
|
+
if cycles.empty?
|
478
|
+
max_counts = 0
|
479
|
+
else
|
480
|
+
max_counts = cycles.values.max
|
481
|
+
end
|
482
|
+
#log.debug "Most cycles found #{max_counts}." if log.debug?
|
483
|
+
return max_counts
|
484
|
+
end
|
485
|
+
end
|
486
|
+
|
487
|
+
class DynamicProgrammingProblem
|
488
|
+
attr_accessor :min_distance, :known_paths
|
489
|
+
|
490
|
+
def initialize
|
491
|
+
@known_paths = []
|
492
|
+
end
|
493
|
+
end
|
494
|
+
|
495
|
+
# Like a Hash, but also contains a list of keys that end in the
|
496
|
+
# terminal node
|
497
|
+
class ProblemSet < Hash
|
498
|
+
# Array of keys to this hash that end in the terminal onode
|
499
|
+
attr_accessor :terminal_node_keys
|
500
|
+
end
|
501
|
+
|
502
|
+
class ProblemTrailFinder
|
503
|
+
include Bio::FinishM::Logging
|
504
|
+
|
505
|
+
def initialize(graph, initial_path)
|
506
|
+
@graph = graph
|
507
|
+
@pqueue = DS::AnyPriorityQueue.new {|a,b| a < b}
|
508
|
+
@pqueue.enqueue initial_path.copy, 0
|
509
|
+
end
|
510
|
+
|
511
|
+
def dequeue
|
512
|
+
@pqueue.dequeue
|
513
|
+
end
|
514
|
+
|
515
|
+
def length
|
516
|
+
@pqueue.length
|
517
|
+
end
|
518
|
+
|
519
|
+
def push_next_neighbours(current_path)
|
520
|
+
next_nodes = current_path.neighbours_of_last_node(@graph)
|
521
|
+
log.debug "Pushing #{next_nodes.length} new neighbours of #{current_path.last}" if log.debug?
|
522
|
+
#TODO: not neccessary to copy all paths, can just continue one of them
|
523
|
+
next_nodes.each do |n|
|
524
|
+
log.debug "Pushing neighbour to stack: #{n}" if log.debug?
|
525
|
+
path = current_path.copy
|
526
|
+
path.add_oriented_node n
|
527
|
+
@pqueue.enqueue path, path.length_in_bp
|
528
|
+
end
|
529
|
+
end
|
530
|
+
end
|
531
|
+
end
|
532
|
+
|
533
|
+
|