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 +1,153 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
require 'optparse'
|
4
|
+
require 'bio-logger'
|
5
|
+
require 'systemu'
|
6
|
+
|
7
|
+
SCRIPT_NAME = File.basename(__FILE__); LOG_NAME = SCRIPT_NAME.gsub('.rb','')
|
8
|
+
|
9
|
+
# Parse command line options into the options hash
|
10
|
+
options = {
|
11
|
+
:logger => 'stderr',
|
12
|
+
:log_level => 'info',
|
13
|
+
}
|
14
|
+
o = OptionParser.new do |opts|
|
15
|
+
opts.banner = "
|
16
|
+
Usage: #{SCRIPT_NAME} -b <contigs_against_assembly.blast_outfmt6.csv>
|
17
|
+
|
18
|
+
Takes a set of contigs, and an assembly. Works out if there are any contigs where there is a blast hit spanning of the contigs using two of the assembly's contig ends.\n\n"
|
19
|
+
|
20
|
+
opts.on("--query FASTA_FILE", "new contigs fasta file [Required]") do |arg|
|
21
|
+
options[:query_file] = arg
|
22
|
+
end
|
23
|
+
opts.on("--blastdb FASTA_FILE_FORMATTED", "basename of makeblastdb output [Required]") do |arg|
|
24
|
+
options[:blastdb] = arg
|
25
|
+
end
|
26
|
+
|
27
|
+
# logger options
|
28
|
+
opts.separator "\nVerbosity:\n\n"
|
29
|
+
opts.on("-q", "--quiet", "Run quietly, set logging to ERROR level [default INFO]") {options[:log_level] = 'error'}
|
30
|
+
opts.on("--logger filename",String,"Log to file [default #{options[:logger]}]") { |name| options[:logger] = name}
|
31
|
+
opts.on("--trace options",String,"Set log level [default INFO]. e.g. '--trace debug' to set logging level to DEBUG"){|s| options[:log_level] = s}
|
32
|
+
end; o.parse!
|
33
|
+
if ARGV.length != 0 or options[:query_file].nil? or options[:blastdb].nil?
|
34
|
+
$stderr.puts o
|
35
|
+
exit 1
|
36
|
+
end
|
37
|
+
# Setup logging
|
38
|
+
Bio::Log::CLI.logger(options[:logger]); Bio::Log::CLI.trace(options[:log_level]); log = Bio::Log::LoggerPlus.new(LOG_NAME); Bio::Log::CLI.configure(LOG_NAME)
|
39
|
+
|
40
|
+
|
41
|
+
# Read in the blast file
|
42
|
+
blast_results = []
|
43
|
+
class BlastResult
|
44
|
+
attr_accessor :qseqid, :sseqid, :pident, :length, :mismatch, :gapopen, :qstart, :qend, :sstart, :subject_end, :evalue, :bitscore, :query_length, :subject_length
|
45
|
+
|
46
|
+
attr_accessor :cutoff_inwards
|
47
|
+
|
48
|
+
def initialize
|
49
|
+
@cutoff_inwards = 500
|
50
|
+
end
|
51
|
+
|
52
|
+
def hits_end_of_subject?
|
53
|
+
@subject_end >= @subject_length-@cutoff_inwards and @length >= 100
|
54
|
+
end
|
55
|
+
|
56
|
+
def hits_start_of_subject?
|
57
|
+
@sstart <= @cutoff_inwards and @length >= 100
|
58
|
+
end
|
59
|
+
|
60
|
+
def hits_end_of_query?
|
61
|
+
@qend >= @query_length-@cutoff_inwards and @length >= 100
|
62
|
+
end
|
63
|
+
|
64
|
+
def hits_start_of_query?
|
65
|
+
@qstart <= @cutoff_inwards and @length >= 100
|
66
|
+
end
|
67
|
+
end
|
68
|
+
|
69
|
+
status, blast_output, stderr = systemu "blastn -query #{options[:query_file].inspect} -db #{options[:blastdb].inspect} -outfmt '6 qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore qlen slen' -evalue 1e-5"
|
70
|
+
raise stderr unless stderr==""
|
71
|
+
raise "bad status running blast" unless status.exitstatus == 0
|
72
|
+
log.debug "Finished running blast, presumably successfully"
|
73
|
+
|
74
|
+
blast_output.each_line do |line|
|
75
|
+
res = BlastResult.new
|
76
|
+
row = line.chomp.split "\t"
|
77
|
+
[:qseqid, :sseqid, :pident, :length, :mismatch, :gapopen, :qstart,
|
78
|
+
:qend, :sstart, :subject_end, :evalue, :bitscore,
|
79
|
+
:query_length, :subject_length].each_with_index do |attr, i|
|
80
|
+
res.send "#{attr}=".to_sym, row[i]
|
81
|
+
end
|
82
|
+
[:length, :mismatch, :gapopen, :qstart,
|
83
|
+
:qend, :sstart, :subject_end,:query_length, :subject_length].each do |attr|
|
84
|
+
res.send "#{attr}=".to_sym, res.send(attr).to_i
|
85
|
+
end
|
86
|
+
[:pident, :evalue, :bitscore].each do |attr|
|
87
|
+
res.send "#{attr}=".to_sym, res.send(attr).to_f
|
88
|
+
end
|
89
|
+
|
90
|
+
blast_results.push res
|
91
|
+
end
|
92
|
+
log.info "Parsed #{blast_results.length} blast results e.g. #{blast_results[0].inspect}"
|
93
|
+
|
94
|
+
|
95
|
+
query_to_blast_results = {}
|
96
|
+
hit_to_blast_results = {}
|
97
|
+
blast_results.each do |result|
|
98
|
+
query_to_blast_results[result.qseqid] ||= []
|
99
|
+
query_to_blast_results[result.qseqid].push result
|
100
|
+
|
101
|
+
hit_to_blast_results[result.sseqid] ||= []
|
102
|
+
hit_to_blast_results[result.sseqid].push result
|
103
|
+
end
|
104
|
+
|
105
|
+
# For each query sequence, does it map to the ends of both contigs
|
106
|
+
header = %w(query subject1 subject2 qstart1? qend1? sstart1? send1? qstart2? qend2? sstart2? send2?).join("\t")
|
107
|
+
query_to_blast_results.each do |query_id, hits|
|
108
|
+
query_length = hits[0].query_length
|
109
|
+
keepers = []
|
110
|
+
|
111
|
+
hits.each do |hit|
|
112
|
+
# perfect if it hits the start or the end (but not both) of both the query and the subject, unless it is circular
|
113
|
+
if hit.hits_start_of_query? ^ hit.hits_end_of_query? and
|
114
|
+
hit.hits_start_of_subject? ^ hit.hits_end_of_subject?
|
115
|
+
keepers.push hit
|
116
|
+
elsif hit.hits_start_of_query? or hit.hits_end_of_query? or
|
117
|
+
hit.hits_start_of_subject? or hit.hits_end_of_subject?
|
118
|
+
log.info "There's a half-correct hit for #{query_id}: qstart? #{hit.hits_start_of_query?} qend #{hit.hits_end_of_query?} "+
|
119
|
+
"sstart #{hit.hits_start_of_subject?} send #{hit.hits_end_of_subject?}, to subject sequence #{hit.sseqid}"
|
120
|
+
end
|
121
|
+
end
|
122
|
+
|
123
|
+
if keepers.empty?
|
124
|
+
log.debug "no latchings found for #{query_id}"
|
125
|
+
elsif keepers.length == 1
|
126
|
+
log.info "Query #{query_id} only latches on to a single end, maybe manually inspect"
|
127
|
+
elsif keepers.length == 2
|
128
|
+
log.debug "Query #{query_id} has 2 keepers!"
|
129
|
+
q = keepers.collect{|hit| hit.hits_start_of_query?}.join
|
130
|
+
s = keepers.collect{|hit| hit.hits_start_of_subject?}.join
|
131
|
+
if (q == 'truefalse' or q == 'falsetrue') and
|
132
|
+
(s == 'truefalse' or s == 'falsetrue')
|
133
|
+
outs = (0..1).collect{|i|
|
134
|
+
[
|
135
|
+
keepers[i].hits_start_of_query?,
|
136
|
+
keepers[i].hits_end_of_query?,
|
137
|
+
keepers[i].hits_start_of_subject?,
|
138
|
+
keepers[i].hits_end_of_subject?,
|
139
|
+
]
|
140
|
+
}.flatten
|
141
|
+
unless header.nil?
|
142
|
+
puts header
|
143
|
+
header = nil
|
144
|
+
end
|
145
|
+
puts [query_id, keepers[0].sseqid, keepers[1].sseqid, outs].flatten.join("\t")
|
146
|
+
else
|
147
|
+
log.info "Query #{query_id} has 2 keepers, but they are fighting it seems"
|
148
|
+
end
|
149
|
+
else
|
150
|
+
log.info "More than 2 keepers found for #{query_id}, manual inspection likely required"
|
151
|
+
end
|
152
|
+
end
|
153
|
+
|
data/bin/finishm
ADDED
@@ -0,0 +1,143 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
require 'optparse'
|
4
|
+
require 'bio-logger'
|
5
|
+
require 'bio-velvet'
|
6
|
+
require 'pp'
|
7
|
+
|
8
|
+
SCRIPT_NAME = File.basename(__FILE__); LOG_NAME = 'finishm'
|
9
|
+
$:.unshift File.join(File.dirname(__FILE__),'..','lib')
|
10
|
+
require 'priner'
|
11
|
+
|
12
|
+
# Parse command line options into the options hash
|
13
|
+
options = {
|
14
|
+
:logger => 'stderr',
|
15
|
+
:log_level => 'info',
|
16
|
+
}
|
17
|
+
global = OptionParser.new do |opts|
|
18
|
+
opts.banner = "
|
19
|
+
Usage: #{SCRIPT_NAME} <command> [<arguments>]
|
20
|
+
|
21
|
+
FinishM is a collection of tasks related to assembly and metagenome assembly. Available commands:
|
22
|
+
|
23
|
+
wander\tTry to connect contigs (experimental)
|
24
|
+
gapfill\tFill assembly gaps (N characters) (experimental)
|
25
|
+
explore\tWhat happens in the graph beyond the end of my contig(s)? (experimental)
|
26
|
+
visualise\tVisualise the DeBruijn graph (experimental)
|
27
|
+
|
28
|
+
Commands for PCR finishing:
|
29
|
+
|
30
|
+
primers\tdesign primers for multi-primer multi-lane PCR experimental setup (experimental)
|
31
|
+
primers_check\ttest a set of primers for incompatibility (experimental)
|
32
|
+
finish\tprocess results from multi-primer multi-lane PCR experimental setup (experimental)
|
33
|
+
|
34
|
+
Utility modes:
|
35
|
+
|
36
|
+
sequence\tGiven a defined sequence of nodes, what is the corresponding sequence?
|
37
|
+
count_paths\tCount the number of paths through assembly graph
|
38
|
+
find_orfs\tFind possible open reading frames in assembly graph
|
39
|
+
\n\n"
|
40
|
+
end
|
41
|
+
global.order!
|
42
|
+
|
43
|
+
operator = nil
|
44
|
+
subcommands = {
|
45
|
+
'primers' => lambda {OptionParser.new do |opts|
|
46
|
+
operator = Bio::FinishM::Primers.new
|
47
|
+
operator.add_options(opts, options)
|
48
|
+
end},
|
49
|
+
'primers_check' => lambda {OptionParser.new do |opts|
|
50
|
+
operator = Bio::FinishM::Primers::Checker.new
|
51
|
+
operator.add_options(opts, options)
|
52
|
+
end},
|
53
|
+
'finish' => lambda {OptionParser.new do |opts|
|
54
|
+
operator = Bio::FinishM::Finisher.new
|
55
|
+
operator.add_options(opts, options)
|
56
|
+
end},
|
57
|
+
'gapfill' => lambda {OptionParser.new do |opts|
|
58
|
+
operator = Bio::FinishM::GapFiller.new
|
59
|
+
operator.add_options(opts, options)
|
60
|
+
end},
|
61
|
+
'wander' => lambda {OptionParser.new do |opts|
|
62
|
+
operator = Bio::FinishM::Wanderer.new
|
63
|
+
operator.add_options(opts, options)
|
64
|
+
end},
|
65
|
+
'fluff' => lambda {OptionParser.new do |opts|
|
66
|
+
operator = Bio::FinishM::Fluff.new
|
67
|
+
operator.add_options(opts, options)
|
68
|
+
end},
|
69
|
+
'explore' => lambda {OptionParser.new do |opts|
|
70
|
+
operator = Bio::FinishM::Explorer.new
|
71
|
+
operator.add_options(opts, options)
|
72
|
+
end},
|
73
|
+
'assemble' => lambda {OptionParser.new do |opts|
|
74
|
+
operator = Bio::FinishM::Assembler.new
|
75
|
+
operator.add_options(opts, options)
|
76
|
+
end},
|
77
|
+
'visualise' => lambda {OptionParser.new do |opts|
|
78
|
+
operator = Bio::FinishM::Visualise.new
|
79
|
+
operator.add_options(opts, options)
|
80
|
+
end},
|
81
|
+
'sequence' => lambda {OptionParser.new do |opts|
|
82
|
+
operator = Bio::FinishM::Sequence.new
|
83
|
+
operator.add_options(opts, options)
|
84
|
+
end},
|
85
|
+
'roundup' => lambda {OptionParser.new do |opts|
|
86
|
+
operator = Bio::FinishM::RoundUp.new
|
87
|
+
operator.add_options(opts, options)
|
88
|
+
end},
|
89
|
+
'count_paths' => lambda {OptionParser.new do |opts|
|
90
|
+
operator = Bio::FinishM::PathCounter.new
|
91
|
+
operator.add_options(opts, options)
|
92
|
+
end},
|
93
|
+
'find_orfs' => lambda {OptionParser.new do |opts|
|
94
|
+
operator = Bio::FinishM::ORFsFinder.new
|
95
|
+
operator.add_options(opts, options)
|
96
|
+
end}
|
97
|
+
}
|
98
|
+
|
99
|
+
subcommand = nil
|
100
|
+
if ARGV[0] and ARGV[0].match(/finishm$/) #if debugging e.g. 'pry finishm wander ..'. But be careful of finishm no arguments
|
101
|
+
subcommand = ARGV.shift
|
102
|
+
subcommand = ARGV.shift
|
103
|
+
else
|
104
|
+
# not debugging
|
105
|
+
subcommand = ARGV.shift
|
106
|
+
end
|
107
|
+
|
108
|
+
if subcommand.nil?
|
109
|
+
$stderr.puts global
|
110
|
+
exit 1
|
111
|
+
elsif !subcommands[subcommand]
|
112
|
+
$stderr.puts "Unrecognized subcommand: #{subcommand}"
|
113
|
+
exit 1
|
114
|
+
else
|
115
|
+
# Add options specific for subcommand
|
116
|
+
opts = subcommands[subcommand].call
|
117
|
+
|
118
|
+
# Add global logging options
|
119
|
+
opts.separator "\nVerbosity:\n\n"
|
120
|
+
opts.on("-q", "--quiet", "Run quietly, set logging to ERROR level [default INFO]") {options[:log_level] = 'error'}
|
121
|
+
opts.on("--logger filename",String,"Log to file [default #{options[:logger]}]") { |name| options[:logger] = name}
|
122
|
+
opts.on("--trace options",String,"Set log level [default INFO]. e.g. '--trace debug' to set logging level to DEBUG"){|s| options[:log_level] = s}
|
123
|
+
opts.separator "\n"
|
124
|
+
opts.parse!
|
125
|
+
|
126
|
+
# Setup logging
|
127
|
+
Bio::Log::CLI.logger(options[:logger]); Bio::Log::CLI.trace(options[:log_level]); log = Bio::Log::LoggerPlus.new(LOG_NAME); Bio::Log::CLI.configure(LOG_NAME); log.outputters.each {|o| o.formatter = Log4r::PatternFormatter.new(:pattern => "%5l %c %d: %m", :date_pattern => '%d/%m %T')}
|
128
|
+
Bio::Log::LoggerPlus.new 'bio-velvet'; Bio::Log::CLI.configure 'bio-velvet'
|
129
|
+
|
130
|
+
log.debug "Running FinishM #{subcommand} with these options: #{PP.pp(options, "").gsub(/\n$/,'')}"
|
131
|
+
|
132
|
+
error_message = operator.validate_options(options, ARGV)
|
133
|
+
if error_message.nil? or error_message == false
|
134
|
+
operator.run options, ARGV
|
135
|
+
else
|
136
|
+
$stderr.puts
|
137
|
+
$stderr.puts "ERROR parsing options: #{error_message}"
|
138
|
+
$stderr.puts
|
139
|
+
$stderr.puts opts
|
140
|
+
exit 1
|
141
|
+
end
|
142
|
+
end
|
143
|
+
|
@@ -0,0 +1,55 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
require 'optparse'
|
4
|
+
require 'bio-logger'
|
5
|
+
|
6
|
+
SCRIPT_NAME = File.basename(__FILE__); LOG_NAME = SCRIPT_NAME.gsub('.rb','')
|
7
|
+
|
8
|
+
# Parse command line options into the options hash
|
9
|
+
options = {
|
10
|
+
:logger => 'stderr',
|
11
|
+
:log_level => 'info',
|
12
|
+
}
|
13
|
+
o = OptionParser.new do |opts|
|
14
|
+
opts.banner = "
|
15
|
+
Usage: #{SCRIPT_NAME} <arguments>
|
16
|
+
|
17
|
+
Description of what this program does...\n\n"
|
18
|
+
|
19
|
+
opts.on("--velvet-pregraph GRAPH_FILE", "PreGraph file output from velveth [required]") do |arg|
|
20
|
+
options[:velvet_pregraph_file] = arg
|
21
|
+
end
|
22
|
+
|
23
|
+
# logger options
|
24
|
+
opts.separator "\nVerbosity:\n\n"
|
25
|
+
opts.on("-q", "--quiet", "Run quietly, set logging to ERROR level [default INFO]") {options[:log_level] = 'error'}
|
26
|
+
opts.on("--logger filename",String,"Log to file [default #{options[:logger]}]") { |name| options[:logger] = name}
|
27
|
+
opts.on("--trace options",String,"Set log level [default INFO]. e.g. '--trace debug' to set logging level to DEBUG"){|s| options[:log_level] = s}
|
28
|
+
end; o.parse!
|
29
|
+
if ARGV.length != 0 or options[:velvet_pregraph_file].nil?
|
30
|
+
$stderr.puts o
|
31
|
+
exit 1
|
32
|
+
end
|
33
|
+
# Setup logging
|
34
|
+
Bio::Log::CLI.logger(options[:logger]); Bio::Log::CLI.trace(options[:log_level]); log = Bio::Log::LoggerPlus.new(LOG_NAME); Bio::Log::CLI.configure(LOG_NAME)
|
35
|
+
|
36
|
+
|
37
|
+
# Read in the velvet graph
|
38
|
+
log.info "Parsing graph from #{options[:velvet_pregraph_file]}"
|
39
|
+
graph = Bio::Velvet::Graph.parse_from_file(options[:velvet_pregraph_file])
|
40
|
+
log.info "Finished parsing graph, found #{graph.number_of_nodes} nodes"
|
41
|
+
|
42
|
+
# Log the number of nodes and arcs in the current graph
|
43
|
+
|
44
|
+
# Read in the fasta file of immutable nodes, and extract the two most immutable
|
45
|
+
# Log that they are found
|
46
|
+
|
47
|
+
# Determine that the graph is connected or not between the two most immutable nodes, using some graph theoretic algorithm
|
48
|
+
# If the graph is not connected, then there is no hope, exit
|
49
|
+
|
50
|
+
# Go through the graph to get a list of the cap nodes
|
51
|
+
# Log the number of cap nodes found
|
52
|
+
|
53
|
+
# Trim off all the cap nodes back to cross nodes, keeping track of the lengths
|
54
|
+
|
55
|
+
# Print the graph in graphviz dot format
|
@@ -0,0 +1,241 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
require 'optparse'
|
4
|
+
require 'bio-logger'
|
5
|
+
require 'bio-velvet'
|
6
|
+
require 'tempfile'
|
7
|
+
require 'pp'
|
8
|
+
|
9
|
+
SCRIPT_NAME = File.basename(__FILE__); LOG_NAME = 'finishm'
|
10
|
+
$:.unshift File.join(File.dirname(__FILE__),'..','lib')
|
11
|
+
require 'priner'
|
12
|
+
|
13
|
+
# Parse command line options into the options hash
|
14
|
+
options = {
|
15
|
+
:logger => 'stderr',
|
16
|
+
:log_level => 'info',
|
17
|
+
:velvet_kmer_size => 43,#TODO: these options should be exposed to the user, and perhaps not guessed at
|
18
|
+
:contig_end_length => 200,
|
19
|
+
:output_assembly_path => '/tmp/velvetAssembly',
|
20
|
+
:graph_search_leash_length => 3000,
|
21
|
+
:assembly_coverage_cutoff => 1.5,
|
22
|
+
}
|
23
|
+
o = OptionParser.new do |opts|
|
24
|
+
opts.banner = "
|
25
|
+
Usage: #{SCRIPT_NAME} --reads <read_file> --contig <contig_file>
|
26
|
+
|
27
|
+
Takes a set of reads and a contig that contains gap characters. Then it tries to fill in
|
28
|
+
these N characters. It is possible that there is multiple ways to close the gap - in that case
|
29
|
+
each is reported. \n\n"
|
30
|
+
|
31
|
+
|
32
|
+
opts.on("--reads FILE", "gzipped fastq file of reads to perform the gap closing with [required]") do |arg|
|
33
|
+
options[:reads_file] = arg
|
34
|
+
end
|
35
|
+
opts.on("--contig FILE", "fasta file of single contig containing Ns that are to be closed [required]") do |arg|
|
36
|
+
options[:contig_file] = arg
|
37
|
+
end
|
38
|
+
opts.on("--output-trails-fasta PATH", "Output found paths to this file in fasta format [default: off]") do |arg|
|
39
|
+
options[:overall_trail_output_fasta_file] = arg
|
40
|
+
end
|
41
|
+
|
42
|
+
opts.separator "\nOptional arguments:\n\n"
|
43
|
+
opts.on("--overhang NUM", "Start assembling this far from the gap [default: #{options[:contig_end_length]}]") do |arg|
|
44
|
+
options[:contig_end_length] = arg.to_i
|
45
|
+
end
|
46
|
+
opts.on("--start OFFSET", "Start trying to fill from this position in the contig, requires --stop [default: found from position of Ns}]") do |arg|
|
47
|
+
options[:start_offset] = arg.to_i-1
|
48
|
+
end
|
49
|
+
opts.on("--stop OFFSET", "Start trying to fill to this position in the contig, requires --start [default: found from position of Ns}]") do |arg|
|
50
|
+
options[:end_offset] = arg.to_i-1
|
51
|
+
end
|
52
|
+
opts.on("--assembly-png PATH", "Output assembly as a PNG file [default: off]") do |arg|
|
53
|
+
options[:output_graph_png] = arg
|
54
|
+
end
|
55
|
+
opts.on("--assembly-svg PATH", "Output assembly as an SVG file [default: off]") do |arg|
|
56
|
+
options[:output_graph_svg] = arg
|
57
|
+
end
|
58
|
+
opts.on("--assembly-dot PATH", "Output assembly as an DOT file [default: off]") do |arg|
|
59
|
+
options[:output_graph_dot] = arg
|
60
|
+
end
|
61
|
+
opts.on("--velvet-kmer KMER", "kmer size to use with velvet [default: #{options[:velvet_kmer_size]}]") do |arg|
|
62
|
+
options[:velvet_kmer_size] = arg.to_i
|
63
|
+
end
|
64
|
+
|
65
|
+
opts.separator "\nDebug-related options:\n\n"
|
66
|
+
|
67
|
+
|
68
|
+
|
69
|
+
# logger options
|
70
|
+
opts.separator "\nVerbosity:\n\n"
|
71
|
+
opts.on("-q", "--quiet", "Run quietly, set logging to ERROR level [default INFO]") {options[:log_level] = 'error'}
|
72
|
+
opts.on("--logger filename",String,"Log to file [default #{options[:logger]}]") { |name| options[:logger] = name}
|
73
|
+
opts.on("--trace options",String,"Set log level [default INFO]. e.g. '--trace debug' to set logging level to DEBUG"){|s| options[:log_level] = s}
|
74
|
+
end; o.parse!
|
75
|
+
if ARGV.length != 0 or options[:reads_file].nil? or options[:contig_file].nil? or options[:overall_trail_output_fasta_file].nil?
|
76
|
+
$stderr.puts o
|
77
|
+
exit 1
|
78
|
+
end
|
79
|
+
# Setup logging
|
80
|
+
Bio::Log::CLI.logger(options[:logger]); Bio::Log::CLI.trace(options[:log_level]); log = Bio::Log::LoggerPlus.new(LOG_NAME); Bio::Log::CLI.configure(LOG_NAME)
|
81
|
+
Bio::Log::LoggerPlus.new 'bio-velvet'; Bio::Log::CLI.configure 'bio-velvet'
|
82
|
+
log.outputters[0].formatter = Log4r::PatternFormatter.new(:pattern => "%5l %c %d: %m", :date_pattern => '%d/%m %T')
|
83
|
+
|
84
|
+
log.debug "Running finishm with options: #{PP.pp(options, "").gsub(/\n$/,'')}" if log.debug?
|
85
|
+
|
86
|
+
# Find where the Ns are
|
87
|
+
n_region_start = nil
|
88
|
+
n_region_end = nil
|
89
|
+
sequence = nil
|
90
|
+
Bio::FlatFile.foreach(options[:contig_file]) do |seq|
|
91
|
+
if sequence
|
92
|
+
raise Exception, "Sorry, this script can only handle single sequences to be gap filled at the moment"
|
93
|
+
end
|
94
|
+
|
95
|
+
sequence = seq.seq
|
96
|
+
|
97
|
+
if options[:start_offset] and options[:end_offset]
|
98
|
+
log.info "Trying to gap fill from #{options[:start_offset]+1} to #{options[:end_offset]+1}"
|
99
|
+
n_region_start = options[:start_offset]
|
100
|
+
n_region_end = options[:end_offset]
|
101
|
+
else
|
102
|
+
log.info "Determining where to fill from the presence of Ns"
|
103
|
+
|
104
|
+
matches = sequence.match(/(N+)/i)
|
105
|
+
if !matches
|
106
|
+
raise "Unable to find any gaps in the input sequence. That was a bit too easy.."
|
107
|
+
end
|
108
|
+
n_region_start = matches.offset(0)[0]
|
109
|
+
n_region_end = n_region_start + matches[1].length
|
110
|
+
log.info "Detected a gap between #{n_region_start} and #{n_region_end}"
|
111
|
+
end
|
112
|
+
|
113
|
+
# Check to make sure we are sufficiently distant from the ends
|
114
|
+
if n_region_start < options[:contig_end_length] or
|
115
|
+
sequence.length - n_region_end < options[:contig_end_length]
|
116
|
+
raise "The gap is too close to the end of the contig, sorry"
|
117
|
+
end
|
118
|
+
end
|
119
|
+
|
120
|
+
# Do the assembly
|
121
|
+
graph = nil
|
122
|
+
if options[:previously_serialized_parsed_graph_file].nil?
|
123
|
+
velvet_result = nil
|
124
|
+
if options[:previous_assembly].nil? #If assembly has not already been carried out
|
125
|
+
Tempfile.open('anchors.fa') do |tempfile|
|
126
|
+
tempfile.puts ">anchor1"
|
127
|
+
tempfile.puts sequence[n_region_start-options[:contig_end_length]-1...n_region_start]
|
128
|
+
tempfile.puts ">anchor2"
|
129
|
+
#Have to be in reverse, because the node finder finds the node at the start of the read, not the end
|
130
|
+
fwd2 = Bio::Sequence::NA.new(sequence[n_region_end..(n_region_end+options[:contig_end_length])])
|
131
|
+
tempfile.puts fwd2.reverse_complement.to_s
|
132
|
+
tempfile.close
|
133
|
+
log.debug "Inputting anchors into the assembly: #{File.open(tempfile.path).read}" if log.debug?
|
134
|
+
|
135
|
+
log.info "Assembling sampled reads with velvet"
|
136
|
+
# Bit of a hack, but have to use -short1 as the anchors because then start and end anchors will have node IDs 1,2,... etc.
|
137
|
+
velvet_result = Bio::Velvet::Runner.new.velvet(
|
138
|
+
options[:velvet_kmer_size],
|
139
|
+
"-short #{tempfile.path} -short2 -fastq.gz #{options[:reads_file]}",
|
140
|
+
"-read_trkg yes -cov_cutoff #{options[:assembly_coverage_cutoff]}",
|
141
|
+
:output_assembly_path => options[:output_assembly_path]
|
142
|
+
)
|
143
|
+
if log.debug?
|
144
|
+
log.debug "velveth stdout: #{velvet_result.velveth_stdout}"
|
145
|
+
log.debug "velveth stderr: #{velvet_result.velveth_stderr}"
|
146
|
+
log.debug "velvetg stdout: #{velvet_result.velvetg_stdout}"
|
147
|
+
log.debug "velvetg stderr: #{velvet_result.velvetg_stderr}"
|
148
|
+
end
|
149
|
+
log.info "Finished running assembly"
|
150
|
+
end
|
151
|
+
else
|
152
|
+
log.info "Using previous assembly stored at #{options[:previous_assembly]}"
|
153
|
+
velvet_result = Bio::Velvet::Result.new
|
154
|
+
velvet_result.result_directory = options[:previous_assembly]
|
155
|
+
end
|
156
|
+
|
157
|
+
log.info "Parsing the graph output from velvet"
|
158
|
+
graph = Bio::Velvet::Graph.parse_from_file(File.join velvet_result.result_directory, 'LastGraph')
|
159
|
+
log.info "Finished parsing graph: found #{graph.nodes.length} nodes and #{graph.arcs.length} arcs"
|
160
|
+
|
161
|
+
if options[:serialize_parsed_graph_file]
|
162
|
+
log.info "Storing a binary version of the graph file for later use at #{options[:serialize_parsed_graph_file]}"
|
163
|
+
File.open(options[:serialize_parsed_graph_file],'wb') do |f|
|
164
|
+
f.print Marshal.dump(graph)
|
165
|
+
end
|
166
|
+
log.info "Stored a binary representation of the velvet graph at #{options[:serialize_parsed_graph_file]}"
|
167
|
+
end
|
168
|
+
|
169
|
+
if options[:assembly_coverage_cutoff]
|
170
|
+
log.info "Removing low-coverage nodes from the graph (less than #{options[:assembly_coverage_cutoff]})"
|
171
|
+
cutoffer = Bio::AssemblyGraphAlgorithms::CoverageBasedGraphFilter.new
|
172
|
+
deleted_nodes, deleted_arcs = cutoffer.remove_low_coverage_nodes(graph, options[:assembly_coverage_cutoff], :whitelisted_sequences => [1,2])
|
173
|
+
|
174
|
+
log.info "Removed #{deleted_nodes.length} nodes and #{deleted_arcs.length} arcs from the graph due to low coverage"
|
175
|
+
log.info "Now there is #{graph.nodes.length} nodes and #{graph.arcs.length} arcs remaining"
|
176
|
+
end
|
177
|
+
else
|
178
|
+
log.info "Restoring graph file from #{options[:previously_serialized_parsed_graph_file]}.."
|
179
|
+
graph = Marshal.load(File.open(options[:previously_serialized_parsed_graph_file]))
|
180
|
+
log.info "Restoration complete"
|
181
|
+
end
|
182
|
+
|
183
|
+
|
184
|
+
# Find the anchor nodes again
|
185
|
+
finder = Bio::AssemblyGraphAlgorithms::NodeFinder.new
|
186
|
+
log.info "Finding node representing the end of the each contig"
|
187
|
+
i = 1
|
188
|
+
anchor_sequence_ids = [1,2]
|
189
|
+
start_node, start_node_forward = finder.find_unique_node_with_sequence_id(graph, 1)
|
190
|
+
end_node, end_node_forward = finder.find_unique_node_with_sequence_id(graph, 2)
|
191
|
+
if start_node and end_node
|
192
|
+
log.info "Found both anchoring nodes in the graph: #{start_node.node_id}/#{start_node_forward} and #{end_node.node_id}/#{end_node_forward}"
|
193
|
+
else
|
194
|
+
log.error "start node not found" if start_node.nil?
|
195
|
+
log.error "end node not found" if end_node.nil?
|
196
|
+
raise "Unable to find both anchor reads from the assembly, cannot continue. This is probably an error with this script, not you."
|
197
|
+
end
|
198
|
+
|
199
|
+
log.info "Removing nodes unconnected to either the start or the end from the graph.."
|
200
|
+
original_num_nodes = graph.nodes.length
|
201
|
+
original_num_arcs = graph.arcs.length
|
202
|
+
filter = Bio::AssemblyGraphAlgorithms::ConnectivityBasedGraphFilter.new
|
203
|
+
filter.remove_unconnected_nodes(graph, [start_node, end_node])
|
204
|
+
log.info "Removed #{original_num_nodes-graph.nodes.length} nodes and #{original_num_arcs-graph.arcs.length} arcs"
|
205
|
+
|
206
|
+
|
207
|
+
if options[:output_graph_png]
|
208
|
+
log.info "Converting assembly to a graphviz PNG"
|
209
|
+
viser = Bio::Assembly::ABVisualiser.new
|
210
|
+
gv = viser.graphviz(graph, {:start_node_id => start_node.node_id, :end_node_id => end_node.node_id})
|
211
|
+
gv.output :png => options[:output_graph_png], :use => :neato
|
212
|
+
end
|
213
|
+
if options[:output_graph_svg]
|
214
|
+
log.info "Converting assembly to a graphviz SVG"
|
215
|
+
viser = Bio::Assembly::ABVisualiser.new
|
216
|
+
gv = viser.graphviz(graph, {:start_node_id => start_node.node_id, :end_node_id => end_node.node_id})
|
217
|
+
gv.output :svg => options[:output_graph_svg], :use => :neato
|
218
|
+
end
|
219
|
+
if options[:output_graph_dot]
|
220
|
+
log.info "Converting assembly to a graphviz DOT"
|
221
|
+
viser = Bio::Assembly::ABVisualiser.new
|
222
|
+
gv = viser.graphviz(graph, {:start_node_id => start_node.node_id, :end_node_id => end_node.node_id, :digraph => false})
|
223
|
+
gv.output :dot => options[:output_graph_dot]
|
224
|
+
end
|
225
|
+
|
226
|
+
|
227
|
+
|
228
|
+
log.info "Searching for trails between the nodes within the assembly graph"
|
229
|
+
cartographer = Bio::AssemblyGraphAlgorithms::AcyclicConnectionFinder.new
|
230
|
+
trails = cartographer.find_trails_between_nodes(graph, start_node, end_node, options[:graph_search_leash_length], start_node_forward)
|
231
|
+
log.info "Found #{trails.length} trail(s) in total"
|
232
|
+
|
233
|
+
|
234
|
+
log.debug "Outputing trail sequences"
|
235
|
+
File.open(options[:overall_trail_output_fasta_file],'w') do |f|
|
236
|
+
trails.each_with_index do |trail, i|
|
237
|
+
f.puts ">trail#{i+1}"
|
238
|
+
f.puts trail.sequence
|
239
|
+
end
|
240
|
+
end
|
241
|
+
|