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 +1,303 @@
|
|
1
|
+
class Bio::FinishM::Finisher
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
def add_options(opts, options)
|
5
|
+
opts.banner = "\nUsage: finishm finish <options>\n\n"
|
6
|
+
|
7
|
+
options.merge!({
|
8
|
+
:min_leftover_length => false,
|
9
|
+
:kmer_coverage_target => 1,
|
10
|
+
:contig_end_length => 300,
|
11
|
+
:graph_search_leash_length => 20000,
|
12
|
+
:reads_to_assemble => nil,
|
13
|
+
})
|
14
|
+
|
15
|
+
# TODO: make a better interface for this. Maybe specify an entire genome, and then "Contig_1 end, Contig_3 start" or something
|
16
|
+
# Look at the last 300bp of the first contig.
|
17
|
+
extract_exactly_one_contig_from_file = lambda do |fasta_file_path|
|
18
|
+
contig = nil
|
19
|
+
Bio::FlatFile.foreach(Bio::FastaFormat, fasta_file_path) do |e|
|
20
|
+
if contig.nil?
|
21
|
+
contig = e.seq
|
22
|
+
else
|
23
|
+
raise "Multiple sequences found in a contig file! I need exactly one"
|
24
|
+
end
|
25
|
+
end
|
26
|
+
raise "I need a contig to be in the start contig file" if contig.nil?
|
27
|
+
Bio::Sequence::NA.new(contig.to_s)
|
28
|
+
end
|
29
|
+
|
30
|
+
opts.on("--pattern PATTERN", "kmer abundance pattern e.g. '0111001110' [required]") do |arg|
|
31
|
+
options[:pattern] = arg
|
32
|
+
end
|
33
|
+
opts.on("--kmer-abundances FILE", "kmer multiple abundance file [required]") do |arg|
|
34
|
+
options[:kmer_multiple_abundance_file] = arg
|
35
|
+
end
|
36
|
+
opts.on("--upper-threshold NUM", "kmer frequency cutoff to saying 'present' [required]") do |arg|
|
37
|
+
options[:upper_threshold] = arg.to_i
|
38
|
+
end
|
39
|
+
opts.on("--lower-threshold NUM", "kmer frequency cutoff to saying 'not present' [required]") do |arg|
|
40
|
+
options[:lower_threshold] = arg.to_i
|
41
|
+
end
|
42
|
+
opts.on("--reads FILES", "comma-separated list of sequence reads files in the same order as the pattern was supplied [required]") do |arg|
|
43
|
+
options[:reads_files] = arg.split(',').collect{|r| File.absolute_path r}
|
44
|
+
end
|
45
|
+
opts.on("--start-contig FASTA", "path to a fasta file with the starting contig in it (only). Assumes we are building off the end of this contig [required]") do |arg|
|
46
|
+
options[:start_contig] = extract_exactly_one_contig_from_file.call arg
|
47
|
+
end
|
48
|
+
opts.on("--end-contig FASTA", "path to a fasta file with the ending contig in it (only). Assumes we are building onto the start of this contig [required]") do |arg|
|
49
|
+
options[:end_contig] = extract_exactly_one_contig_from_file.call arg
|
50
|
+
end
|
51
|
+
|
52
|
+
opts.separator "\nOptional arguments:\n\n"
|
53
|
+
opts.on("--min-leftover-read-length NUMBER", "when searching for reads with kmers, require the kmer to be at the beginning or end of the selected read [default: #{options[:min_leftover_length]}]") do |arg|
|
54
|
+
options[:min_leftover_length] = arg.to_i
|
55
|
+
end
|
56
|
+
opts.on("--kmer-coverage-target NUMBER", "when searching for reads with kmers, require this many copies per kmer [default: #{options[:kmer_coverage_target]}]") do |arg|
|
57
|
+
options[:kmer_coverage_target] = arg.to_i
|
58
|
+
end
|
59
|
+
opts.on("--already-patterned-reads FILE", "Attempt to assemble the reads in the specified file, useful for re-assembly [default: off]") do |arg|
|
60
|
+
options[:already_patterned_reads] = arg
|
61
|
+
end
|
62
|
+
opts.on("--assembly-png PATH", "Output assembly as a PNG file [default: off]") do |arg|
|
63
|
+
options[:output_graph_png] = arg
|
64
|
+
end
|
65
|
+
opts.on("--assembly-svg PATH", "Output assembly as an SVG file [default: off]") do |arg|
|
66
|
+
options[:output_graph_svg] = arg
|
67
|
+
end
|
68
|
+
opts.on("--assembly-dot PATH", "Output assembly as an DOT file [default: off]") do |arg|
|
69
|
+
options[:output_graph_dot] = arg
|
70
|
+
end
|
71
|
+
opts.on("--assembly-coverage-cutoff NUMBER", "Require this much coverage in each node, all other nodes are removed [default: #{options[:assembly_coverage_cutoff]}]") do |arg|
|
72
|
+
options[:assembly_coverage_cutoff] = arg.to_f
|
73
|
+
end
|
74
|
+
opts.on("--contig-end-length LENGTH", "Number of base pairs to start into the ends of the contigs [default: #{options[:contig_end_length]}]") do |arg|
|
75
|
+
options[:contig_end_length] = arg.to_i
|
76
|
+
end
|
77
|
+
|
78
|
+
Bio::FinishM::GraphGenerator.new.add_options opts, options
|
79
|
+
end
|
80
|
+
|
81
|
+
def validate_options(options, argv)
|
82
|
+
#TODO: give a better description of the error that has occurred
|
83
|
+
if argv.length != 0
|
84
|
+
return "Dangling argument(s) found e.g. #{argv[0]}"
|
85
|
+
elsif options[:already_patterned_reads]
|
86
|
+
else
|
87
|
+
[:upper_threshold,
|
88
|
+
:lower_threshold,
|
89
|
+
:pattern,
|
90
|
+
:kmer_multiple_abundance_file,
|
91
|
+
:reads_files].each do |sym|
|
92
|
+
if options[sym].nil?
|
93
|
+
return "No option found to specify #{sym}"
|
94
|
+
end
|
95
|
+
end
|
96
|
+
end
|
97
|
+
return nil #if here, options all were parsed successfully
|
98
|
+
end
|
99
|
+
|
100
|
+
#TODO: this method is too long - split it up by refactoring
|
101
|
+
def run(options, argv)
|
102
|
+
pooled_reads_filename = 'pooled_sampled_reads.fasta' #TODO: remove this constant into a tempfile or something.
|
103
|
+
if options[:already_patterned_reads] #If skipping read extraction
|
104
|
+
pooled_reads_filename = options[:already_patterned_reads]
|
105
|
+
|
106
|
+
else
|
107
|
+
# Parse pattern from cmdline
|
108
|
+
desired_pattern = KmerAbundancePattern.new
|
109
|
+
desired_pattern.parse_from_human(options[:pattern])
|
110
|
+
if options[:reads_files].length != desired_pattern.length
|
111
|
+
raise "Number of entries in the pattern #{desired_pattern.length} and number of reads files #{options[:reads].length} not equivalent!"
|
112
|
+
end
|
113
|
+
|
114
|
+
# Collect the kmers that will be used to find trusted reads i.e.
|
115
|
+
# Go through each line of the kmer abundance file, looking for kmers that suit the pattern
|
116
|
+
input_file = File.open options[:kmer_multiple_abundance_file]
|
117
|
+
csv = CSV.new(input_file, :col_sep => ' ')
|
118
|
+
|
119
|
+
whitelist_kmers = []
|
120
|
+
blacklist_kmers = []
|
121
|
+
csv.each do |row|
|
122
|
+
max_i = row.length - 2 if max_i.nil?
|
123
|
+
|
124
|
+
kmer = row[0]
|
125
|
+
counts = row[1...row.length].collect{|s| s.to_i}
|
126
|
+
this_pattern = []
|
127
|
+
counts.each_with_index do |count, i|
|
128
|
+
if count > options[:upper_threshold]
|
129
|
+
this_pattern[i] = true
|
130
|
+
elsif count < options[:lower_threshold]
|
131
|
+
this_pattern[i] = false
|
132
|
+
else
|
133
|
+
# coverage was in no man's land between thresholds.
|
134
|
+
# Ignore this kmer as noise.
|
135
|
+
this_pattern[i] = '-'
|
136
|
+
end
|
137
|
+
end
|
138
|
+
#log.debug "Found pattern #{this_pattern} from kmer #{kmer}, which has abundances #{counts}" if log.debug?
|
139
|
+
|
140
|
+
if desired_pattern.consistent_with? this_pattern
|
141
|
+
whitelist_kmers.push row[0]
|
142
|
+
else
|
143
|
+
# kmer is not present when it should be
|
144
|
+
blacklist_kmers.push row[0]
|
145
|
+
end
|
146
|
+
end
|
147
|
+
log.info "After parsing the kmer multiple abundance file, found #{whitelist_kmers.length} kmers that matched the pattern, and #{blacklist_kmers.length} that didn't"
|
148
|
+
unless whitelist_kmers.length > 0
|
149
|
+
log.error "No kmers found that satisfy the given pattern, exiting.."
|
150
|
+
exit 1
|
151
|
+
end
|
152
|
+
|
153
|
+
|
154
|
+
#outdir = options[:output_directory]
|
155
|
+
#Dir.mkdir outdir unless Dir.exist?(outdir)
|
156
|
+
|
157
|
+
# grep the pattern out from the raw reads, subsampling so as to not overwhelm the assembler
|
158
|
+
#Tempfile.open('whitelist') do |white|
|
159
|
+
File.open 'whitelist', 'w' do |white| #TODO: remove 'whitelist' file as a constant
|
160
|
+
white.puts whitelist_kmers.join("\n")
|
161
|
+
white.close
|
162
|
+
|
163
|
+
#Tempfile.open('blacklist') do |black|
|
164
|
+
File.open('black','w') do |black|
|
165
|
+
black.puts blacklist_kmers.join("\n")
|
166
|
+
black.close
|
167
|
+
|
168
|
+
threadpool = []
|
169
|
+
sampled_read_files = []
|
170
|
+
log.info "Extracting reads that contain suitable kmers"
|
171
|
+
options[:reads_files].each_with_index do |file, i|
|
172
|
+
next unless desired_pattern[i] #Don't extract reads from reads where those reads should not have been amplified
|
173
|
+
|
174
|
+
sampled = File.basename(file)+'.sampled_reads.fasta'
|
175
|
+
sampled_read_files.push sampled
|
176
|
+
|
177
|
+
grep_path = "#{ ENV['HOME'] }/git/priner/bin/read_selection_by_kmer " #TODO: this won't work on other people's systems.
|
178
|
+
if options[:min_leftover_length]
|
179
|
+
grep_path += "--min-leftover-length #{options[:min_leftover_length]} "
|
180
|
+
end
|
181
|
+
thr = Thread.new do
|
182
|
+
grep_cmd = "#{grep_path} --whitelist #{white.path} --blacklist #{black.path} --reads #{file} --kmer-coverage-target #{options[:kmer_coverage_target]} > #{sampled}"
|
183
|
+
log.debug "Running cmd: #{grep_cmd}"
|
184
|
+
status, stdout, stderr = systemu grep_cmd
|
185
|
+
log.debug stderr
|
186
|
+
|
187
|
+
raise unless status.exitstatus == 0
|
188
|
+
log.debug "Finished extracting reads from #{file}"
|
189
|
+
end
|
190
|
+
threadpool.push thr
|
191
|
+
end
|
192
|
+
threadpool.each do |thread| thread.join; end #wait until everything is finito
|
193
|
+
|
194
|
+
log.info "Finished extracting reads for sampling. Now pooling sampled reads"
|
195
|
+
pool_cmd = "cat #{sampled_read_files.join ' '} >#{pooled_reads_filename}"
|
196
|
+
log.debug "Running cmd: #{pool_cmd}"
|
197
|
+
status, stdout, stderr = systemu pool_cmd
|
198
|
+
raise stderr if stderr != ''
|
199
|
+
raise unless status.exitstatus == 0
|
200
|
+
end
|
201
|
+
end
|
202
|
+
end
|
203
|
+
|
204
|
+
log.info "Extracting dummy reads from the ends of contigs to use as anchors"
|
205
|
+
start_contig = options[:start_contig]
|
206
|
+
end_contig = options[:end_contig]
|
207
|
+
if [start_contig.length, end_contig.length].min < 2*options[:contig_end_length]
|
208
|
+
log.warn "Choice of initial/terminal nodes to perform graph search with may not be optimal due to the small contig size"
|
209
|
+
end
|
210
|
+
if [start_contig.length, end_contig.length].min < options[:contig_end_length]
|
211
|
+
log.error "At least one contig too small to proceed with current code base, need to fix the code to allow such a small contig"
|
212
|
+
exit 1
|
213
|
+
end
|
214
|
+
|
215
|
+
probe_sequences = [
|
216
|
+
start_contig[start_contig.length-options[:contig_end_length]...start_contig.length],
|
217
|
+
Bio::Sequence::NA.new(end_contig[0...options[:contig_end_length]]).reverse_complement.to_s
|
218
|
+
]
|
219
|
+
read_input = Bio::FinishM::ReadInput.new
|
220
|
+
read_input.fasta_singles = [pooled_reads_filename]
|
221
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
222
|
+
graph = finishm_graph.graph
|
223
|
+
start_node = finishm_graph.probe_nodes[0]
|
224
|
+
start_node_forward = finishm_graph.probe_node_directions[0]
|
225
|
+
end_node = finishm_graph.probe_nodes[1]
|
226
|
+
end_node_forward = finishm_graph.probe_node_directions[1]
|
227
|
+
|
228
|
+
log.info "Node(s) found that are suitable as initial and terminal nodes in the graph search, respectively: #{start_node.node_id} and #{end_node.node_id}"
|
229
|
+
|
230
|
+
log.info "Removing nodes unconnected to either the start or the end from the graph.."
|
231
|
+
original_num_nodes = graph.nodes.length
|
232
|
+
original_num_arcs = graph.arcs.length
|
233
|
+
filter = Bio::AssemblyGraphAlgorithms::ConnectivityBasedGraphFilter.new
|
234
|
+
filter.remove_unconnected_nodes(graph, [start_node, end_node])
|
235
|
+
log.info "Removed #{original_num_nodes-graph.nodes.length} nodes and #{original_num_arcs-graph.arcs.length} arcs"
|
236
|
+
|
237
|
+
if options[:output_graph_png] or options[:output_graph_svg] or options[:output_graph_dot]
|
238
|
+
viser = Bio::Assembly::ABVisualiser.new
|
239
|
+
log.info "Preparing GraphViz object for output"
|
240
|
+
gv = viser.graphviz(graph, {:start_node_id => start_node.node_id, :end_node_id => end_node.node_id})
|
241
|
+
|
242
|
+
if options[:output_graph_png]
|
243
|
+
log.info "Converting assembly to a graphviz PNG #{options[:output_graph_png] }"
|
244
|
+
gv.output :png => options[:output_graph_png], :use => :neato
|
245
|
+
end
|
246
|
+
if options[:output_graph_svg]
|
247
|
+
log.info "Converting assembly to a graphviz SVG #{options[:output_graph_svg] }"
|
248
|
+
gv.output :svg => options[:output_graph_svg], :use => :neato
|
249
|
+
end
|
250
|
+
if options[:output_graph_dot]
|
251
|
+
log.info "Converting assembly to a graphviz DOT #{options[:output_graph_dot] }"
|
252
|
+
gv.output :dot => options[:output_graph_dot]
|
253
|
+
end
|
254
|
+
end
|
255
|
+
|
256
|
+
log.info "Searching for trails between the initial and terminal nodes, within the assembly graph"
|
257
|
+
cartographer = Bio::AssemblyGraphAlgorithms::AcyclicConnectionFinder.new
|
258
|
+
#raise "Untested connection finder below"
|
259
|
+
#trails = cartographer.find_all_trails_between_nodes(graph, start_node, end_node, options[:graph_search_leash_length], start_node_forward)
|
260
|
+
trails = cartographer.find_trails_between_nodes(graph, start_node, end_node, options[:graph_search_leash_length], start_node_forward)
|
261
|
+
log.info "Found #{trails.length} trail(s) between the initial and terminal nodes"
|
262
|
+
|
263
|
+
# log.info "Reading kmer abundances from #{options[:kmer_multiple_abundance_file]}.."
|
264
|
+
# kmer_hash = Bio::KmerMultipleAbundanceHash.parse_from_file options[:kmer_multiple_abundance_file]
|
265
|
+
# log.info "Finished reading the kmer abundances"
|
266
|
+
|
267
|
+
# if options[:trail_kmer_coverage_file]
|
268
|
+
# log.info "Writing out kmer coverages to #{options[:trail_kmer_coverage_file]}.."
|
269
|
+
# writer = Bio::AssemblyGraphAlgorithms::KmerCoverageWriter.new
|
270
|
+
# io = File.open(options[:trail_kmer_coverage_file],'w')
|
271
|
+
# writer.write(io, trails, kmer_hash)
|
272
|
+
# log.info "Finished writing"
|
273
|
+
# end
|
274
|
+
|
275
|
+
# log.info "Filtering trail(s) based on kmer coverage, requiring each kmer in the path to have a minimum of #{options[:kmer_path_filter_min_coverage]} coverage in patterned reads, except for the #{options[:kmer_path_end_exclusion_length]}bp at the ends"
|
276
|
+
# kmer_path_filter = Bio::AssemblyGraphAlgorithms::KmerCoverageBasedPathFilter.new
|
277
|
+
# thresholds = desired_pattern.collect{|c| c == true ? 1 : 0}
|
278
|
+
# log.info "Using thresholds for filtering: #{thresholds}"
|
279
|
+
# trails = kmer_path_filter.filter(trails, kmer_hash, thresholds, :exclude_ending_length => options[:kmer_path_end_exclusion_length])
|
280
|
+
# log.info "After filtering remained #{trails.length} trails"
|
281
|
+
|
282
|
+
printer = Bio::AssemblyGraphAlgorithms::ContigPrinter.new
|
283
|
+
trails.each_with_index do |trail, i|
|
284
|
+
log.debug "Before attachment to the contig, sequence of the trail was #{trail.sequence}" if log.debug?
|
285
|
+
acon = Bio::AssemblyGraphAlgorithms::ContigPrinter::AnchoredConnection.new
|
286
|
+
acon.start_probe_read_id = 1
|
287
|
+
acon.end_probe_read_id = 2
|
288
|
+
acon.start_probe_node = start_node
|
289
|
+
acon.end_probe_node = end_node
|
290
|
+
acon.start_probe_contig_offset = options[:contig_end_length]
|
291
|
+
acon.end_probe_contig_offset = options[:contig_end_length]
|
292
|
+
acon.paths = [trail]
|
293
|
+
log.debug "AnchoredConnection object to print for this trail: #{acon.inspect}" if log.debug?
|
294
|
+
|
295
|
+
puts ">trail#{i+1}"
|
296
|
+
puts printer.one_connection_between_two_contigs(
|
297
|
+
finishm_graph.graph,
|
298
|
+
probe_sequences[0],
|
299
|
+
acon,
|
300
|
+
probe_sequences[1])
|
301
|
+
end
|
302
|
+
end
|
303
|
+
end
|
@@ -0,0 +1,122 @@
|
|
1
|
+
class Bio::FinishM::Fluff
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
def add_options(optparse_object, options)
|
5
|
+
optparse_object.banner = "\nUsage: finishm fluff --contigs <contig_file> --fastq-gz <reads..> --output-fluff-file <output.fa>
|
6
|
+
|
7
|
+
Takes a set of contigs, and places probes across them (e.g. every 2kb), and then explores the
|
8
|
+
graph from each of these probes, taking all paths within some leash length, including the 'fluff'
|
9
|
+
which is not the same path as along the contig. Prints out all of these paths to a fasta file.\n\n"
|
10
|
+
|
11
|
+
options.merge!({
|
12
|
+
:probe_spacing => 2000,
|
13
|
+
:probe_length => 100,
|
14
|
+
:graph_search_leash_length => 20000,
|
15
|
+
})
|
16
|
+
|
17
|
+
optparse_object.separator "\nRequired arguments:\n\n"
|
18
|
+
optparse_object.on("--contigs FILE", "fasta file containing contigs to find the fluff on [required]") do |arg|
|
19
|
+
options[:contigs_file] = arg
|
20
|
+
end
|
21
|
+
optparse_object.on("--output-fluff-file PATH", "Output found paths to this file in fasta format [required]") do |arg|
|
22
|
+
options[:output_fluff_file] = arg
|
23
|
+
end
|
24
|
+
optparse_object.separator "\nThere must be some definition of reads too:\n\n" #TODO improve this help
|
25
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
26
|
+
|
27
|
+
optparse_object.separator "\nOptional arguments:\n\n"
|
28
|
+
optparse_object.on("--probe-spacing NUM", Integer, "Distance between probe points in the contig [default: #{options[:probe_spacing]}]") do |arg|
|
29
|
+
options[:probe_spacing] = arg
|
30
|
+
end
|
31
|
+
optparse_object.on("--probe-size NUM", Integer, "Length of the probe to be inserted into the velvet graph. Must be greater than graph kmer length. [default: #{options[:probe_length]}]") do |arg|
|
32
|
+
options[:probe_length] = arg
|
33
|
+
end
|
34
|
+
optparse_object.on("--leash-length NUM", Integer, "Don't explore too far in the graph, only this far and not much more [default: #{options[:graph_search_leash_length]}]") do |arg|
|
35
|
+
options[:graph_search_leash_length] = arg
|
36
|
+
end
|
37
|
+
optparse_object.on("--assembly-png PATH", "Output assembly as a PNG file [default: off]") do |arg|
|
38
|
+
options[:output_graph_png] = arg
|
39
|
+
end
|
40
|
+
optparse_object.on("--assembly-svg PATH", "Output assembly as an SVG file [default: off]") do |arg|
|
41
|
+
options[:output_graph_svg] = arg
|
42
|
+
end
|
43
|
+
optparse_object.on("--assembly-dot PATH", "Output assembly as an DOT file [default: off]") do |arg|
|
44
|
+
options[:output_graph_dot] = arg
|
45
|
+
end
|
46
|
+
|
47
|
+
Bio::FinishM::GraphGenerator.new.add_options optparse_object, options
|
48
|
+
end
|
49
|
+
|
50
|
+
def validate_options(options, argv)
|
51
|
+
#TODO: give a better description of the error that has occurred
|
52
|
+
#TODO: require reads options
|
53
|
+
if argv.length != 0
|
54
|
+
return "Dangling argument(s) found e.g. #{argv[0]}"
|
55
|
+
else
|
56
|
+
[
|
57
|
+
:contigs_file,
|
58
|
+
:output_fluff_file
|
59
|
+
].each do |sym|
|
60
|
+
if options[sym].nil?
|
61
|
+
return "No option found to specify #{sym}."
|
62
|
+
end
|
63
|
+
end
|
64
|
+
|
65
|
+
unless options[:velvet_kmer_size] < options[:probe_length]
|
66
|
+
return "The probe length must be greater than the kmer length, otherwise it will not be incorporated into the kmer graph"
|
67
|
+
end
|
68
|
+
|
69
|
+
#if return nil from here, options all were parsed successfully
|
70
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
71
|
+
end
|
72
|
+
end
|
73
|
+
|
74
|
+
def run(options, argv)
|
75
|
+
# Read in all the contigs sequences
|
76
|
+
probe_sequences = []
|
77
|
+
sequence_names = []
|
78
|
+
Bio::FlatFile.foreach(options[:contigs_file]) do |seq|
|
79
|
+
sequence_names.push seq.definition
|
80
|
+
|
81
|
+
sequence = seq.seq
|
82
|
+
0.step(sequence.length-1-options[:probe_length], options[:probe_spacing]) do |offset|
|
83
|
+
# Only probe in the forward direction
|
84
|
+
probe_sequence = sequence[offset...offset+options[:probe_length]]
|
85
|
+
probe_sequences.push probe_sequence
|
86
|
+
end
|
87
|
+
end
|
88
|
+
log.info "Searching from #{probe_sequences.length} different probes from #{sequence_names.length} contigs)"
|
89
|
+
|
90
|
+
# Generate the graph with the probe sequences in it.
|
91
|
+
read_input = Bio::FinishM::ReadInput.new
|
92
|
+
read_input.parse_options options
|
93
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
94
|
+
|
95
|
+
# Loop over the ends, trying to make connections from each one
|
96
|
+
fluffer = Bio::AssemblyGraphAlgorithms::Fluffer.new
|
97
|
+
fluffings = fluffer.fluff(finishm_graph, options[:graph_search_leash_length])
|
98
|
+
log.debug "Found these fluffings: #{fluffings}" if log.debug?
|
99
|
+
log.info "Found #{fluffings.collect{|sets| sets.length}.reduce(:+)} paths in total" if log.info?
|
100
|
+
|
101
|
+
if options[:output_graph_png] or options[:output_graph_svg] or options[:output_graph_dot]
|
102
|
+
log.info "Converting assembly to a graphviz PNG"
|
103
|
+
viser = Bio::Assembly::ABVisualiser.new
|
104
|
+
gv = viser.graphviz(finishm_graph.graph, {:start_node_ids => finishm_graph.probe_nodes.collect{|node| node.node_id}})
|
105
|
+
|
106
|
+
gv.output :png => options[:output_graph_png], :use => :neato if options[:output_graph_png]
|
107
|
+
gv.output :svg => options[:output_graph_svg], :use => :neato if options[:output_graph_svg]
|
108
|
+
gv.output :dot => options[:output_graph_dot] if options[:output_graph_dot]
|
109
|
+
end
|
110
|
+
|
111
|
+
# Print out the sequences
|
112
|
+
File.open(options[:output_fluff_file], 'w') do |output|
|
113
|
+
fluffings.each_with_index do |path_set, probe_number|
|
114
|
+
path_set.each_with_index do |path, path_number|
|
115
|
+
fate = path_set.fates[path_number]
|
116
|
+
output.puts ">probe#{probe_number+1}_path#{path_number+1} #{fate}"
|
117
|
+
output.puts path.sequence
|
118
|
+
end
|
119
|
+
end
|
120
|
+
end
|
121
|
+
end
|
122
|
+
end
|
@@ -0,0 +1,325 @@
|
|
1
|
+
require 'tmpdir'
|
2
|
+
|
3
|
+
class Bio::FinishM::GapFiller
|
4
|
+
include Bio::FinishM::Logging
|
5
|
+
|
6
|
+
def add_options(optparse_object, options)
|
7
|
+
optparse_object.banner = "\nUsage: finishm gapfill --contigs <contigs_file> --fastq-gz <reads..> --output-fasta <output.fa>
|
8
|
+
|
9
|
+
Takes a set of reads and a contig that contains gap characters. Then it tries to fill in
|
10
|
+
these N characters. It is possible that there is multiple ways to close the gap - in that case
|
11
|
+
each can be reported.
|
12
|
+
|
13
|
+
example: finishm gapfill --contigs to_gapfill.fasta --fastq-gz reads.1.fq.gz,reads.2.fq.gz --output-fasta output.fasta
|
14
|
+
\n"
|
15
|
+
|
16
|
+
options.merge!({
|
17
|
+
:contig_end_length => 200,
|
18
|
+
:graph_search_leash_length => 20000,
|
19
|
+
})
|
20
|
+
|
21
|
+
optparse_object.separator "\nRequired arguments:\n\n"
|
22
|
+
optparse_object.on("--contigs FILE", "fasta file of single contig containing Ns that are to be closed [required]") do |arg|
|
23
|
+
options[:contigs_file] = arg
|
24
|
+
end
|
25
|
+
optparse_object.on("--output-fasta PATH", "Output the gap-filled sequence to this file [required]") do |arg|
|
26
|
+
options[:overall_fasta_file] = arg
|
27
|
+
end
|
28
|
+
|
29
|
+
optparse_object.separator "\nThere must be some definition of of how to do the assembly, or else a path to a previous assembly directory:\n\n"
|
30
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
31
|
+
Bio::FinishM::GraphGenerator.new.add_options optparse_object, options
|
32
|
+
|
33
|
+
optparse_object.separator "\nGraph search options:\n\n"
|
34
|
+
optparse_object.on("--overhang NUM", Integer, "Start assembling this many base pairs back from the gap [default: #{options[:contig_end_length] }]") do |arg|
|
35
|
+
options[:contig_end_length] = arg
|
36
|
+
end
|
37
|
+
optparse_object.on("--leash-length NUM", Integer, "Don't explore too far in the graph, only this many base pairs and not (much) more [default: #{options[:graph_search_leash_length] }]") do |arg|
|
38
|
+
options[:graph_search_leash_length] = arg
|
39
|
+
end
|
40
|
+
optparse_object.on("--recoherence-kmer NUM", Integer, "Use a kmer longer than the original velvet one, to help remove bubbles and circular paths [default: none]") do |arg|
|
41
|
+
options[:recoherence_kmer] = arg
|
42
|
+
end
|
43
|
+
|
44
|
+
optparse_object.separator "\nVisualisation options (of all joins):\n\n"
|
45
|
+
optparse_object.on("--assembly-png PATH", "Output assembly as a PNG file [default: off]") do |arg|
|
46
|
+
options[:output_graph_png] = arg
|
47
|
+
end
|
48
|
+
optparse_object.on("--assembly-svg PATH", "Output assembly as an SVG file [default: off]") do |arg|
|
49
|
+
options[:output_graph_svg] = arg
|
50
|
+
end
|
51
|
+
optparse_object.on("--assembly-dot PATH", "Output assembly as an DOT file [default: off]") do |arg|
|
52
|
+
options[:output_graph_dot] = arg
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
def validate_options(options, argv)
|
57
|
+
#TODO: give a better description of the error that has occurred
|
58
|
+
#TODO: require reads options
|
59
|
+
if argv.length != 0
|
60
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
61
|
+
else
|
62
|
+
[
|
63
|
+
:contigs_file,
|
64
|
+
:overall_fasta_file
|
65
|
+
].each do |sym|
|
66
|
+
if options[sym].nil?
|
67
|
+
return "No option found to specify #{sym}"
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
#if return nil from here, options all were parsed successfully
|
72
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
def run(options, argv)
|
77
|
+
# Read in all the contigs sequences and work out where the gaps are
|
78
|
+
genome = Bio::FinishM::InputGenome.new(
|
79
|
+
options[:contigs_file],
|
80
|
+
options[:contig_end_length],
|
81
|
+
options
|
82
|
+
)
|
83
|
+
|
84
|
+
|
85
|
+
scaffolds = Bio::FinishM::ScaffoldBreaker.new.break_scaffolds(options[:contigs_file])
|
86
|
+
gaps = []
|
87
|
+
output_fasta_file = File.open(options[:overall_fasta_file],'w')
|
88
|
+
num_without_gaps = 0
|
89
|
+
scaffolds.each do |scaffold|
|
90
|
+
sgaps = scaffold.gaps
|
91
|
+
if sgaps.empty?
|
92
|
+
num_without_gaps += 1
|
93
|
+
output_fasta_file.puts ">#{scaffold.name }"
|
94
|
+
output_fasta_file.puts scaffold.sequence
|
95
|
+
else
|
96
|
+
gaps.push scaffold.gaps
|
97
|
+
end
|
98
|
+
end
|
99
|
+
gaps.flatten!
|
100
|
+
log.info "Detected #{gaps.length} gap(s) from #{scaffolds.length} different contig(s). #{num_without_gaps } contig(s) were gap-free."
|
101
|
+
|
102
|
+
# Create probe sequences
|
103
|
+
probe_sequences = []
|
104
|
+
gaps.each do |gap|
|
105
|
+
sequence = gap.scaffold.sequence
|
106
|
+
|
107
|
+
if gap.start < options[:contig_end_length] or gap.stop > sequence.length - options[:contig_end_length]
|
108
|
+
log.warn "Found a gap that was too close to the end of a contig, skipping it: #{gap.coords}"
|
109
|
+
next
|
110
|
+
end
|
111
|
+
|
112
|
+
log.debug "Processing gap number #{gap.number}, #{gap.coords}"
|
113
|
+
first_coords = [
|
114
|
+
gap.start-options[:contig_end_length]-1,
|
115
|
+
gap.start-1,
|
116
|
+
]
|
117
|
+
second_coords = [
|
118
|
+
gap.stop,
|
119
|
+
(gap.stop+options[:contig_end_length]),
|
120
|
+
]
|
121
|
+
log.debug "Coordinates of the probes are #{first_coords} and #{second_coords}"
|
122
|
+
second = sequence[second_coords[0]..second_coords[1]]
|
123
|
+
probes = [
|
124
|
+
sequence[first_coords[0]...first_coords[1]],
|
125
|
+
Bio::Sequence::NA.new(second).reverse_complement.to_s,
|
126
|
+
]
|
127
|
+
#TODO: this could probably be handled better.. e.g. if the amount of sequence is too small, just throw it out and make one big gap
|
128
|
+
if probes[0].match(/N/i) or probes[1].match(/N/i)
|
129
|
+
log.warn "Noticed gap that was too close together, skipping: #{gap.coords}"
|
130
|
+
next
|
131
|
+
end
|
132
|
+
probe_sequences.push probes[0]
|
133
|
+
probe_sequences.push probes[1]
|
134
|
+
end
|
135
|
+
log.debug "Generated #{probe_sequences.length} probes e.g. #{probe_sequences[0] }"
|
136
|
+
|
137
|
+
|
138
|
+
# Generate the graph with the probe sequences in it.
|
139
|
+
read_input = Bio::FinishM::ReadInput.new
|
140
|
+
read_input.parse_options options
|
141
|
+
# Own the tmpdir, if one is to be used - need to re-read the LastGraph later on see..
|
142
|
+
assembly_directory = options[:output_assembly_path]
|
143
|
+
assembly_directory ||= options[:previous_assembly]
|
144
|
+
using_tmp_assembly_directory = false
|
145
|
+
if assembly_directory.nil?
|
146
|
+
using_tmp_assembly_directory = true
|
147
|
+
assembly_directory = Dir.mktmpdir
|
148
|
+
options[:output_assembly_path] = assembly_directory
|
149
|
+
end
|
150
|
+
|
151
|
+
# Do the actual graph building and/or initial reading
|
152
|
+
options[:parse_sequences] = true
|
153
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
154
|
+
|
155
|
+
# Output optional graphics.
|
156
|
+
if options[:output_graph_png] or options[:output_graph_svg] or options[:output_graph_dot]
|
157
|
+
viser = Bio::Assembly::ABVisualiser.new
|
158
|
+
# TODO: make these visualise more than one join somehow
|
159
|
+
gv = viser.graphviz(finishm_graph.graph, {
|
160
|
+
:start_node_id => finishm_graph.probe_nodes[0].node_id,
|
161
|
+
:end_node_id => finishm_graph.probe_nodes[1].node_id})
|
162
|
+
|
163
|
+
if options[:output_graph_png]
|
164
|
+
log.info "Converting assembly to a graphviz PNG"
|
165
|
+
gv.output :png => options[:output_graph_png], :use => :neato
|
166
|
+
end
|
167
|
+
if options[:output_graph_svg]
|
168
|
+
log.info "Converting assembly to a graphviz SVG"
|
169
|
+
gv.output :svg => options[:output_graph_svg], :use => :neato
|
170
|
+
end
|
171
|
+
if options[:output_graph_dot]
|
172
|
+
log.info "Converting assembly to a graphviz DOT"
|
173
|
+
gv.output :dot => options[:output_graph_dot]
|
174
|
+
end
|
175
|
+
end
|
176
|
+
|
177
|
+
# Clean up the tmdir, if one was used.
|
178
|
+
if using_tmp_assembly_directory
|
179
|
+
log.debug "Removing tmpdir that held the assembly `#{assembly_directory}'.."
|
180
|
+
FileUtils.remove_entry assembly_directory
|
181
|
+
end
|
182
|
+
|
183
|
+
# Do the gap-filling and print out the results
|
184
|
+
printer = Bio::AssemblyGraphAlgorithms::ContigPrinter.new
|
185
|
+
num_total_trails = 0
|
186
|
+
num_singly_filled = 0
|
187
|
+
num_unbridgable = 0
|
188
|
+
|
189
|
+
output_trails_file = nil
|
190
|
+
output_trails_file = File.open(options[:overall_trail_output_fasta_file],'w') unless options[:overall_trail_output_fasta_file].nil?
|
191
|
+
|
192
|
+
# Print the fasta output for the scaffold
|
193
|
+
print_scaffold = lambda do |last_scaffold, gapfilled_sequence|
|
194
|
+
output_fasta_file.puts ">#{last_scaffold.name }"
|
195
|
+
#gapfilled_sequence += last_scaffold.contigs[last_scaffold.contigs.length-1].sequence #add last contig
|
196
|
+
output_fasta_file.puts gapfilled_sequence
|
197
|
+
end
|
198
|
+
# Lambda to add a gap the the String representing the scaffold
|
199
|
+
#TODO: if the trail is not filled then the wrong sequence is currently printed. BUG???
|
200
|
+
filler = lambda do |anchored_connection, following_contig, gapfilled_sequence, gap|
|
201
|
+
gapfilled = nil
|
202
|
+
if anchored_connection.paths.length == 1
|
203
|
+
# If there is only 1 trail, then output scaffolding information
|
204
|
+
num_singly_filled += 1
|
205
|
+
|
206
|
+
gapfilled = printer.one_connection_between_two_contigs(
|
207
|
+
finishm_graph.graph,
|
208
|
+
gapfilled_sequence,
|
209
|
+
anchored_connection,
|
210
|
+
following_contig.sequence
|
211
|
+
)
|
212
|
+
else
|
213
|
+
# Otherwise don't make any assumptions
|
214
|
+
num_unbridgable += 1 if anchored_connection.paths.empty?
|
215
|
+
# TODO: even the there is multiple trails, better info can still be output here
|
216
|
+
gapfilled = gapfilled_sequence + 'N'*gap.length + following_contig.sequence
|
217
|
+
end
|
218
|
+
gapfilled #return this string
|
219
|
+
end
|
220
|
+
|
221
|
+
log.info "Searching for trails between the nodes within the assembly graph"
|
222
|
+
log.info "Using contig overhang length #{options[:contig_end_length] } and leash length #{options[:graph_search_leash_length] }"
|
223
|
+
gapfilled_sequence = ''
|
224
|
+
last_scaffold = nil
|
225
|
+
|
226
|
+
(0...(probe_sequences.length / 2)).collect{|i| i*2}.each do |start_probe_index|
|
227
|
+
gap_number = start_probe_index / 2
|
228
|
+
gap = gaps[gap_number]
|
229
|
+
log.info "Now working through gap number #{gap_number+1}: #{gap.coords}"
|
230
|
+
|
231
|
+
probe_index1 = start_probe_index
|
232
|
+
probe_index2 = start_probe_index+1
|
233
|
+
|
234
|
+
connection = gapfill(finishm_graph, probe_index1, probe_index2, options)
|
235
|
+
log.info "Found #{connection.paths.length} trails for #{gap.coords}"
|
236
|
+
|
237
|
+
unless output_trails_file.nil?
|
238
|
+
# print the sequences of the trails if asked for:
|
239
|
+
trails.each_with_index do |trail, i|
|
240
|
+
#TODO: need to output this as something more sensible e.g. VCF format
|
241
|
+
output_trails_file.puts ">#{gap.coords}_trail#{i+1}"
|
242
|
+
output_trails_file.puts trail.sequence
|
243
|
+
end
|
244
|
+
end
|
245
|
+
num_total_trails += connection.paths.length
|
246
|
+
|
247
|
+
# Output the updated sequence. Fill in the sequence if there is only 1 trail
|
248
|
+
if gap.scaffold == last_scaffold
|
249
|
+
# We are still building the current scaffold
|
250
|
+
#gapfilled_sequence += gap.scaffold.contigs[gap.number].sequence
|
251
|
+
log.debug "Before adding next chunk of contig, length of scaffold being built is #{gapfilled_sequence.length}" if log.debug?
|
252
|
+
gapfilled_sequence = filler.call connection, gap.scaffold.contigs[gap.number+1], gapfilled_sequence, gap
|
253
|
+
log.debug "After adding next chunk of contig, length of scaffold being built is #{gapfilled_sequence.length}" if log.debug?
|
254
|
+
else
|
255
|
+
# We are onto a new scaffold. Print the previous one (unless this the first one)
|
256
|
+
unless last_scaffold.nil?
|
257
|
+
# print the gapfilled (or not) scaffold.
|
258
|
+
print_scaffold.call(last_scaffold, gapfilled_sequence)
|
259
|
+
end
|
260
|
+
#reset
|
261
|
+
last_scaffold = gap.scaffold
|
262
|
+
|
263
|
+
#add the current gap (and the contig before it)
|
264
|
+
log.debug "Before adding first chunk of contig, length of scaffold being built is #{gapfilled_sequence.length}"
|
265
|
+
gapfilled_sequence = gap.scaffold.contigs[gap.number].sequence
|
266
|
+
log.debug "After adding first chunk of contig, length of scaffold being built is #{gapfilled_sequence.length}"
|
267
|
+
gapfilled_sequence = filler.call connection, gap.scaffold.contigs[gap.number+1], gapfilled_sequence, gap
|
268
|
+
log.debug "After adding first gap sequence and next contig, gapfilled sequence length is #{gapfilled_sequence.length}"
|
269
|
+
end
|
270
|
+
end
|
271
|
+
print_scaffold.call(last_scaffold, gapfilled_sequence) # print the last scaffold
|
272
|
+
|
273
|
+
log.info "#{num_unbridgable } gaps had no suitable bridging paths in the graph within the leash, and found #{num_total_trails} trails in total."
|
274
|
+
log.info "Filled #{num_singly_filled } out of #{gaps.length } gaps."
|
275
|
+
|
276
|
+
output_trails_file.close unless output_trails_file.nil?
|
277
|
+
output_fasta_file.close
|
278
|
+
end
|
279
|
+
|
280
|
+
# Given a finishm graph, gapfill from the first probe to the second. Return a
|
281
|
+
# Bio::AssemblyGraphAlgorithms::ContigPrinter::AnchoredConnection object
|
282
|
+
def gapfill(finishm_graph, probe_index1, probe_index2, options)
|
283
|
+
start_onode = finishm_graph.velvet_oriented_node(probe_index1)
|
284
|
+
end_onode_inward = finishm_graph.velvet_oriented_node(probe_index2)
|
285
|
+
unless start_onode and end_onode_inward
|
286
|
+
raise "Unable to retrieve both probes from the graph for gap #{gap_number} (#{gap.coords}), fail"
|
287
|
+
end
|
288
|
+
|
289
|
+
# The probe from finishm_graph points in the wrong direction for path finding
|
290
|
+
end_onode = Bio::Velvet::Graph::OrientedNodeTrail::OrientedNode.new
|
291
|
+
end_onode.node = end_onode_inward.node
|
292
|
+
end_onode.first_side = end_onode_inward.starts_at_start? ? Bio::Velvet::Graph::OrientedNodeTrail::END_IS_FIRST : Bio::Velvet::Graph::OrientedNodeTrail::START_IS_FIRST
|
293
|
+
|
294
|
+
adjusted_leash_length = finishm_graph.adjusted_leash_length(probe_index1, options[:graph_search_leash_length])
|
295
|
+
log.debug "Using adjusted leash length #{adjusted_leash_length }" if log.debug?
|
296
|
+
|
297
|
+
cartographer = Bio::AssemblyGraphAlgorithms::AcyclicConnectionFinder.new
|
298
|
+
trails = cartographer.find_trails_between_nodes(
|
299
|
+
finishm_graph.graph, start_onode, end_onode, adjusted_leash_length, {
|
300
|
+
:recoherence_kmer => options[:recoherence_kmer],
|
301
|
+
:sequences => finishm_graph.velvet_sequences,
|
302
|
+
:max_explore_nodes => options[:max_explore_nodes],
|
303
|
+
:max_gapfill_paths => options[:max_gapfill_paths],
|
304
|
+
}
|
305
|
+
)
|
306
|
+
if trails.circular_paths_detected
|
307
|
+
log.warn "Circular path detected here, not attempting to gapfill"
|
308
|
+
end
|
309
|
+
# Convert the trails into OrientedNodePaths
|
310
|
+
trails = trails.collect do |trail|
|
311
|
+
path = Bio::Velvet::Graph::OrientedNodeTrail.new
|
312
|
+
path.trail = trail
|
313
|
+
path
|
314
|
+
end
|
315
|
+
|
316
|
+
acon = Bio::AssemblyGraphAlgorithms::ContigPrinter::AnchoredConnection.new
|
317
|
+
acon.start_probe_noded_read = finishm_graph.probe_node_reads[probe_index1]
|
318
|
+
acon.end_probe_noded_read = finishm_graph.probe_node_reads[probe_index2]
|
319
|
+
acon.start_probe_contig_offset = options[:contig_end_length]
|
320
|
+
acon.end_probe_contig_offset = options[:contig_end_length]
|
321
|
+
acon.paths = trails
|
322
|
+
|
323
|
+
return acon
|
324
|
+
end
|
325
|
+
end
|