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 +1,270 @@
|
|
1
|
+
class Bio::FinishM::Wanderer
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
DEFAULT_OPTIONS = {
|
5
|
+
:contig_end_length => 200,
|
6
|
+
:graph_search_leash_length => 20000,
|
7
|
+
:unscaffold_first => false,
|
8
|
+
:recoherence_kmer => 1,
|
9
|
+
}
|
10
|
+
|
11
|
+
def add_options(optparse_object, options)
|
12
|
+
optparse_object.banner = "\nUsage: finishm wander --contigs <contig_file> --fastq-gz <reads..> --output-connections <output.csv> --output-scaffolds <output.fasta>
|
13
|
+
|
14
|
+
Takes a set of contigs/scaffolds from a genome and finds connections in the graph between them. A connection here is given as
|
15
|
+
the length of the shortest path between them, without actually computing all the paths.
|
16
|
+
|
17
|
+
This can be used for scaffolding, because if a contig end only connects to one other contig end, then
|
18
|
+
those contigs might be scaffolded together.
|
19
|
+
|
20
|
+
This method can also be used for 'pre-scaffolding', in the following sense. If the shortest path between
|
21
|
+
two contig ends is 10kb, and a mate pair library with insert size 2kb suggests a linkage
|
22
|
+
between the two ends, then the mate pair linkage is likely false (as long as there is sufficient
|
23
|
+
coverage in the reads, and not overwhelming amounts of strain heterogeneity, etc.).
|
24
|
+
|
25
|
+
Example:
|
26
|
+
|
27
|
+
finishm wander --contigs contigs.fasta --fastq-gz reads.1.fq.gz,reads.2.fq.gz --output-scaffolds scaffolds.fasta
|
28
|
+
|
29
|
+
That will create a collapsed de-Bruijn graph from reads.1.fq.gz and reads.2.fq.gz, then try to find connections between
|
30
|
+
the starts and the ends of the contigs in contigs.fasta through the de-Bruijn graph. The new scaffolds are then
|
31
|
+
output to scaffolds.fasta
|
32
|
+
|
33
|
+
\n\n"
|
34
|
+
|
35
|
+
options.merge!(DEFAULT_OPTIONS)
|
36
|
+
|
37
|
+
optparse_object.separator "\nRequired arguments:\n\n"
|
38
|
+
optparse_object.on("--contigs FILE", "fasta file of single contig containing Ns that are to be closed [required]") do |arg|
|
39
|
+
options[:contigs_file] = arg
|
40
|
+
end
|
41
|
+
|
42
|
+
optparse_object.separator "\nOutput modes:\n\n"
|
43
|
+
optparse_object.on("--output-scaffolds FILE", "Output scaffolds in FASTA format [required]") do |arg|
|
44
|
+
options[:output_scaffolds_file] = arg
|
45
|
+
end
|
46
|
+
optparse_object.on("--output-connections FILE", "Output connections in tab-separated format [required]") do |arg|
|
47
|
+
options[:output_connection_file] = arg
|
48
|
+
end
|
49
|
+
|
50
|
+
optparse_object.separator "\nThere must be some definition of reads too:\n\n" #TODO improve this help
|
51
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
52
|
+
|
53
|
+
optparse_object.separator "\nOptional arguments:\n\n"
|
54
|
+
optparse_object.on("--overhang NUM", Integer, "Start assembling this far from the ends of the contigs [default: #{options[:contig_end_length] }]") do |arg|
|
55
|
+
options[:contig_end_length] = arg.to_i
|
56
|
+
end
|
57
|
+
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|
|
58
|
+
options[:recoherence_kmer] = arg
|
59
|
+
end
|
60
|
+
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|
|
61
|
+
options[:graph_search_leash_length] = arg
|
62
|
+
end
|
63
|
+
optparse_object.on("--unscaffold-first", "Break the scaffolds in the contigs file apart, and then wander between the resultant contigs. [default: #{options[:unscaffold_first] }]") do
|
64
|
+
options[:unscaffold_first] = true
|
65
|
+
end
|
66
|
+
optparse_object.on("--proceed-on-short-contigs", "By default, when overly short contigs are encountered, finishm croaks. This option stops the croaking [default: #{options[:proceed_on_short_contigs] }]") do
|
67
|
+
options[:proceed_on_short_contigs] = true
|
68
|
+
end
|
69
|
+
|
70
|
+
Bio::FinishM::GraphGenerator.new.add_options optparse_object, options
|
71
|
+
end
|
72
|
+
|
73
|
+
def validate_options(options, argv)
|
74
|
+
#TODO: give a better description of the error that has occurred
|
75
|
+
#TODO: require reads options
|
76
|
+
if argv.length != 0
|
77
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
78
|
+
else
|
79
|
+
[
|
80
|
+
:contigs_file,
|
81
|
+
].each do |sym|
|
82
|
+
if options[sym].nil?
|
83
|
+
return "No option found to specify #{sym}."
|
84
|
+
end
|
85
|
+
end
|
86
|
+
if options[:output_scaffolds_file].nil? and
|
87
|
+
options[:output_connection_file].nil?
|
88
|
+
return "Need to specify either output scaffolds or output connections file"
|
89
|
+
end
|
90
|
+
|
91
|
+
#if return nil from here, options all were parsed successfully
|
92
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
93
|
+
end
|
94
|
+
end
|
95
|
+
|
96
|
+
def run(options, argv=[])
|
97
|
+
# Read in all the contigs sequences, removing those that are too short
|
98
|
+
probe_sequences = []
|
99
|
+
contig_sequences = []
|
100
|
+
contig_names = []
|
101
|
+
overly_short_sequence_count = 0
|
102
|
+
process_sequence = lambda do |name, seq|
|
103
|
+
if seq.length < 2*options[:contig_end_length]
|
104
|
+
log.warn "Not attempting to make connections from this contig, as it is overly short: #{name}"
|
105
|
+
overly_short_sequence_count += 1
|
106
|
+
nil
|
107
|
+
else
|
108
|
+
contig_sequences.push seq.to_s
|
109
|
+
contig_names.push name
|
110
|
+
|
111
|
+
sequence = seq.seq
|
112
|
+
fwd2 = Bio::Sequence::NA.new(sequence[0...options[:contig_end_length]])
|
113
|
+
probe_sequences.push fwd2.reverse_complement.to_s
|
114
|
+
|
115
|
+
probe_sequences.push sequence[(sequence.length-options[:contig_end_length])...sequence.length]
|
116
|
+
|
117
|
+
# 'return' the probe indices that have been assigned
|
118
|
+
[probe_sequences.length-2, probe_sequences.length-1]
|
119
|
+
end
|
120
|
+
end
|
121
|
+
|
122
|
+
scaffolds = nil #Array of Bio::FinishM::ScaffoldBreaker::Scaffold objects.
|
123
|
+
scaffolded_contig_to_probe_ids = {}
|
124
|
+
if options[:unscaffold_first]
|
125
|
+
log.info "Unscaffolding scaffolds (before trying to connect them together again)"
|
126
|
+
scaffolds = Bio::FinishM::ScaffoldBreaker.new.break_scaffolds options[:contigs_file]
|
127
|
+
scaffolds.each do |scaffold|
|
128
|
+
scaffold.contigs.each do |contig|
|
129
|
+
process_sequence.call contig.name, contig.sequence
|
130
|
+
end
|
131
|
+
end
|
132
|
+
else
|
133
|
+
# Else don't split up any of the sequences
|
134
|
+
log.info "Reading input sequences.."
|
135
|
+
Bio::FlatFile.foreach(options[:contigs_file]) do |seq|
|
136
|
+
process_sequence.call seq.definition, seq.seq
|
137
|
+
end
|
138
|
+
end
|
139
|
+
|
140
|
+
if overly_short_sequence_count > 0
|
141
|
+
unless options[:proceed_on_short_contigs]
|
142
|
+
raise "Not proceding as some contigs are too short (length < 2 * overhang). You might try: "+
|
143
|
+
"(1) omitting the smaller contigs, (2) reducing the --overhang parameter, or "+
|
144
|
+
"(3) using --proceed-on-short-contigs to continue optimistically ignoring the #{overly_short_sequence_count} short contigs"
|
145
|
+
end
|
146
|
+
end
|
147
|
+
|
148
|
+
log.info "Searching from #{probe_sequences.length} different contig ends (#{probe_sequences.length / 2} contigs)"
|
149
|
+
|
150
|
+
# Generate the graph with the probe sequences in it.
|
151
|
+
read_input = Bio::FinishM::ReadInput.new
|
152
|
+
read_input.parse_options options
|
153
|
+
finishm_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
154
|
+
|
155
|
+
log.info "Finding possible connections with recoherence kmer of #{options[:recoherence_kmer] }"
|
156
|
+
all_connections = probed_graph_to_connections(finishm_graph, options)
|
157
|
+
log.debug "Finished actual wandering, found #{all_connections.length} connections" if log.debug?
|
158
|
+
|
159
|
+
# Determine scaffolding connections
|
160
|
+
interpreter = Bio::FinishM::ConnectionInterpreter.new(all_connections, (0...contig_sequences.length))
|
161
|
+
connections = interpreter.doubly_single_contig_connections
|
162
|
+
log.debug "Found #{connections.length} connections between contigs that can be used for scaffolding" if log.debug?
|
163
|
+
scaffolds = interpreter.scaffolds(connections)
|
164
|
+
|
165
|
+
# Gather some stats
|
166
|
+
circular_scaffold_names = []
|
167
|
+
num_contigs_in_circular_scaffolds = 0
|
168
|
+
num_singleton_contigs = 0
|
169
|
+
num_scaffolded_contigs = 0
|
170
|
+
scaffolds.each do |scaffold|
|
171
|
+
if scaffold.circular?
|
172
|
+
circular_scaffold_names.push name
|
173
|
+
num_contigs_in_circular_scaffolds += scaffold.contigs.length
|
174
|
+
elsif scaffold.contigs.length == 1
|
175
|
+
num_singleton_contigs += 1
|
176
|
+
else
|
177
|
+
num_scaffolded_contigs += scaffold.contigs.length
|
178
|
+
end
|
179
|
+
end
|
180
|
+
log.info "Found #{circular_scaffold_names.length} circular scaffolds encompassing #{num_contigs_in_circular_scaffolds} contigs"
|
181
|
+
log.info "#{num_scaffolded_contigs} contigs were incorporated into scaffolds"
|
182
|
+
log.info "#{num_singleton_contigs} contigs were not incorporated into any scaffolds"
|
183
|
+
|
184
|
+
unless options[:output_scaffolds_file].nil?
|
185
|
+
File.open(options[:output_scaffolds_file],'w') do |scaffold_file|
|
186
|
+
scaffolds.each_with_index do |scaffold, i|
|
187
|
+
name = nil
|
188
|
+
if scaffold.contigs.length == 1
|
189
|
+
name = "scaffold#{i+1}"
|
190
|
+
else
|
191
|
+
name = "scaffold#{i+1}"
|
192
|
+
end
|
193
|
+
if scaffold.circular?
|
194
|
+
name += ' circular'
|
195
|
+
end
|
196
|
+
|
197
|
+
scaffold_file.puts ">#{name}"
|
198
|
+
# Output the NA sequence wrapped
|
199
|
+
seq = scaffold.sequence(contig_sequences)
|
200
|
+
scaffold_file.puts seq.gsub(/(.{80})/,"\\1\n").gsub(/\n$/,'')
|
201
|
+
end
|
202
|
+
end
|
203
|
+
end
|
204
|
+
|
205
|
+
# Write out all connections to the given file if wanted
|
206
|
+
unless options[:output_connection_file].nil?
|
207
|
+
File.open(options[:output_connection_file], 'w') do |out|
|
208
|
+
all_connections.each do |conn|
|
209
|
+
out.puts [
|
210
|
+
"#{contig_names[conn.probe1.sequence_index]}:#{conn.probe1.side}",
|
211
|
+
"#{contig_names[conn.probe2.sequence_index]}:#{conn.probe2.side}",
|
212
|
+
conn.distance
|
213
|
+
].join("\t")
|
214
|
+
end
|
215
|
+
end
|
216
|
+
end
|
217
|
+
|
218
|
+
log.info "All done."
|
219
|
+
end
|
220
|
+
|
221
|
+
# Given a probed graph, wander between all the nodes, and then return an
|
222
|
+
# instance of Bio::FinishM::ConnectionInterpreter::Scaffold. Required options:
|
223
|
+
# * :graph_search_leash_length
|
224
|
+
# * :recoherence_kmer
|
225
|
+
def probed_graph_to_connections(finishm_graph, options)
|
226
|
+
# Loop over the ends, trying to make connections from each one
|
227
|
+
cartographer = Bio::AssemblyGraphAlgorithms::SingleCoherentWanderer.new
|
228
|
+
|
229
|
+
first_connections = cartographer.wander(finishm_graph, options[:graph_search_leash_length], options[:recoherence_kmer], finishm_graph.velvet_sequences, options)
|
230
|
+
log.debug "Initially found #{first_connections.length} connections with less distance than the leash length" if log.debug?
|
231
|
+
|
232
|
+
probe_descriptions = []
|
233
|
+
(0...finishm_graph.probe_nodes.length).each do |i|
|
234
|
+
desc = Bio::FinishM::ConnectionInterpreter::Probe.new
|
235
|
+
if i % 2 == 0
|
236
|
+
desc.side = :start
|
237
|
+
desc.sequence_index = i / 2
|
238
|
+
else
|
239
|
+
desc.side = :end
|
240
|
+
desc.sequence_index = (i-1) / 2
|
241
|
+
end
|
242
|
+
probe_descriptions.push desc
|
243
|
+
end
|
244
|
+
|
245
|
+
# Gather connections ready for output
|
246
|
+
distance_calibrator = Bio::AssemblyGraphAlgorithms::AcyclicConnectionFinder.new
|
247
|
+
all_connections = []
|
248
|
+
first_connections.each do |node_indices, distance|
|
249
|
+
calibrated_distance = distance_calibrator.calibrate_distance_accounting_for_probes(
|
250
|
+
finishm_graph,
|
251
|
+
node_indices[0],
|
252
|
+
node_indices[1],
|
253
|
+
distance
|
254
|
+
)
|
255
|
+
|
256
|
+
# It is possible that a connection just larger than the leash length is returned.
|
257
|
+
# weed these out.
|
258
|
+
conn = Bio::FinishM::ConnectionInterpreter::Connection.new
|
259
|
+
conn.probe1 = probe_descriptions[node_indices[0]]
|
260
|
+
conn.probe2 = probe_descriptions[node_indices[1]]
|
261
|
+
conn.distance = calibrated_distance
|
262
|
+
if calibrated_distance > options[:graph_search_leash_length]
|
263
|
+
log.debug "Disregarding connection #{conn} because it was ultimately outside the allowable leash length" if log.debug?
|
264
|
+
else
|
265
|
+
all_connections.push conn
|
266
|
+
end
|
267
|
+
end
|
268
|
+
return all_connections
|
269
|
+
end
|
270
|
+
end
|
@@ -0,0 +1,79 @@
|
|
1
|
+
# A pattern of presence/absence/neither across a run of kmers
|
2
|
+
class KmerAbundancePattern < Array
|
3
|
+
def binary_string
|
4
|
+
to_return = ''
|
5
|
+
each do |present|
|
6
|
+
to_return += case present
|
7
|
+
when true
|
8
|
+
'1'
|
9
|
+
when false
|
10
|
+
'0'
|
11
|
+
when '-'
|
12
|
+
'-'
|
13
|
+
else
|
14
|
+
raise "Unexpected pattern atom found: #{present}"
|
15
|
+
end
|
16
|
+
end
|
17
|
+
to_return
|
18
|
+
end
|
19
|
+
|
20
|
+
# Parse a 100001011 type representation
|
21
|
+
def parse_from_human(boolean_pattern)
|
22
|
+
self[0...length] = [] #remove the last pattern if it existed
|
23
|
+
boolean_pattern.each_char do |char|
|
24
|
+
if char == '1'
|
25
|
+
push true
|
26
|
+
elsif char == '0'
|
27
|
+
push false
|
28
|
+
elsif char == '-'
|
29
|
+
push nil
|
30
|
+
else
|
31
|
+
raise "Unexpected pattern character: #{char}"
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
35
|
+
|
36
|
+
# Return true if this pattern is exactly the same
|
37
|
+
# as another pattern
|
38
|
+
#
|
39
|
+
# e.g. 101 is same_as? 101 but not 111 or 110
|
40
|
+
def same_as?(another_pattern)
|
41
|
+
unless length == another_pattern.length
|
42
|
+
raise "Unexpected comparison of this pattern #{inspect} with another: #{another_pattern.inspect}"
|
43
|
+
end
|
44
|
+
each_with_index do |bool, i|
|
45
|
+
return false if bool != another_pattern[i]
|
46
|
+
end
|
47
|
+
return true
|
48
|
+
end
|
49
|
+
|
50
|
+
# Return true if another_pattern shows presence in all places
|
51
|
+
# where this pattern is present, (but maybe more)
|
52
|
+
#
|
53
|
+
# e.g. 101 is consisten with 101 and 111, but not 011
|
54
|
+
#
|
55
|
+
# Behaviour not defined when the first (this) pattern
|
56
|
+
# includes no-man's land components
|
57
|
+
def consistent_with?(another_pattern)
|
58
|
+
unless length == another_pattern.length
|
59
|
+
raise "Unexpected comparison of this pattern #{inspect} with another: #{another_pattern.inspect}"
|
60
|
+
end
|
61
|
+
each_with_index do |bool, i|
|
62
|
+
raise unless [true, false].include?(bool)
|
63
|
+
return false if bool and another_pattern[i] == false
|
64
|
+
end
|
65
|
+
return true
|
66
|
+
end
|
67
|
+
|
68
|
+
def parse_from_kmer_abundance(abundances, lower_limit, upper_limit)
|
69
|
+
abundances.each do |a|
|
70
|
+
if a>=upper_limit
|
71
|
+
push true
|
72
|
+
elsif a<=lower_limit
|
73
|
+
push false
|
74
|
+
else
|
75
|
+
push '-'
|
76
|
+
end
|
77
|
+
end
|
78
|
+
end
|
79
|
+
end
|
@@ -0,0 +1,48 @@
|
|
1
|
+
require 'csv'
|
2
|
+
|
3
|
+
module Bio
|
4
|
+
# A class to work with a kmer abundance file format,
|
5
|
+
# where the kmer is first, then each abundance after that (no headings, space separated)
|
6
|
+
class KmerMultipleAbundanceHash < Hash
|
7
|
+
include Bio::FinishM::Logging
|
8
|
+
|
9
|
+
def self.parse_from_file(path)
|
10
|
+
obj = self.new
|
11
|
+
kmer_length = nil
|
12
|
+
num_abundances = nil
|
13
|
+
CSV.foreach(path, :col_sep => ' ') do |row|
|
14
|
+
kmer = row[0].upcase
|
15
|
+
abundances = row[1...row.length]
|
16
|
+
|
17
|
+
kmer_length ||= kmer.length
|
18
|
+
if kmer.length != kmer_length
|
19
|
+
raise "inconsistent length of kmer found in kmer abundance file, in line: #{row.inspect}"
|
20
|
+
end
|
21
|
+
num_abundances ||= abundances.length
|
22
|
+
if num_abundances != abundances.length
|
23
|
+
raise "inconsistent number of abundances found in kmer abundance file, in line: #{row.inspect}"
|
24
|
+
end
|
25
|
+
obj[kmer] = abundances
|
26
|
+
end
|
27
|
+
return obj
|
28
|
+
end
|
29
|
+
|
30
|
+
def kmer_length
|
31
|
+
each do |kmer, abundances|
|
32
|
+
return kmer.length
|
33
|
+
end
|
34
|
+
end
|
35
|
+
|
36
|
+
def number_of_abundances
|
37
|
+
each do |kmer, abundances|
|
38
|
+
return abundances.length
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
def [](kmer)
|
43
|
+
abundances = super(kmer.upcase)
|
44
|
+
abundances ||= [0]*number_of_abundances
|
45
|
+
return abundances
|
46
|
+
end
|
47
|
+
end
|
48
|
+
end
|
@@ -0,0 +1,88 @@
|
|
1
|
+
class OligoDesigner
|
2
|
+
# Given a sequence, find the subsequence that starts at the 5' end (ie the
|
3
|
+
# start of the string), and ends when the melting temperature is maximal but
|
4
|
+
# below the max_temperature requires the oligotm program to be available on the
|
5
|
+
# cmd line.
|
6
|
+
#
|
7
|
+
# * nucleotide_string: the full sequence that we are choosing oligos from
|
8
|
+
# * max_temperature: the maximal temperature to start things off at
|
9
|
+
# * gc_clamp: require this many G or C residues at the 3' end of the oligo.
|
10
|
+
def just_below(nucleotide_string, max_temperature, gc_clamp=0)
|
11
|
+
# initial conditions
|
12
|
+
guess = 0
|
13
|
+
guess_temp = 0
|
14
|
+
|
15
|
+
# loop around
|
16
|
+
while guess_temp < max_temperature
|
17
|
+
guess += 1
|
18
|
+
guess_temp = melting_temperature nucleotide_string[0..guess-1]
|
19
|
+
|
20
|
+
# break if there's we've reached the end of the line
|
21
|
+
return nucleotide_string if guess > nucleotide_string.length
|
22
|
+
end
|
23
|
+
return nucleotide_string[0..guess-2]
|
24
|
+
end
|
25
|
+
|
26
|
+
# Rank oligomers within some constraints.
|
27
|
+
def possible_oligos_ordered_by_temperature_difference(nucleotide_string, min_temperature, best_temperature, max_temperature, gc_clamp)
|
28
|
+
default_distance = lambda do |seq, tm|
|
29
|
+
# fails constraints if not enough GC clamp
|
30
|
+
if seq[seq.length-gc_clamp..seq.length-1].gsub(/[gc]/i,'').length > 0
|
31
|
+
false
|
32
|
+
else
|
33
|
+
# the sequence is within contraints. The melting temperature closest to the best wins.
|
34
|
+
tm_diff = (best_temperature-tm).abs
|
35
|
+
tm_diff
|
36
|
+
end
|
37
|
+
end
|
38
|
+
|
39
|
+
# initial conditions
|
40
|
+
guess = 0
|
41
|
+
guess_temp = 0
|
42
|
+
# arrays to fill with possible possibles
|
43
|
+
oligos = []
|
44
|
+
|
45
|
+
# loop around, until max temperature is reached
|
46
|
+
while guess_temp < max_temperature
|
47
|
+
guess += 1
|
48
|
+
seq = nucleotide_string[0..guess-1]
|
49
|
+
guess_temp = melting_temperature seq
|
50
|
+
|
51
|
+
# Add it to the list if there is enough temperature
|
52
|
+
if guess_temp > min_temperature and guess_temp < max_temperature
|
53
|
+
o = Oligo.new
|
54
|
+
o.sequence = seq
|
55
|
+
o.tm = guess_temp
|
56
|
+
oligos.push o
|
57
|
+
end
|
58
|
+
|
59
|
+
# break if there's we've reached the end of the line
|
60
|
+
break if guess > nucleotide_string.length-1
|
61
|
+
end
|
62
|
+
|
63
|
+
# Convert sequences into distances
|
64
|
+
oligos.each do |oligo|
|
65
|
+
oligo.distance = default_distance.call(oligo.sequence, oligo.tm)
|
66
|
+
end
|
67
|
+
|
68
|
+
# remove sequences that don't meet the constraints, and sort the rest with
|
69
|
+
# smallest distance first
|
70
|
+
return oligos.reject{|o| o.distance == false}.sort{|a,b|
|
71
|
+
a.distance<=>b.distance
|
72
|
+
}.collect{|o| o.sequence}
|
73
|
+
end
|
74
|
+
alias_method :order, :possible_oligos_ordered_by_temperature_difference
|
75
|
+
|
76
|
+
# A simple method to return the melting temperature of a particular nucleotide
|
77
|
+
# string. Uses oligotm on the command line.
|
78
|
+
def melting_temperature(nucleotide_string)
|
79
|
+
#`oligotm -tp 1 -sc 1 -n 0.8 -d 500 -mv 0 -dv 50 '#{nucleotide_string}'`.to_f
|
80
|
+
`oligotm -tp 1 -sc 1 -n 0.2 -d 2 -mv 1 '#{nucleotide_string}'`.to_f
|
81
|
+
end
|
82
|
+
|
83
|
+
private
|
84
|
+
|
85
|
+
class Oligo
|
86
|
+
attr_accessor :sequence, :tm, :distance
|
87
|
+
end
|
88
|
+
end
|