wwood-bioruby 1.2.11
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +205 -0
- data/README_DEV.rdoc +285 -0
- data/VERSION.yml +4 -0
- data/bin/bioruby +44 -0
- data/bin/br_biofetch.rb +47 -0
- data/bin/br_bioflat.rb +293 -0
- data/bin/br_biogetseq.rb +45 -0
- data/bin/br_pmfetch.rb +421 -0
- data/lib/bio.rb +306 -0
- data/lib/bio/alignment.rb +2518 -0
- data/lib/bio/appl/bl2seq/report.rb +334 -0
- data/lib/bio/appl/blast.rb +505 -0
- data/lib/bio/appl/blast/ddbj.rb +142 -0
- data/lib/bio/appl/blast/format0.rb +1438 -0
- data/lib/bio/appl/blast/format8.rb +83 -0
- data/lib/bio/appl/blast/genomenet.rb +263 -0
- data/lib/bio/appl/blast/ncbioptions.rb +220 -0
- data/lib/bio/appl/blast/remote.rb +105 -0
- data/lib/bio/appl/blast/report.rb +767 -0
- data/lib/bio/appl/blast/rexml.rb +144 -0
- data/lib/bio/appl/blast/rpsblast.rb +277 -0
- data/lib/bio/appl/blast/wublast.rb +635 -0
- data/lib/bio/appl/blast/xmlparser.rb +236 -0
- data/lib/bio/appl/blat/report.rb +530 -0
- data/lib/bio/appl/clustalw.rb +219 -0
- data/lib/bio/appl/clustalw/report.rb +152 -0
- data/lib/bio/appl/emboss.rb +203 -0
- data/lib/bio/appl/fasta.rb +235 -0
- data/lib/bio/appl/fasta/format10.rb +325 -0
- data/lib/bio/appl/gcg/msf.rb +212 -0
- data/lib/bio/appl/gcg/seq.rb +195 -0
- data/lib/bio/appl/genscan/report.rb +552 -0
- data/lib/bio/appl/hmmer.rb +126 -0
- data/lib/bio/appl/hmmer/report.rb +683 -0
- data/lib/bio/appl/iprscan/report.rb +374 -0
- data/lib/bio/appl/mafft.rb +259 -0
- data/lib/bio/appl/mafft/report.rb +226 -0
- data/lib/bio/appl/muscle.rb +52 -0
- data/lib/bio/appl/paml/baseml.rb +95 -0
- data/lib/bio/appl/paml/baseml/report.rb +32 -0
- data/lib/bio/appl/paml/codeml.rb +242 -0
- data/lib/bio/appl/paml/codeml/rates.rb +67 -0
- data/lib/bio/appl/paml/codeml/report.rb +67 -0
- data/lib/bio/appl/paml/common.rb +348 -0
- data/lib/bio/appl/paml/common_report.rb +38 -0
- data/lib/bio/appl/paml/yn00.rb +103 -0
- data/lib/bio/appl/paml/yn00/report.rb +32 -0
- data/lib/bio/appl/phylip/alignment.rb +133 -0
- data/lib/bio/appl/phylip/distance_matrix.rb +96 -0
- data/lib/bio/appl/probcons.rb +41 -0
- data/lib/bio/appl/psort.rb +548 -0
- data/lib/bio/appl/psort/report.rb +542 -0
- data/lib/bio/appl/pts1.rb +263 -0
- data/lib/bio/appl/sim4.rb +124 -0
- data/lib/bio/appl/sim4/report.rb +485 -0
- data/lib/bio/appl/sosui/report.rb +151 -0
- data/lib/bio/appl/spidey/report.rb +593 -0
- data/lib/bio/appl/targetp/report.rb +267 -0
- data/lib/bio/appl/tcoffee.rb +55 -0
- data/lib/bio/appl/tmhmm/report.rb +231 -0
- data/lib/bio/command.rb +593 -0
- data/lib/bio/compat/features.rb +157 -0
- data/lib/bio/compat/references.rb +128 -0
- data/lib/bio/data/aa.rb +353 -0
- data/lib/bio/data/codontable.rb +722 -0
- data/lib/bio/data/na.rb +223 -0
- data/lib/bio/db.rb +329 -0
- data/lib/bio/db/aaindex.rb +357 -0
- data/lib/bio/db/biosql/biosql_to_biosequence.rb +67 -0
- data/lib/bio/db/biosql/sequence.rb +508 -0
- data/lib/bio/db/embl/common.rb +352 -0
- data/lib/bio/db/embl/embl.rb +500 -0
- data/lib/bio/db/embl/embl_to_biosequence.rb +85 -0
- data/lib/bio/db/embl/format_embl.rb +190 -0
- data/lib/bio/db/embl/sptr.rb +1283 -0
- data/lib/bio/db/embl/swissprot.rb +42 -0
- data/lib/bio/db/embl/trembl.rb +41 -0
- data/lib/bio/db/embl/uniprot.rb +42 -0
- data/lib/bio/db/fantom.rb +597 -0
- data/lib/bio/db/fasta.rb +410 -0
- data/lib/bio/db/fasta/defline.rb +532 -0
- data/lib/bio/db/fasta/fasta_to_biosequence.rb +63 -0
- data/lib/bio/db/fasta/format_fasta.rb +97 -0
- data/lib/bio/db/genbank/common.rb +307 -0
- data/lib/bio/db/genbank/ddbj.rb +22 -0
- data/lib/bio/db/genbank/format_genbank.rb +187 -0
- data/lib/bio/db/genbank/genbank.rb +250 -0
- data/lib/bio/db/genbank/genbank_to_biosequence.rb +86 -0
- data/lib/bio/db/genbank/genpept.rb +60 -0
- data/lib/bio/db/genbank/refseq.rb +18 -0
- data/lib/bio/db/gff.rb +1846 -0
- data/lib/bio/db/go.rb +481 -0
- data/lib/bio/db/kegg/brite.rb +41 -0
- data/lib/bio/db/kegg/compound.rb +131 -0
- data/lib/bio/db/kegg/drug.rb +98 -0
- data/lib/bio/db/kegg/enzyme.rb +148 -0
- data/lib/bio/db/kegg/expression.rb +155 -0
- data/lib/bio/db/kegg/genes.rb +263 -0
- data/lib/bio/db/kegg/genome.rb +241 -0
- data/lib/bio/db/kegg/glycan.rb +166 -0
- data/lib/bio/db/kegg/keggtab.rb +357 -0
- data/lib/bio/db/kegg/kgml.rb +256 -0
- data/lib/bio/db/kegg/orthology.rb +136 -0
- data/lib/bio/db/kegg/reaction.rb +82 -0
- data/lib/bio/db/kegg/taxonomy.rb +331 -0
- data/lib/bio/db/lasergene.rb +209 -0
- data/lib/bio/db/litdb.rb +107 -0
- data/lib/bio/db/medline.rb +326 -0
- data/lib/bio/db/nbrf.rb +191 -0
- data/lib/bio/db/newick.rb +658 -0
- data/lib/bio/db/nexus.rb +1854 -0
- data/lib/bio/db/pdb.rb +29 -0
- data/lib/bio/db/pdb/atom.rb +77 -0
- data/lib/bio/db/pdb/chain.rb +210 -0
- data/lib/bio/db/pdb/chemicalcomponent.rb +224 -0
- data/lib/bio/db/pdb/model.rb +148 -0
- data/lib/bio/db/pdb/pdb.rb +1911 -0
- data/lib/bio/db/pdb/residue.rb +176 -0
- data/lib/bio/db/pdb/utils.rb +399 -0
- data/lib/bio/db/prosite.rb +597 -0
- data/lib/bio/db/rebase.rb +456 -0
- data/lib/bio/db/soft.rb +404 -0
- data/lib/bio/db/transfac.rb +375 -0
- data/lib/bio/db/url.rb +42 -0
- data/lib/bio/feature.rb +139 -0
- data/lib/bio/io/biosql/biodatabase.rb +64 -0
- data/lib/bio/io/biosql/bioentry.rb +29 -0
- data/lib/bio/io/biosql/bioentry_dbxref.rb +11 -0
- data/lib/bio/io/biosql/bioentry_path.rb +12 -0
- data/lib/bio/io/biosql/bioentry_qualifier_value.rb +10 -0
- data/lib/bio/io/biosql/bioentry_reference.rb +10 -0
- data/lib/bio/io/biosql/bioentry_relationship.rb +10 -0
- data/lib/bio/io/biosql/biosequence.rb +11 -0
- data/lib/bio/io/biosql/comment.rb +7 -0
- data/lib/bio/io/biosql/config/database.yml +20 -0
- data/lib/bio/io/biosql/dbxref.rb +13 -0
- data/lib/bio/io/biosql/dbxref_qualifier_value.rb +12 -0
- data/lib/bio/io/biosql/location.rb +32 -0
- data/lib/bio/io/biosql/location_qualifier_value.rb +11 -0
- data/lib/bio/io/biosql/ontology.rb +10 -0
- data/lib/bio/io/biosql/reference.rb +9 -0
- data/lib/bio/io/biosql/seqfeature.rb +32 -0
- data/lib/bio/io/biosql/seqfeature_dbxref.rb +11 -0
- data/lib/bio/io/biosql/seqfeature_path.rb +11 -0
- data/lib/bio/io/biosql/seqfeature_qualifier_value.rb +20 -0
- data/lib/bio/io/biosql/seqfeature_relationship.rb +11 -0
- data/lib/bio/io/biosql/taxon.rb +12 -0
- data/lib/bio/io/biosql/taxon_name.rb +9 -0
- data/lib/bio/io/biosql/term.rb +27 -0
- data/lib/bio/io/biosql/term_dbxref.rb +11 -0
- data/lib/bio/io/biosql/term_path.rb +12 -0
- data/lib/bio/io/biosql/term_relationship.rb +13 -0
- data/lib/bio/io/biosql/term_relationship_term.rb +11 -0
- data/lib/bio/io/biosql/term_synonym.rb +10 -0
- data/lib/bio/io/das.rb +461 -0
- data/lib/bio/io/dbget.rb +194 -0
- data/lib/bio/io/ddbjxml.rb +638 -0
- data/lib/bio/io/ebisoap.rb +158 -0
- data/lib/bio/io/ensembl.rb +229 -0
- data/lib/bio/io/fastacmd.rb +163 -0
- data/lib/bio/io/fetch.rb +195 -0
- data/lib/bio/io/flatfile.rb +482 -0
- data/lib/bio/io/flatfile/autodetection.rb +545 -0
- data/lib/bio/io/flatfile/bdb.rb +253 -0
- data/lib/bio/io/flatfile/buffer.rb +237 -0
- data/lib/bio/io/flatfile/index.rb +1381 -0
- data/lib/bio/io/flatfile/indexer.rb +805 -0
- data/lib/bio/io/flatfile/splitter.rb +297 -0
- data/lib/bio/io/higet.rb +73 -0
- data/lib/bio/io/hinv.rb +442 -0
- data/lib/bio/io/keggapi.rb +805 -0
- data/lib/bio/io/ncbirest.rb +733 -0
- data/lib/bio/io/ncbisoap.rb +155 -0
- data/lib/bio/io/pubmed.rb +307 -0
- data/lib/bio/io/registry.rb +292 -0
- data/lib/bio/io/soapwsdl.rb +119 -0
- data/lib/bio/io/sql.rb +186 -0
- data/lib/bio/location.rb +867 -0
- data/lib/bio/map.rb +410 -0
- data/lib/bio/pathway.rb +960 -0
- data/lib/bio/reference.rb +602 -0
- data/lib/bio/sequence.rb +456 -0
- data/lib/bio/sequence/aa.rb +152 -0
- data/lib/bio/sequence/adapter.rb +108 -0
- data/lib/bio/sequence/common.rb +310 -0
- data/lib/bio/sequence/compat.rb +123 -0
- data/lib/bio/sequence/dblink.rb +54 -0
- data/lib/bio/sequence/format.rb +358 -0
- data/lib/bio/sequence/format_raw.rb +23 -0
- data/lib/bio/sequence/generic.rb +24 -0
- data/lib/bio/sequence/na.rb +491 -0
- data/lib/bio/shell.rb +44 -0
- data/lib/bio/shell/core.rb +578 -0
- data/lib/bio/shell/demo.rb +146 -0
- data/lib/bio/shell/interface.rb +218 -0
- data/lib/bio/shell/irb.rb +95 -0
- data/lib/bio/shell/object.rb +71 -0
- data/lib/bio/shell/plugin/blast.rb +42 -0
- data/lib/bio/shell/plugin/codon.rb +218 -0
- data/lib/bio/shell/plugin/das.rb +58 -0
- data/lib/bio/shell/plugin/emboss.rb +23 -0
- data/lib/bio/shell/plugin/entry.rb +105 -0
- data/lib/bio/shell/plugin/flatfile.rb +101 -0
- data/lib/bio/shell/plugin/keggapi.rb +181 -0
- data/lib/bio/shell/plugin/midi.rb +430 -0
- data/lib/bio/shell/plugin/obda.rb +45 -0
- data/lib/bio/shell/plugin/psort.rb +56 -0
- data/lib/bio/shell/plugin/seq.rb +247 -0
- data/lib/bio/shell/plugin/soap.rb +87 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/bioruby_generator.rb +29 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_classes.rhtml +4 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_log.rhtml +27 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_methods.rhtml +11 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_modules.rhtml +4 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_variables.rhtml +7 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-bg.gif +0 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-gem.png +0 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-link.gif +0 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.css +368 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.rhtml +47 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_controller.rb +144 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_helper.rb +47 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/commands.rhtml +8 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/history.rhtml +10 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/index.rhtml +26 -0
- data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/spinner.gif +0 -0
- data/lib/bio/shell/script.rb +25 -0
- data/lib/bio/shell/setup.rb +109 -0
- data/lib/bio/shell/web.rb +102 -0
- data/lib/bio/tree.rb +852 -0
- data/lib/bio/util/color_scheme.rb +191 -0
- data/lib/bio/util/color_scheme/buried.rb +59 -0
- data/lib/bio/util/color_scheme/helix.rb +59 -0
- data/lib/bio/util/color_scheme/hydropathy.rb +64 -0
- data/lib/bio/util/color_scheme/nucleotide.rb +31 -0
- data/lib/bio/util/color_scheme/strand.rb +59 -0
- data/lib/bio/util/color_scheme/taylor.rb +50 -0
- data/lib/bio/util/color_scheme/turn.rb +59 -0
- data/lib/bio/util/color_scheme/zappo.rb +50 -0
- data/lib/bio/util/contingency_table.rb +370 -0
- data/lib/bio/util/restriction_enzyme.rb +228 -0
- data/lib/bio/util/restriction_enzyme/analysis.rb +249 -0
- data/lib/bio/util/restriction_enzyme/analysis_basic.rb +217 -0
- data/lib/bio/util/restriction_enzyme/cut_symbol.rb +107 -0
- data/lib/bio/util/restriction_enzyme/double_stranded.rb +321 -0
- data/lib/bio/util/restriction_enzyme/double_stranded/aligned_strands.rb +130 -0
- data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair.rb +103 -0
- data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair_in_enzyme_notation.rb +38 -0
- data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations.rb +76 -0
- data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations_in_enzyme_notation.rb +107 -0
- data/lib/bio/util/restriction_enzyme/enzymes.yaml +7061 -0
- data/lib/bio/util/restriction_enzyme/range/cut_range.rb +24 -0
- data/lib/bio/util/restriction_enzyme/range/cut_ranges.rb +47 -0
- data/lib/bio/util/restriction_enzyme/range/horizontal_cut_range.rb +67 -0
- data/lib/bio/util/restriction_enzyme/range/sequence_range.rb +257 -0
- data/lib/bio/util/restriction_enzyme/range/sequence_range/calculated_cuts.rb +242 -0
- data/lib/bio/util/restriction_enzyme/range/sequence_range/fragment.rb +51 -0
- data/lib/bio/util/restriction_enzyme/range/sequence_range/fragments.rb +41 -0
- data/lib/bio/util/restriction_enzyme/range/vertical_cut_range.rb +77 -0
- data/lib/bio/util/restriction_enzyme/single_strand.rb +200 -0
- data/lib/bio/util/restriction_enzyme/single_strand/cut_locations_in_enzyme_notation.rb +135 -0
- data/lib/bio/util/restriction_enzyme/single_strand_complement.rb +23 -0
- data/lib/bio/util/restriction_enzyme/string_formatting.rb +111 -0
- data/lib/bio/util/sirna.rb +288 -0
- data/test/data/HMMER/hmmpfam.out +64 -0
- data/test/data/HMMER/hmmsearch.out +88 -0
- data/test/data/SOSUI/sample.report +11 -0
- data/test/data/TMHMM/sample.report +21 -0
- data/test/data/aaindex/DAYM780301 +30 -0
- data/test/data/aaindex/PRAM900102 +20 -0
- data/test/data/bl2seq/cd8a_cd8b_blastp.bl2seq +53 -0
- data/test/data/bl2seq/cd8a_p53_e-5blastp.bl2seq +37 -0
- data/test/data/blast/2.2.15.blastp.m7 +876 -0
- data/test/data/blast/b0002.faa +15 -0
- data/test/data/blast/b0002.faa.m0 +128 -0
- data/test/data/blast/b0002.faa.m7 +65 -0
- data/test/data/blast/b0002.faa.m8 +1 -0
- data/test/data/blast/blastp-multi.m7 +188 -0
- data/test/data/command/echoarg2.bat +1 -0
- data/test/data/embl/AB090716.embl +65 -0
- data/test/data/embl/AB090716.embl.rel89 +63 -0
- data/test/data/fasta/example1.txt +75 -0
- data/test/data/fasta/example2.txt +21 -0
- data/test/data/genscan/sample.report +63 -0
- data/test/data/iprscan/merged.raw +32 -0
- data/test/data/iprscan/merged.txt +74 -0
- data/test/data/paml/codeml/control_file.txt +30 -0
- data/test/data/paml/codeml/output.txt +78 -0
- data/test/data/paml/codeml/rates +217 -0
- data/test/data/prosite/prosite.dat +2233 -0
- data/test/data/refseq/nm_126355.entret +64 -0
- data/test/data/rpsblast/misc.rpsblast +193 -0
- data/test/data/soft/GDS100_partial.soft +92 -0
- data/test/data/soft/GSE3457_family_partial.soft +874 -0
- data/test/data/uniprot/p53_human.uniprot +1456 -0
- data/test/functional/bio/appl/test_pts1.rb +115 -0
- data/test/functional/bio/io/test_ensembl.rb +229 -0
- data/test/functional/bio/io/test_soapwsdl.rb +52 -0
- data/test/functional/bio/sequence/test_output_embl.rb +51 -0
- data/test/functional/bio/test_command.rb +301 -0
- data/test/runner.rb +14 -0
- data/test/unit/bio/appl/bl2seq/test_report.rb +134 -0
- data/test/unit/bio/appl/blast/test_ncbioptions.rb +112 -0
- data/test/unit/bio/appl/blast/test_report.rb +1135 -0
- data/test/unit/bio/appl/blast/test_rpsblast.rb +398 -0
- data/test/unit/bio/appl/genscan/test_report.rb +182 -0
- data/test/unit/bio/appl/hmmer/test_report.rb +342 -0
- data/test/unit/bio/appl/iprscan/test_report.rb +338 -0
- data/test/unit/bio/appl/mafft/test_report.rb +63 -0
- data/test/unit/bio/appl/paml/codeml/test_rates.rb +45 -0
- data/test/unit/bio/appl/paml/codeml/test_report.rb +45 -0
- data/test/unit/bio/appl/paml/test_codeml.rb +174 -0
- data/test/unit/bio/appl/sosui/test_report.rb +81 -0
- data/test/unit/bio/appl/targetp/test_report.rb +146 -0
- data/test/unit/bio/appl/test_blast.rb +277 -0
- data/test/unit/bio/appl/test_fasta.rb +130 -0
- data/test/unit/bio/appl/test_psort.rb +57 -0
- data/test/unit/bio/appl/test_pts1.rb +77 -0
- data/test/unit/bio/appl/tmhmm/test_report.rb +126 -0
- data/test/unit/bio/data/test_aa.rb +90 -0
- data/test/unit/bio/data/test_codontable.rb +107 -0
- data/test/unit/bio/data/test_na.rb +80 -0
- data/test/unit/bio/db/embl/test_common.rb +117 -0
- data/test/unit/bio/db/embl/test_embl.rb +214 -0
- data/test/unit/bio/db/embl/test_embl_rel89.rb +219 -0
- data/test/unit/bio/db/embl/test_embl_to_bioseq.rb +203 -0
- data/test/unit/bio/db/embl/test_sptr.rb +1812 -0
- data/test/unit/bio/db/embl/test_uniprot.rb +31 -0
- data/test/unit/bio/db/kegg/test_genes.rb +45 -0
- data/test/unit/bio/db/pdb/test_pdb.rb +152 -0
- data/test/unit/bio/db/test_aaindex.rb +197 -0
- data/test/unit/bio/db/test_fasta.rb +250 -0
- data/test/unit/bio/db/test_gff.rb +1190 -0
- data/test/unit/bio/db/test_lasergene.rb +95 -0
- data/test/unit/bio/db/test_medline.rb +127 -0
- data/test/unit/bio/db/test_newick.rb +293 -0
- data/test/unit/bio/db/test_nexus.rb +364 -0
- data/test/unit/bio/db/test_prosite.rb +1437 -0
- data/test/unit/bio/db/test_rebase.rb +101 -0
- data/test/unit/bio/db/test_soft.rb +138 -0
- data/test/unit/bio/db/test_url.rb +36 -0
- data/test/unit/bio/io/flatfile/test_autodetection.rb +375 -0
- data/test/unit/bio/io/flatfile/test_buffer.rb +251 -0
- data/test/unit/bio/io/flatfile/test_splitter.rb +369 -0
- data/test/unit/bio/io/test_ddbjxml.rb +80 -0
- data/test/unit/bio/io/test_ensembl.rb +109 -0
- data/test/unit/bio/io/test_fastacmd.rb +42 -0
- data/test/unit/bio/io/test_flatfile.rb +505 -0
- data/test/unit/bio/io/test_soapwsdl.rb +32 -0
- data/test/unit/bio/sequence/test_aa.rb +115 -0
- data/test/unit/bio/sequence/test_common.rb +373 -0
- data/test/unit/bio/sequence/test_compat.rb +69 -0
- data/test/unit/bio/sequence/test_dblink.rb +58 -0
- data/test/unit/bio/sequence/test_na.rb +330 -0
- data/test/unit/bio/shell/plugin/test_seq.rb +185 -0
- data/test/unit/bio/test_alignment.rb +1025 -0
- data/test/unit/bio/test_command.rb +349 -0
- data/test/unit/bio/test_db.rb +96 -0
- data/test/unit/bio/test_feature.rb +144 -0
- data/test/unit/bio/test_location.rb +599 -0
- data/test/unit/bio/test_map.rb +230 -0
- data/test/unit/bio/test_pathway.rb +499 -0
- data/test/unit/bio/test_reference.rb +252 -0
- data/test/unit/bio/test_sequence.rb +329 -0
- data/test/unit/bio/test_shell.rb +18 -0
- data/test/unit/bio/test_tree.rb +593 -0
- data/test/unit/bio/util/restriction_enzyme/analysis/test_calculated_cuts.rb +299 -0
- data/test/unit/bio/util/restriction_enzyme/analysis/test_cut_ranges.rb +103 -0
- data/test/unit/bio/util/restriction_enzyme/analysis/test_sequence_range.rb +240 -0
- data/test/unit/bio/util/restriction_enzyme/double_stranded/test_aligned_strands.rb +101 -0
- data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair.rb +75 -0
- data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair_in_enzyme_notation.rb +73 -0
- data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations.rb +53 -0
- data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations_in_enzyme_notation.rb +104 -0
- data/test/unit/bio/util/restriction_enzyme/single_strand/test_cut_locations_in_enzyme_notation.rb +83 -0
- data/test/unit/bio/util/restriction_enzyme/test_analysis.rb +246 -0
- data/test/unit/bio/util/restriction_enzyme/test_cut_symbol.rb +44 -0
- data/test/unit/bio/util/restriction_enzyme/test_double_stranded.rb +115 -0
- data/test/unit/bio/util/restriction_enzyme/test_single_strand.rb +147 -0
- data/test/unit/bio/util/restriction_enzyme/test_single_strand_complement.rb +147 -0
- data/test/unit/bio/util/restriction_enzyme/test_string_formatting.rb +60 -0
- data/test/unit/bio/util/test_color_scheme.rb +33 -0
- data/test/unit/bio/util/test_contingency_table.rb +94 -0
- data/test/unit/bio/util/test_restriction_enzyme.rb +42 -0
- data/test/unit/bio/util/test_sirna.rb +245 -0
- metadata +543 -0
@@ -0,0 +1,126 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/hmmer.rb - HMMER wrapper
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2002
|
5
|
+
# Toshiaki Katayama <k@bioruby.org>
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: hmmer.rb,v 1.9 2007/04/05 23:35:39 trevor Exp $
|
9
|
+
#
|
10
|
+
|
11
|
+
require 'bio/command'
|
12
|
+
require 'shellwords'
|
13
|
+
|
14
|
+
module Bio
|
15
|
+
|
16
|
+
# == Description
|
17
|
+
#
|
18
|
+
# A wapper for HMMER programs (hmmsearch or hmmpfam).
|
19
|
+
#
|
20
|
+
# === Examples
|
21
|
+
#
|
22
|
+
# require 'bio'
|
23
|
+
# program = 'hmmsearch' # or 'hmmpfam'
|
24
|
+
# hmmfile = 'test.hmm'
|
25
|
+
# seqfile = 'test.faa'
|
26
|
+
#
|
27
|
+
# factory = Bio::HMMER.new(program, hmmfile, seqfile)
|
28
|
+
# report = factory.query
|
29
|
+
# report.class # => Bio::HMMER::Report
|
30
|
+
#
|
31
|
+
# === References
|
32
|
+
#
|
33
|
+
# * HMMER
|
34
|
+
# http://hmmer.wustl.edu/
|
35
|
+
#
|
36
|
+
class HMMER
|
37
|
+
|
38
|
+
autoload :Report, 'bio/appl/hmmer/report'
|
39
|
+
|
40
|
+
# Prgrams name. (hmmsearch or hmmpfam).
|
41
|
+
attr_accessor :program
|
42
|
+
|
43
|
+
# Name of hmmfile.
|
44
|
+
attr_accessor :hmmfile
|
45
|
+
|
46
|
+
# Name of seqfile.
|
47
|
+
attr_accessor :seqfile
|
48
|
+
|
49
|
+
# Command line options.
|
50
|
+
attr_accessor :options
|
51
|
+
|
52
|
+
# Shows the raw output from the hmmer search.
|
53
|
+
attr_reader :output
|
54
|
+
|
55
|
+
# Sets a program name, a profile hmm file name, a query sequence file name
|
56
|
+
# and options in string.
|
57
|
+
#
|
58
|
+
# Program names: hmmsearch, hmmpfam
|
59
|
+
#
|
60
|
+
def initialize(program, hmmfile, seqfile, options = [])
|
61
|
+
@program = program
|
62
|
+
@hmmfile = hmmfile
|
63
|
+
@seqfile = seqfile
|
64
|
+
@output = ''
|
65
|
+
|
66
|
+
begin
|
67
|
+
@options = options.to_ary
|
68
|
+
rescue NameError #NoMethodError
|
69
|
+
# backward compatibility
|
70
|
+
@options = Shellwords.shellwords(options)
|
71
|
+
end
|
72
|
+
end
|
73
|
+
|
74
|
+
|
75
|
+
# Gets options by String.
|
76
|
+
# backward compatibility.
|
77
|
+
def option
|
78
|
+
Bio::Command.make_command_line(@options)
|
79
|
+
end
|
80
|
+
|
81
|
+
|
82
|
+
# Sets options by String.
|
83
|
+
# backward compatibility.
|
84
|
+
def option=(str)
|
85
|
+
@options = Shellwords.shellwords(str)
|
86
|
+
end
|
87
|
+
|
88
|
+
|
89
|
+
# Executes the hmmer search and returns the report
|
90
|
+
# (Bio::HMMER::Report object).
|
91
|
+
def query
|
92
|
+
cmd = [ @program, *@options ]
|
93
|
+
cmd.concat([ @hmmfile, @seqfile ])
|
94
|
+
|
95
|
+
report = nil
|
96
|
+
|
97
|
+
@output = Bio::Command.query_command(cmd, nil)
|
98
|
+
report = parse_result(@output)
|
99
|
+
|
100
|
+
return report
|
101
|
+
end
|
102
|
+
|
103
|
+
private
|
104
|
+
|
105
|
+
def parse_result(data)
|
106
|
+
Report.new(data)
|
107
|
+
end
|
108
|
+
|
109
|
+
end # class HMMER
|
110
|
+
|
111
|
+
end # module Bio
|
112
|
+
|
113
|
+
|
114
|
+
|
115
|
+
if __FILE__ == $0
|
116
|
+
|
117
|
+
require 'pp'
|
118
|
+
|
119
|
+
program = ARGV.shift # hmmsearch, hmmpfam
|
120
|
+
hmmfile = ARGV.shift
|
121
|
+
seqfile = ARGV.shift
|
122
|
+
|
123
|
+
factory = Bio::HMMER.new(program, hmmfile, seqfile)
|
124
|
+
pp factory.query
|
125
|
+
|
126
|
+
end
|
@@ -0,0 +1,683 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/hmmer/report.rb - hmmsearch, hmmpfam parserer
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2002
|
5
|
+
# Hiroshi Suga <suga@biophys.kyoto-u.ac.jp>,
|
6
|
+
# Copyright:: Copyright (C) 2005
|
7
|
+
# Masashi Fujita <fujita@kuicr.kyoto-u.ac.jp>
|
8
|
+
# License:: The Ruby License
|
9
|
+
#
|
10
|
+
# $Id:$
|
11
|
+
#
|
12
|
+
# == Description
|
13
|
+
#
|
14
|
+
# Parser class for hmmsearch and hmmpfam in the HMMER package.
|
15
|
+
#
|
16
|
+
# == Examples
|
17
|
+
#
|
18
|
+
# #for multiple reports in a single output file (example.hmmpfam)
|
19
|
+
# Bio::HMMER.reports(File.read("example.hmmpfam")) do |report|
|
20
|
+
# report.program['name']
|
21
|
+
# report.parameter['HMM file']
|
22
|
+
# report.query_info['Query sequence']
|
23
|
+
# report.hits.each do |hit|
|
24
|
+
# hit.accession
|
25
|
+
# hit.description
|
26
|
+
# hit.score
|
27
|
+
# hit.evalue
|
28
|
+
# hit.hsps.each do |hsp|
|
29
|
+
# hsp.accession
|
30
|
+
# hsp.domain
|
31
|
+
# hsp.evalue
|
32
|
+
# hsp.midline
|
33
|
+
# end
|
34
|
+
# end
|
35
|
+
#
|
36
|
+
# == References
|
37
|
+
#
|
38
|
+
# * HMMER
|
39
|
+
# http://hmmer.wustl.edu/
|
40
|
+
#
|
41
|
+
|
42
|
+
require 'bio/appl/hmmer'
|
43
|
+
|
44
|
+
module Bio
|
45
|
+
|
46
|
+
|
47
|
+
class HMMER
|
48
|
+
|
49
|
+
# A reader interface for multiple reports text into a report
|
50
|
+
# (Bio::HMMER::Report).
|
51
|
+
#
|
52
|
+
# === Examples
|
53
|
+
#
|
54
|
+
# # Iterator
|
55
|
+
# Bio::HMMER.reports(reports_text) do |report|
|
56
|
+
# report
|
57
|
+
# end
|
58
|
+
#
|
59
|
+
# # Array
|
60
|
+
# reports = Bio::HMMER.reports(reports_text)
|
61
|
+
#
|
62
|
+
def self.reports(multiple_report_text)
|
63
|
+
ary = []
|
64
|
+
multiple_report_text.each_line("\n//\n") do |report|
|
65
|
+
if block_given?
|
66
|
+
yield Report.new(report)
|
67
|
+
else
|
68
|
+
ary << Report.new(report)
|
69
|
+
end
|
70
|
+
end
|
71
|
+
return ary
|
72
|
+
end
|
73
|
+
|
74
|
+
|
75
|
+
# A parser class for a search report by hmmsearch or hmmpfam program in the
|
76
|
+
# HMMER package.
|
77
|
+
#
|
78
|
+
# === Examples
|
79
|
+
#
|
80
|
+
# Examples
|
81
|
+
# #for multiple reports in a single output file (example.hmmpfam)
|
82
|
+
# Bio::HMMER.reports(File.read("example.hmmpfam")) do |report|
|
83
|
+
# report.program['name']
|
84
|
+
# report.parameter['HMM file']
|
85
|
+
# report.query_info['Query sequence']
|
86
|
+
# report.hits.each do |hit|
|
87
|
+
# hit.accession
|
88
|
+
# hit.description
|
89
|
+
# hit.score
|
90
|
+
# hit.evalue
|
91
|
+
# hit.hsps.each do |hsp|
|
92
|
+
# hsp.accession
|
93
|
+
# hsp.domain
|
94
|
+
# hsp.evalue
|
95
|
+
# hsp.midline
|
96
|
+
# end
|
97
|
+
# end
|
98
|
+
#
|
99
|
+
# === References
|
100
|
+
#
|
101
|
+
# * HMMER
|
102
|
+
# http://hmmer.wustl.edu/
|
103
|
+
#
|
104
|
+
class Report
|
105
|
+
|
106
|
+
# Delimiter of each entry for Bio::FlatFile support.
|
107
|
+
DELIMITER = RS = "\n//\n"
|
108
|
+
|
109
|
+
|
110
|
+
# A Hash contains program information used.
|
111
|
+
# Valid keys are 'name', 'version', 'copyright' and 'license'.
|
112
|
+
attr_reader :program
|
113
|
+
|
114
|
+
# A hash contains parameters used.
|
115
|
+
# Valid keys are 'HMM file' and 'Sequence file'.
|
116
|
+
attr_reader :parameter
|
117
|
+
|
118
|
+
# A hash contains the query information.
|
119
|
+
# Valid keys are 'query sequence', 'Accession' and 'Description'.
|
120
|
+
attr_reader :query_info
|
121
|
+
|
122
|
+
#
|
123
|
+
attr_reader :hits
|
124
|
+
|
125
|
+
# Returns an Array of Bio::HMMER::Report::Hsp objects.
|
126
|
+
# Under special circumstances, some HSPs do not have
|
127
|
+
# parent Hit objects. If you want to access such HSPs,
|
128
|
+
# use this method.
|
129
|
+
attr_reader :hsps
|
130
|
+
|
131
|
+
# statistics by hmmsearch.
|
132
|
+
attr_reader :histogram
|
133
|
+
|
134
|
+
# statistics by hmmsearch. Keys are 'mu', 'lambda', 'chi-sq statistic' and 'P(chi-square)'.
|
135
|
+
attr_reader :statistical_detail
|
136
|
+
|
137
|
+
# statistics by hmmsearch.
|
138
|
+
attr_reader :total_seq_searched
|
139
|
+
|
140
|
+
# statistics by hmmsearch. Keys are 'Total memory', 'Satisfying E cutoff' and 'Total hits'.
|
141
|
+
attr_reader :whole_seq_top_hits
|
142
|
+
|
143
|
+
# statistics by hmmsearch. Keys are 'Total memory', 'Satisfying E cutoff' and 'Total hits'.
|
144
|
+
attr_reader :domain_top_hits
|
145
|
+
|
146
|
+
|
147
|
+
# Parses a HMMER search report (by hmmpfam or hmmsearch program) and
|
148
|
+
# reutrns a Bio::HMMER::Report object.
|
149
|
+
#
|
150
|
+
# === Examples
|
151
|
+
#
|
152
|
+
# hmmpfam_report = Bio::HMMER::Report.new(File.read("hmmpfam.out"))
|
153
|
+
#
|
154
|
+
# hmmsearch_report = Bio::HMMER::Report.new(File.read("hmmsearch.out"))
|
155
|
+
#
|
156
|
+
def initialize(data)
|
157
|
+
|
158
|
+
# The input data is divided into six data fields, i.e. header,
|
159
|
+
# query infomation, hits, HSPs, alignments and search statistics.
|
160
|
+
# However, header and statistics data don't necessarily exist.
|
161
|
+
subdata, is_hmmsearch = get_subdata(data)
|
162
|
+
|
163
|
+
# if header exists, parse it
|
164
|
+
if subdata["header"]
|
165
|
+
@program, @parameter = parse_header_data(subdata["header"])
|
166
|
+
else
|
167
|
+
@program, @parameter = [{}, {}]
|
168
|
+
end
|
169
|
+
|
170
|
+
@query_info = parse_query_info(subdata["query"])
|
171
|
+
@hits = parse_hit_data(subdata["hit"])
|
172
|
+
@hsps = parse_hsp_data(subdata["hsp"], is_hmmsearch)
|
173
|
+
|
174
|
+
if @hsps != []
|
175
|
+
# split alignment subdata into an array of alignments
|
176
|
+
aln_ary = subdata["alignment"].split(/^\S+.*?\n/).slice(1..-1)
|
177
|
+
|
178
|
+
# append alignment information to corresponding Hsp
|
179
|
+
aln_ary.each_with_index do |aln, i|
|
180
|
+
@hsps[i].set_alignment(aln)
|
181
|
+
end
|
182
|
+
end
|
183
|
+
|
184
|
+
# assign each Hsp object to its parent Hit
|
185
|
+
hits_hash = {}
|
186
|
+
@hits.each do |hit|
|
187
|
+
hits_hash[hit.accession] = hit
|
188
|
+
end
|
189
|
+
@hsps.each do |hsp|
|
190
|
+
if hits_hash.has_key?(hsp.accession)
|
191
|
+
hits_hash[hsp.accession].append_hsp(hsp)
|
192
|
+
end
|
193
|
+
end
|
194
|
+
|
195
|
+
# parse statistics (for hmmsearch)
|
196
|
+
if is_hmmsearch
|
197
|
+
@histogram, @statistical_detail, @total_seq_searched, \
|
198
|
+
@whole_seq_top_hits, @domain_top_hits = \
|
199
|
+
parse_stat_data(subdata["statistics"])
|
200
|
+
end
|
201
|
+
|
202
|
+
end
|
203
|
+
|
204
|
+
|
205
|
+
# Iterates each hit (Bio::HMMER::Report::Hit).
|
206
|
+
def each
|
207
|
+
@hits.each do |hit|
|
208
|
+
yield hit
|
209
|
+
end
|
210
|
+
end
|
211
|
+
alias :each_hit :each
|
212
|
+
|
213
|
+
|
214
|
+
# Bio::HMMER::Report#get_subdata
|
215
|
+
def get_subdata(data)
|
216
|
+
subdata = {}
|
217
|
+
header_prefix = '\Ahmm(search|pfam) - search'
|
218
|
+
query_prefix = '^Query (HMM|sequence): .*\nAccession: '
|
219
|
+
hit_prefix = '^Scores for (complete sequences|sequence family)'
|
220
|
+
hsp_prefix = '^Parsed for domains:'
|
221
|
+
aln_prefix = '^Alignments of top-scoring domains:\n'
|
222
|
+
stat_prefix = '^\nHistogram of all scores:'
|
223
|
+
|
224
|
+
# if header exists, get it
|
225
|
+
if data =~ /#{header_prefix}/
|
226
|
+
is_hmmsearch = ($1 == "search") # hmmsearch or hmmpfam
|
227
|
+
subdata["header"] = data[/(\A.+?)(?=#{query_prefix})/m]
|
228
|
+
else
|
229
|
+
is_hmmsearch = false # if no header, assumed to be hmmpfam
|
230
|
+
end
|
231
|
+
|
232
|
+
# get query, Hit and Hsp data
|
233
|
+
subdata["query"] = data[/(#{query_prefix}.+?)(?=#{hit_prefix})/m]
|
234
|
+
subdata["hit"] = data[/(#{hit_prefix}.+?)(?=#{hsp_prefix})/m]
|
235
|
+
subdata["hsp"] = data[/(#{hsp_prefix}.+?)(?=#{aln_prefix})/m]
|
236
|
+
|
237
|
+
# get alignment data
|
238
|
+
if is_hmmsearch
|
239
|
+
data =~ /#{aln_prefix}(.+?)#{stat_prefix}/m
|
240
|
+
subdata["alignment"] = $1
|
241
|
+
else
|
242
|
+
data =~ /#{aln_prefix}(.+?)\/\/\n/m
|
243
|
+
subdata["alignment"] = $1
|
244
|
+
raise "multiple reports found" if $'.length > 0
|
245
|
+
end
|
246
|
+
|
247
|
+
# handle -A option of HMMER
|
248
|
+
cutoff_line = '\t\[output cut off at A = \d+ top alignments\]\n\z'
|
249
|
+
subdata["alignment"].sub!(/#{cutoff_line}/, '')
|
250
|
+
|
251
|
+
# get statistics data
|
252
|
+
subdata["statistics"] = data[/(#{stat_prefix}.+)\z/m]
|
253
|
+
|
254
|
+
[subdata, is_hmmsearch]
|
255
|
+
end
|
256
|
+
private :get_subdata
|
257
|
+
|
258
|
+
|
259
|
+
# Bio::HMMER::Report#parse_header_data
|
260
|
+
def parse_header_data(data)
|
261
|
+
data =~ /\A(.+? - - -$\n)(.+? - - -$\n)\n\z/m
|
262
|
+
program_data = $1
|
263
|
+
parameter_data = $2
|
264
|
+
|
265
|
+
program = {}
|
266
|
+
program['name'], program['version'], program['copyright'], \
|
267
|
+
program['license'] = program_data.split(/\n/)
|
268
|
+
|
269
|
+
parameter = {}
|
270
|
+
parameter_data.each_line do |x|
|
271
|
+
if /^(.+?):\s+(.*?)\s*$/ =~ x
|
272
|
+
parameter[$1] = $2
|
273
|
+
end
|
274
|
+
end
|
275
|
+
|
276
|
+
[program, parameter]
|
277
|
+
end
|
278
|
+
private :parse_header_data
|
279
|
+
|
280
|
+
|
281
|
+
# Bio::HMMER::Report#parse_query_info
|
282
|
+
def parse_query_info(data)
|
283
|
+
hash = {}
|
284
|
+
data.each_line do |x|
|
285
|
+
if /^(.+?):\s+(.*?)\s*$/ =~ x
|
286
|
+
hash[$1] = $2
|
287
|
+
elsif /\s+\[(.+)\]/ =~ x
|
288
|
+
hash['comments'] = $1
|
289
|
+
end
|
290
|
+
end
|
291
|
+
hash
|
292
|
+
end
|
293
|
+
private :parse_query_info
|
294
|
+
|
295
|
+
|
296
|
+
# Bio::HMMER::Report#parse_hit_data
|
297
|
+
def parse_hit_data(data)
|
298
|
+
data.sub!(/.+?---\n/m, '').chop!
|
299
|
+
hits = []
|
300
|
+
return hits if data == "\t[no hits above thresholds]\n"
|
301
|
+
data.each_line do |l|
|
302
|
+
hits.push(Hit.new(l))
|
303
|
+
end
|
304
|
+
hits
|
305
|
+
end
|
306
|
+
private :parse_hit_data
|
307
|
+
|
308
|
+
|
309
|
+
# Bio::HMMER::Report#parse_hsp_data
|
310
|
+
def parse_hsp_data(data, is_hmmsearch)
|
311
|
+
data.sub!(/.+?---\n/m, '').chop!
|
312
|
+
hsps=[]
|
313
|
+
return hsps if data == "\t[no hits above thresholds]\n"
|
314
|
+
data.each_line do |l|
|
315
|
+
hsps.push(Hsp.new(l, is_hmmsearch))
|
316
|
+
end
|
317
|
+
return hsps
|
318
|
+
end
|
319
|
+
private :parse_hsp_data
|
320
|
+
|
321
|
+
|
322
|
+
# Bio::HMMER::Report#parse_stat_data
|
323
|
+
def parse_stat_data(data)
|
324
|
+
data.sub!(/\nHistogram of all scores:\n(.+?)\n\n\n%/m, '')
|
325
|
+
histogram = $1.strip
|
326
|
+
|
327
|
+
statistical_detail = {}
|
328
|
+
data.sub!(/(.+?)\n\n/m, '')
|
329
|
+
$1.each_line do |l|
|
330
|
+
statistical_detail[$1] = $2.to_f if /^\s*(.+?)\s*=\s*(\S+)/ =~ l
|
331
|
+
end
|
332
|
+
|
333
|
+
total_seq_searched = nil
|
334
|
+
data.sub!(/(.+?)\n\n/m, '')
|
335
|
+
$1.each_line do |l|
|
336
|
+
total_seq_searched = $2.to_i if /^\s*(.+)\s*:\s*(\S+)/ =~ l
|
337
|
+
end
|
338
|
+
|
339
|
+
whole_seq_top_hits = {}
|
340
|
+
data.sub!(/(.+?)\n\n/m, '')
|
341
|
+
$1.each_line do |l|
|
342
|
+
if /^\s*(.+?):\s*(\d+)\s*$/ =~ l
|
343
|
+
whole_seq_top_hits[$1] = $2.to_i
|
344
|
+
elsif /^\s*(.+?):\s*(\S+)\s*$/ =~ l
|
345
|
+
whole_seq_top_hits[$1] = $2
|
346
|
+
end
|
347
|
+
end
|
348
|
+
|
349
|
+
domain_top_hits = {}
|
350
|
+
data.each_line do |l|
|
351
|
+
if /^\s*(.+?):\s*(\d+)\s*$/ =~ l
|
352
|
+
domain_top_hits[$1] = $2.to_i
|
353
|
+
elsif /^\s*(.+?):\s*(\S+)\s*$/ =~ l
|
354
|
+
domain_top_hits[$1] = $2
|
355
|
+
end
|
356
|
+
end
|
357
|
+
|
358
|
+
[histogram, statistical_detail, total_seq_searched, \
|
359
|
+
whole_seq_top_hits, domain_top_hits]
|
360
|
+
end
|
361
|
+
private :parse_stat_data
|
362
|
+
|
363
|
+
|
364
|
+
# Container class for HMMER search hits.
|
365
|
+
class Hit
|
366
|
+
|
367
|
+
# An Array of Bio::HMMER::Report::Hsp objects.
|
368
|
+
attr_reader :hsps
|
369
|
+
|
370
|
+
#
|
371
|
+
attr_reader :accession
|
372
|
+
alias target_id accession
|
373
|
+
alias hit_id accession
|
374
|
+
alias entry_id accession
|
375
|
+
|
376
|
+
#
|
377
|
+
attr_reader :description
|
378
|
+
alias definition description
|
379
|
+
|
380
|
+
# Matching scores (total of all HSPs).
|
381
|
+
attr_reader :score
|
382
|
+
alias bit_score score
|
383
|
+
|
384
|
+
# E-value
|
385
|
+
attr_reader :evalue
|
386
|
+
|
387
|
+
# Number of domains
|
388
|
+
attr_reader :num
|
389
|
+
|
390
|
+
# Sets hit data.
|
391
|
+
def initialize(hit_data)
|
392
|
+
@hsps = Array.new
|
393
|
+
if /^(\S+)\s+(.*?)\s+(\S+)\s+(\S+)\s+(\S+)$/ =~ hit_data
|
394
|
+
@accession, @description, @score, @evalue, @num = \
|
395
|
+
[$1, $2, $3.to_f, $4.to_f, $5.to_i]
|
396
|
+
end
|
397
|
+
end
|
398
|
+
|
399
|
+
|
400
|
+
# Iterates on each Hsp object (Bio::HMMER::Report::Hsp).
|
401
|
+
def each
|
402
|
+
@hsps.each do |hsp|
|
403
|
+
yield hsp
|
404
|
+
end
|
405
|
+
end
|
406
|
+
alias :each_hsp :each
|
407
|
+
|
408
|
+
|
409
|
+
# Shows the hit description.
|
410
|
+
def target_def
|
411
|
+
if @hsps.size == 1
|
412
|
+
"<#{@hsps[0].domain}> #{@description}"
|
413
|
+
else
|
414
|
+
"<#{@num.to_s}> #{@description}"
|
415
|
+
end
|
416
|
+
end
|
417
|
+
|
418
|
+
# Appends a Bio::HMMER::Report::Hsp object.
|
419
|
+
def append_hsp(hsp)
|
420
|
+
@hsps << hsp
|
421
|
+
end
|
422
|
+
|
423
|
+
end # class Hit
|
424
|
+
|
425
|
+
|
426
|
+
# Container class for HMMER search hsps.
|
427
|
+
class Hsp
|
428
|
+
|
429
|
+
#
|
430
|
+
attr_reader :accession
|
431
|
+
alias target_id accession
|
432
|
+
|
433
|
+
#
|
434
|
+
attr_reader :domain
|
435
|
+
|
436
|
+
#
|
437
|
+
attr_reader :seq_f
|
438
|
+
|
439
|
+
#
|
440
|
+
attr_reader :seq_t
|
441
|
+
|
442
|
+
#
|
443
|
+
attr_reader :seq_ft
|
444
|
+
|
445
|
+
#
|
446
|
+
attr_reader :hmm_f
|
447
|
+
|
448
|
+
#
|
449
|
+
attr_reader :hmm_t
|
450
|
+
|
451
|
+
#
|
452
|
+
attr_reader :hmm_ft
|
453
|
+
|
454
|
+
# Score
|
455
|
+
attr_reader :score
|
456
|
+
alias bit_score score
|
457
|
+
|
458
|
+
# E-value
|
459
|
+
attr_reader :evalue
|
460
|
+
|
461
|
+
# Alignment midline
|
462
|
+
attr_reader :midline
|
463
|
+
|
464
|
+
#
|
465
|
+
attr_reader :hmmseq
|
466
|
+
|
467
|
+
#
|
468
|
+
attr_reader :flatseq
|
469
|
+
|
470
|
+
#
|
471
|
+
attr_reader :query_frame
|
472
|
+
|
473
|
+
#
|
474
|
+
attr_reader :target_frame
|
475
|
+
|
476
|
+
# CS Line
|
477
|
+
attr_reader :csline
|
478
|
+
|
479
|
+
# RF Line
|
480
|
+
attr_reader :rfline
|
481
|
+
|
482
|
+
# Sets hsps.
|
483
|
+
def initialize(hsp_data, is_hmmsearch)
|
484
|
+
@is_hmmsearch = is_hmmsearch
|
485
|
+
|
486
|
+
@accession, @domain, seq_f, seq_t, @seq_ft, hmm_f, hmm_t, @hmm_ft,\
|
487
|
+
score, evalue = hsp_data.split(' ')
|
488
|
+
@seq_f = seq_f.to_i
|
489
|
+
@seq_t = seq_t.to_i
|
490
|
+
@hmm_f = hmm_f.to_i
|
491
|
+
@hmm_t = hmm_t.to_i
|
492
|
+
@score = score.to_f
|
493
|
+
@evalue = evalue.to_f
|
494
|
+
@hmmseq = ''
|
495
|
+
@flatseq = ''
|
496
|
+
@midline = ''
|
497
|
+
@query_frame = 1
|
498
|
+
@target_frame = 1
|
499
|
+
# CS and RF lines are rarely used.
|
500
|
+
@csline = nil
|
501
|
+
@rfline = nil
|
502
|
+
end
|
503
|
+
|
504
|
+
#
|
505
|
+
def set_alignment(alignment)
|
506
|
+
# First, split the input alignment into an array of
|
507
|
+
# "alignment blocks." One block usually has three lines,
|
508
|
+
# i.e. hmmseq, midline and flatseq.
|
509
|
+
# However, although infrequent, it can contain CS or RF lines.
|
510
|
+
alignment.split(/ (?:\d+|-)\s*\n\n/).each do |blk|
|
511
|
+
lines = blk.split(/\n/)
|
512
|
+
cstmp = (lines[0] =~ /^ {16}CS/) ? lines.shift : nil
|
513
|
+
rftmp = (lines[0] =~ /^ {16}RF/) ? lines.shift : nil
|
514
|
+
aln_width = lines[0][/\S+/].length
|
515
|
+
@csline = @csline.to_s + cstmp[19, aln_width] if cstmp
|
516
|
+
@rfline = @rfline.to_s + rftmp[19, aln_width] if rftmp
|
517
|
+
@hmmseq += lines[0][19, aln_width]
|
518
|
+
@midline += lines[1][19, aln_width]
|
519
|
+
@flatseq += lines[2][19, aln_width]
|
520
|
+
end
|
521
|
+
@csline = @csline[3...-3] if @csline
|
522
|
+
@rfline = @rfline[3...-3] if @rfline
|
523
|
+
@hmmseq = @hmmseq[3...-3]
|
524
|
+
@midline = @midline[3...-3]
|
525
|
+
@flatseq = @flatseq[3...-3]
|
526
|
+
end
|
527
|
+
|
528
|
+
|
529
|
+
#
|
530
|
+
def query_seq
|
531
|
+
@is_hmmsearch ? @hmmseq : @flatseq
|
532
|
+
end
|
533
|
+
|
534
|
+
#
|
535
|
+
def target_seq
|
536
|
+
@is_hmmsearch ? @flatseq : @hmmseq
|
537
|
+
end
|
538
|
+
|
539
|
+
#
|
540
|
+
def target_from
|
541
|
+
@is_hmmsearch ? @seq_f : @hmm_f
|
542
|
+
end
|
543
|
+
|
544
|
+
#
|
545
|
+
def target_to
|
546
|
+
@is_hmmsearch ? @seq_t : @hmm_t
|
547
|
+
end
|
548
|
+
|
549
|
+
#
|
550
|
+
def query_from
|
551
|
+
@is_hmmsearch ? @hmm_f : @seq_f
|
552
|
+
end
|
553
|
+
|
554
|
+
#
|
555
|
+
def query_to
|
556
|
+
@is_hmmsearch ? @hmm_t : @seq_t
|
557
|
+
end
|
558
|
+
|
559
|
+
|
560
|
+
end # class Hsp
|
561
|
+
|
562
|
+
end # class Report
|
563
|
+
|
564
|
+
end # class HMMER
|
565
|
+
|
566
|
+
end # module Bio
|
567
|
+
|
568
|
+
|
569
|
+
if __FILE__ == $0
|
570
|
+
|
571
|
+
=begin
|
572
|
+
|
573
|
+
#
|
574
|
+
# for multiple reports in a single output file (hmmpfam)
|
575
|
+
#
|
576
|
+
Bio::HMMER.reports(ARGF.read) do |report|
|
577
|
+
report.hits.each do |hit|
|
578
|
+
hit.hsps.each do |hsp|
|
579
|
+
end
|
580
|
+
end
|
581
|
+
end
|
582
|
+
|
583
|
+
=end
|
584
|
+
|
585
|
+
begin
|
586
|
+
require 'pp'
|
587
|
+
alias p pp
|
588
|
+
rescue LoadError
|
589
|
+
end
|
590
|
+
|
591
|
+
rep = Bio::HMMER::Report.new(ARGF.read)
|
592
|
+
p rep
|
593
|
+
|
594
|
+
indent = 18
|
595
|
+
|
596
|
+
puts "### hmmer result"
|
597
|
+
print "name : ".rjust(indent)
|
598
|
+
p rep.program['name']
|
599
|
+
print "version : ".rjust(indent)
|
600
|
+
p rep.program['version']
|
601
|
+
print "copyright : ".rjust(indent)
|
602
|
+
p rep.program['copyright']
|
603
|
+
print "license : ".rjust(indent)
|
604
|
+
p rep.program['license']
|
605
|
+
|
606
|
+
print "HMM file : ".rjust(indent)
|
607
|
+
p rep.parameter['HMM file']
|
608
|
+
print "Sequence file : ".rjust(indent)
|
609
|
+
p rep.parameter['Sequence file']
|
610
|
+
|
611
|
+
print "Query sequence : ".rjust(indent)
|
612
|
+
p rep.query_info['Query sequence']
|
613
|
+
print "Accession : ".rjust(indent)
|
614
|
+
p rep.query_info['Accession']
|
615
|
+
print "Description : ".rjust(indent)
|
616
|
+
p rep.query_info['Description']
|
617
|
+
|
618
|
+
rep.each do |hit|
|
619
|
+
puts "## each hit"
|
620
|
+
print "accession : ".rjust(indent)
|
621
|
+
p [ hit.accession, hit.target_id, hit.hit_id, hit.entry_id ]
|
622
|
+
print "description : ".rjust(indent)
|
623
|
+
p [ hit.description, hit.definition ]
|
624
|
+
print "target_def : ".rjust(indent)
|
625
|
+
p hit.target_def
|
626
|
+
print "score : ".rjust(indent)
|
627
|
+
p [ hit.score, hit.bit_score ]
|
628
|
+
print "evalue : ".rjust(indent)
|
629
|
+
p hit.evalue
|
630
|
+
print "num : ".rjust(indent)
|
631
|
+
p hit.num
|
632
|
+
|
633
|
+
hit.each do |hsp|
|
634
|
+
puts "## each hsp"
|
635
|
+
print "accession : ".rjust(indent)
|
636
|
+
p [ hsp.accession, hsp.target_id ]
|
637
|
+
print "domain : ".rjust(indent)
|
638
|
+
p hsp.domain
|
639
|
+
print "seq_f : ".rjust(indent)
|
640
|
+
p hsp.seq_f
|
641
|
+
print "seq_t : ".rjust(indent)
|
642
|
+
p hsp.seq_t
|
643
|
+
print "seq_ft : ".rjust(indent)
|
644
|
+
p hsp.seq_ft
|
645
|
+
print "hmm_f : ".rjust(indent)
|
646
|
+
p hsp.hmm_f
|
647
|
+
print "hmm_t : ".rjust(indent)
|
648
|
+
p hsp.hmm_t
|
649
|
+
print "hmm_ft : ".rjust(indent)
|
650
|
+
p hsp.hmm_ft
|
651
|
+
print "score : ".rjust(indent)
|
652
|
+
p [ hsp.score, hsp.bit_score ]
|
653
|
+
print "evalue : ".rjust(indent)
|
654
|
+
p hsp.evalue
|
655
|
+
print "midline : ".rjust(indent)
|
656
|
+
p hsp.midline
|
657
|
+
print "hmmseq : ".rjust(indent)
|
658
|
+
p hsp.hmmseq
|
659
|
+
print "flatseq : ".rjust(indent)
|
660
|
+
p hsp.flatseq
|
661
|
+
print "query_frame : ".rjust(indent)
|
662
|
+
p hsp.query_frame
|
663
|
+
print "target_frame : ".rjust(indent)
|
664
|
+
p hsp.target_frame
|
665
|
+
|
666
|
+
print "query_seq : ".rjust(indent)
|
667
|
+
p hsp.query_seq # hmmseq, flatseq
|
668
|
+
print "target_seq : ".rjust(indent)
|
669
|
+
p hsp.target_seq # flatseq, hmmseq
|
670
|
+
print "target_from : ".rjust(indent)
|
671
|
+
p hsp.target_from # seq_f, hmm_f
|
672
|
+
print "target_to : ".rjust(indent)
|
673
|
+
p hsp.target_to # seq_t, hmm_t
|
674
|
+
print "query_from : ".rjust(indent)
|
675
|
+
p hsp.query_from # hmm_f, seq_f
|
676
|
+
print "query_to : ".rjust(indent)
|
677
|
+
p hsp.query_to # hmm_t, seq_t
|
678
|
+
end
|
679
|
+
end
|
680
|
+
|
681
|
+
end
|
682
|
+
|
683
|
+
|