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 +1,352 @@
|
|
1
|
+
#
|
2
|
+
# = bio/db/embl.rb - Common methods for EMBL style database classes
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2001-2006
|
5
|
+
# Mitsuteru C. Nakao <n@bioruby.org>
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: common.rb,v 1.12.2.5 2008/05/07 12:22:10 ngoto Exp $
|
9
|
+
#
|
10
|
+
# == Description
|
11
|
+
#
|
12
|
+
# EMBL style databases class
|
13
|
+
#
|
14
|
+
# This module defines a common framework among EMBL, UniProtKB, SWISS-PROT,
|
15
|
+
# TrEMBL. For more details, see the documentations in each embl/*.rb
|
16
|
+
# libraries.
|
17
|
+
#
|
18
|
+
# EMBL style format:
|
19
|
+
# ID - identification (begins each entry; 1 per entry)
|
20
|
+
# AC - accession number (>=1 per entry)
|
21
|
+
# SV - sequence version (1 per entry)
|
22
|
+
# DT - date (2 per entry)
|
23
|
+
# DE - description (>=1 per entry)
|
24
|
+
# KW - keyword (>=1 per entry)
|
25
|
+
# OS - organism species (>=1 per entry)
|
26
|
+
# OC - organism classification (>=1 per entry)
|
27
|
+
# OG - organelle (0 or 1 per entry)
|
28
|
+
# RN - reference number (>=1 per entry)
|
29
|
+
# RC - reference comment (>=0 per entry)
|
30
|
+
# RP - reference positions (>=1 per entry)
|
31
|
+
# RX - reference cross-reference (>=0 per entry)
|
32
|
+
# RA - reference author(s) (>=1 per entry)
|
33
|
+
# RG - reference group (>=0 per entry)
|
34
|
+
# RT - reference title (>=1 per entry)
|
35
|
+
# RL - reference location (>=1 per entry)
|
36
|
+
# DR - database cross-reference (>=0 per entry)
|
37
|
+
# FH - feature table header (0 or 2 per entry)
|
38
|
+
# FT - feature table data (>=0 per entry)
|
39
|
+
# CC - comments or notes (>=0 per entry)
|
40
|
+
# XX - spacer line (many per entry)
|
41
|
+
# SQ - sequence header (1 per entry)
|
42
|
+
# bb - (blanks) sequence data (>=1 per entry)
|
43
|
+
# // - termination line (ends each entry; 1 per entry)
|
44
|
+
#
|
45
|
+
# == Examples
|
46
|
+
#
|
47
|
+
# # Make a new parser class for EMBL style database entry.
|
48
|
+
# require 'bio/db/embl/common'
|
49
|
+
# module Bio
|
50
|
+
# class NEWDB < EMBLDB
|
51
|
+
# include Bio::EMBLDB::Common
|
52
|
+
# end
|
53
|
+
# end
|
54
|
+
#
|
55
|
+
# == References
|
56
|
+
#
|
57
|
+
# * The EMBL Nucleotide Sequence Database
|
58
|
+
# http://www.ebi.ac.uk/embl/
|
59
|
+
#
|
60
|
+
# * The EMBL Nucleotide Sequence Database: Users Manual
|
61
|
+
# http://www.ebi.ac.uk/embl/Documentation/User_manual/usrman.html
|
62
|
+
#
|
63
|
+
# * Swiss-Prot Protein knowledgebase. TrEMBL Computer-annotated supplement
|
64
|
+
# to Swiss-Prot
|
65
|
+
# http://au.expasy.org/sprot/
|
66
|
+
#
|
67
|
+
# * UniProt
|
68
|
+
# http://uniprot.org/
|
69
|
+
#
|
70
|
+
# * The UniProtKB/SwissProt/TrEMBL User Manual
|
71
|
+
# http://www.expasy.org/sprot/userman.html
|
72
|
+
#
|
73
|
+
|
74
|
+
require 'bio/db'
|
75
|
+
require 'bio/reference'
|
76
|
+
require 'bio/compat/references'
|
77
|
+
|
78
|
+
module Bio
|
79
|
+
class EMBLDB
|
80
|
+
module Common
|
81
|
+
|
82
|
+
DELIMITER = "\n//\n"
|
83
|
+
RS = DELIMITER
|
84
|
+
TAGSIZE = 5
|
85
|
+
|
86
|
+
def initialize(entry)
|
87
|
+
super(entry, TAGSIZE)
|
88
|
+
end
|
89
|
+
|
90
|
+
# returns a Array of accession numbers in the AC lines.
|
91
|
+
#
|
92
|
+
# AC Line
|
93
|
+
# "AC A12345; B23456;"
|
94
|
+
# AC [AC1;]+
|
95
|
+
#
|
96
|
+
# Accession numbers format:
|
97
|
+
# 1 2 3 4 5 6
|
98
|
+
# [O,P,Q] [0-9] [A-Z, 0-9] [A-Z, 0-9] [A-Z, 0-9] [0-9]
|
99
|
+
def ac
|
100
|
+
unless @data['AC']
|
101
|
+
tmp = Array.new
|
102
|
+
field_fetch('AC').split(/ /).each do |e|
|
103
|
+
tmp.push(e.sub(/;/,''))
|
104
|
+
end
|
105
|
+
@data['AC'] = tmp
|
106
|
+
end
|
107
|
+
@data['AC']
|
108
|
+
end
|
109
|
+
alias accessions ac
|
110
|
+
|
111
|
+
|
112
|
+
# returns the first accession number in the AC lines
|
113
|
+
def accession
|
114
|
+
ac[0]
|
115
|
+
end
|
116
|
+
|
117
|
+
|
118
|
+
# returns a String int the DE line.
|
119
|
+
#
|
120
|
+
# DE Line
|
121
|
+
def de
|
122
|
+
unless @data['DE']
|
123
|
+
@data['DE'] = fetch('DE')
|
124
|
+
end
|
125
|
+
@data['DE']
|
126
|
+
end
|
127
|
+
alias description de
|
128
|
+
alias definition de # API
|
129
|
+
|
130
|
+
|
131
|
+
|
132
|
+
# returns contents in the OS line.
|
133
|
+
# * Bio::EMBLDB#os -> Array of <OS Hash>
|
134
|
+
# where <OS Hash> is:
|
135
|
+
# [{'name'=>'Human', 'os'=>'Homo sapiens'},
|
136
|
+
# {'name'=>'Rat', 'os'=>'Rattus norveticus'}]
|
137
|
+
# * Bio::SPTR#os[0]['name'] => "Human"
|
138
|
+
# * Bio::SPTR#os[0] => {'name'=>"Human", 'os'=>'Homo sapiens'}
|
139
|
+
# * Bio::STPR#os(0) => "Homo sapiens (Human)"
|
140
|
+
#
|
141
|
+
# OS Line; organism species (>=1)
|
142
|
+
# "OS Trifolium repens (white clover)"
|
143
|
+
#
|
144
|
+
# OS Genus species (name).
|
145
|
+
# OS Genus species (name0) (name1).
|
146
|
+
# OS Genus species (name0) (name1).
|
147
|
+
# OS Genus species (name0), G s0 (name0), and G s (name1).
|
148
|
+
def os(num = nil)
|
149
|
+
unless @data['OS']
|
150
|
+
os = Array.new
|
151
|
+
fetch('OS').split(/, and|, /).each do |tmp|
|
152
|
+
if tmp =~ /([A-Z][a-z]* *[\w\d \:\'\+\-]+[\w\d])/
|
153
|
+
org = $1
|
154
|
+
tmp =~ /(\(.+\))/
|
155
|
+
os.push({'name' => $1, 'os' => org})
|
156
|
+
else
|
157
|
+
raise "Error: OS Line. #{$!}\n#{fetch('OS')}\n"
|
158
|
+
end
|
159
|
+
end
|
160
|
+
@data['OS'] = os
|
161
|
+
end
|
162
|
+
if num
|
163
|
+
# EX. "Trifolium repens (white clover)"
|
164
|
+
"#{@data['OS'][num]['os']} {#data['OS'][num]['name']"
|
165
|
+
end
|
166
|
+
@data['OS']
|
167
|
+
end
|
168
|
+
|
169
|
+
|
170
|
+
# returns contents in the OG line.
|
171
|
+
# * Bio::EMBLDB::Common#og -> [ <ogranella String>* ]
|
172
|
+
#
|
173
|
+
# OG Line; organella (0 or 1/entry)
|
174
|
+
# OG Plastid; Chloroplast.
|
175
|
+
# OG Mitochondrion.
|
176
|
+
# OG Plasmid sym pNGR234a.
|
177
|
+
# OG Plastid; Cyanelle.
|
178
|
+
# OG Plasmid pSymA (megaplasmid 1).
|
179
|
+
# OG Plasmid pNRC100, Plasmid pNRC200, and Plasmid pHH1.
|
180
|
+
def og
|
181
|
+
unless @data['OG']
|
182
|
+
og = Array.new
|
183
|
+
if get('OG').size > 0
|
184
|
+
ogstr = fetch('OG')
|
185
|
+
ogstr.sub!(/\.$/,'')
|
186
|
+
ogstr.sub!(/ and/,'')
|
187
|
+
ogstr.sub!(/;/, ',')
|
188
|
+
ogstr.split(',').each do |tmp|
|
189
|
+
og.push(tmp.strip)
|
190
|
+
end
|
191
|
+
end
|
192
|
+
@data['OG'] = og
|
193
|
+
end
|
194
|
+
@data['OG']
|
195
|
+
end
|
196
|
+
|
197
|
+
|
198
|
+
# returns contents in the OC line.
|
199
|
+
# * Bio::EMBLDB::Common#oc -> [ <organism class String>* ]
|
200
|
+
# OC Line; organism classification (>=1)
|
201
|
+
# OC Eukaryota; Alveolata; Apicomplexa; Piroplasmida; Theileriidae;
|
202
|
+
# OC Theileria.
|
203
|
+
def oc
|
204
|
+
unless @data['OC']
|
205
|
+
begin
|
206
|
+
@data['OC'] = fetch('OC').sub(/.$/,'').split(/;/).map {|e|
|
207
|
+
e.strip
|
208
|
+
}
|
209
|
+
rescue NameError
|
210
|
+
nil
|
211
|
+
end
|
212
|
+
end
|
213
|
+
@data['OC']
|
214
|
+
end
|
215
|
+
|
216
|
+
# returns keywords in the KW line.
|
217
|
+
# * Bio::EMBLDB::Common#kw -> [ <keyword>* ]
|
218
|
+
# KW Line; keyword (>=1)
|
219
|
+
# KW [Keyword;]+
|
220
|
+
def kw
|
221
|
+
unless @data['KW']
|
222
|
+
if get('KW').size > 0
|
223
|
+
tmp = fetch('KW').sub(/.$/,'')
|
224
|
+
@data['KW'] = tmp.split(/;/).map {|e| e.strip }
|
225
|
+
else
|
226
|
+
@data['KW'] = []
|
227
|
+
end
|
228
|
+
end
|
229
|
+
@data['KW']
|
230
|
+
end
|
231
|
+
alias keywords kw
|
232
|
+
|
233
|
+
|
234
|
+
# returns contents in the R lines.
|
235
|
+
# * Bio::EMBLDB::Common#ref -> [ <refernece information Hash>* ]
|
236
|
+
# where <reference information Hash> is:
|
237
|
+
# {'RN' => '', 'RC' => '', 'RP' => '', 'RX' => '',
|
238
|
+
# 'RA' => '', 'RT' => '', 'RL' => '', 'RG' => ''}
|
239
|
+
#
|
240
|
+
# R Lines
|
241
|
+
# * RN RC RP RX RA RT RL RG
|
242
|
+
def ref
|
243
|
+
unless @data['R']
|
244
|
+
ary = Array.new
|
245
|
+
get('R').split(/\nRN /).each do |str|
|
246
|
+
raw = {'RN' => '', 'RC' => '', 'RP' => '', 'RX' => '',
|
247
|
+
'RA' => '', 'RT' => '', 'RL' => '', 'RG' => ''}
|
248
|
+
str = 'RN ' + str unless /^RN / =~ str
|
249
|
+
str.split("\n").each do |line|
|
250
|
+
if /^(R[NPXARLCTG]) (.+)/ =~ line
|
251
|
+
raw[$1] += $2 + ' '
|
252
|
+
else
|
253
|
+
raise "Invalid format in R lines, \n[#{line}]\n"
|
254
|
+
end
|
255
|
+
end
|
256
|
+
raw.each_value {|v|
|
257
|
+
v.strip!
|
258
|
+
v.sub!(/^"/,'')
|
259
|
+
v.sub!(/;$/,'')
|
260
|
+
v.sub!(/"$/,'')
|
261
|
+
}
|
262
|
+
ary.push(raw)
|
263
|
+
end
|
264
|
+
@data['R'] = ary
|
265
|
+
end
|
266
|
+
@data['R']
|
267
|
+
end
|
268
|
+
|
269
|
+
# returns Bio::Reference object from Bio::EMBLDB::Common#ref.
|
270
|
+
# * Bio::EMBLDB::Common#ref -> Bio::References
|
271
|
+
def references
|
272
|
+
unless @data['references']
|
273
|
+
ary = self.ref.map {|ent|
|
274
|
+
hash = Hash.new
|
275
|
+
ent.each {|key, value|
|
276
|
+
case key
|
277
|
+
when 'RN'
|
278
|
+
if /\[(\d+)\]/ =~ value.to_s
|
279
|
+
hash['embl_gb_record_number'] = $1.to_i
|
280
|
+
end
|
281
|
+
when 'RC'
|
282
|
+
unless value.to_s.strip.empty?
|
283
|
+
hash['comments'] ||= []
|
284
|
+
hash['comments'].push value
|
285
|
+
end
|
286
|
+
when 'RP'
|
287
|
+
hash['sequence_position'] = value
|
288
|
+
when 'RA'
|
289
|
+
a = value.split(/\, /)
|
290
|
+
a.each do |x|
|
291
|
+
x.sub!(/( [^ ]+)\z/, ",\\1")
|
292
|
+
end
|
293
|
+
hash['authors'] = a
|
294
|
+
when 'RT'
|
295
|
+
hash['title'] = value
|
296
|
+
when 'RL'
|
297
|
+
if /(.*) (\d+) *(\(([^\)]+)\))?(\, |\:)([a-zA-Z\d]+\-[a-zA-Z\d]+) *\((\d+)\)\.?\z/ =~ value.to_s
|
298
|
+
hash['journal'] = $1.rstrip
|
299
|
+
hash['volume'] = $2
|
300
|
+
hash['issue'] = $4
|
301
|
+
hash['pages'] = $6
|
302
|
+
hash['year'] = $7
|
303
|
+
else
|
304
|
+
hash['journal'] = value
|
305
|
+
end
|
306
|
+
when 'RX' # PUBMED, DOI, (AGRICOLA)
|
307
|
+
value.split(/\. /).each {|item|
|
308
|
+
tag, xref = item.split(/\; /).map {|i| i.strip.sub(/\.\z/, '') }
|
309
|
+
hash[ tag.downcase ] = xref
|
310
|
+
}
|
311
|
+
end
|
312
|
+
}
|
313
|
+
Reference.new(hash)
|
314
|
+
}
|
315
|
+
@data['references'] = ary.extend(Bio::References::BackwardCompatibility)
|
316
|
+
end
|
317
|
+
@data['references']
|
318
|
+
end
|
319
|
+
|
320
|
+
|
321
|
+
# returns contents in the DR line.
|
322
|
+
# * Bio::EMBLDB::Common#dr -> [ <Database cross-reference Hash>* ]
|
323
|
+
# where <Database cross-reference Hash> is:
|
324
|
+
# * Bio::EMBLDB::Common#dr {|k,v| }
|
325
|
+
#
|
326
|
+
# DR Line; defabases cross-reference (>=0)
|
327
|
+
# a cross_ref pre one line
|
328
|
+
# "DR database_identifier; primary_identifier; secondary_identifier."
|
329
|
+
def dr
|
330
|
+
unless @data['DR']
|
331
|
+
tmp = Hash.new
|
332
|
+
self.get('DR').split(/\n/).each do |db|
|
333
|
+
a = db.sub(/^DR /,'').sub(/.$/,'').strip.split(/;[ ]/)
|
334
|
+
dbname = a.shift
|
335
|
+
tmp[dbname] = Array.new unless tmp[dbname]
|
336
|
+
tmp[dbname].push(a)
|
337
|
+
end
|
338
|
+
@data['DR'] = tmp
|
339
|
+
end
|
340
|
+
if block_given?
|
341
|
+
@data['DR'].each do |k,v|
|
342
|
+
yield(k, v)
|
343
|
+
end
|
344
|
+
else
|
345
|
+
@data['DR']
|
346
|
+
end
|
347
|
+
end
|
348
|
+
|
349
|
+
end # module Common
|
350
|
+
end # class EMBLDB
|
351
|
+
end # module Bio
|
352
|
+
|
@@ -0,0 +1,500 @@
|
|
1
|
+
#
|
2
|
+
# = bio/db/embl/embl.rb - EMBL database class
|
3
|
+
#
|
4
|
+
#
|
5
|
+
# Copyright:: Copyright (C) 2001-2007
|
6
|
+
# Mitsuteru C. Nakao <n@bioruby.org>
|
7
|
+
# Jan Aerts <jan.aerts@bbsrc.ac.uk>
|
8
|
+
# License:: The Ruby License
|
9
|
+
#
|
10
|
+
# $Id: embl.rb,v 1.29.2.7 2008/06/17 16:04:36 ngoto Exp $
|
11
|
+
#
|
12
|
+
# == Description
|
13
|
+
#
|
14
|
+
# Parser class for EMBL database entry.
|
15
|
+
#
|
16
|
+
# == Examples
|
17
|
+
#
|
18
|
+
# emb = Bio::EMBL.new($<.read)
|
19
|
+
# emb.entry_id
|
20
|
+
# emb.each_cds do |cds|
|
21
|
+
# cds # A CDS in feature table.
|
22
|
+
# end
|
23
|
+
# emb.seq #=> "ACGT..."
|
24
|
+
#
|
25
|
+
# == References
|
26
|
+
#
|
27
|
+
# * The EMBL Nucleotide Sequence Database
|
28
|
+
# http://www.ebi.ac.uk/embl/
|
29
|
+
#
|
30
|
+
# * The EMBL Nucleotide Sequence Database: Users Manual
|
31
|
+
# http://www.ebi.ac.uk/embl/Documentation/User_manual/usrman.html
|
32
|
+
#
|
33
|
+
|
34
|
+
require 'date'
|
35
|
+
require 'bio/db'
|
36
|
+
require 'bio/db/embl/common'
|
37
|
+
require 'bio/compat/features'
|
38
|
+
require 'bio/compat/references'
|
39
|
+
require 'bio/sequence'
|
40
|
+
require 'bio/sequence/dblink'
|
41
|
+
|
42
|
+
module Bio
|
43
|
+
class EMBL < EMBLDB
|
44
|
+
include Bio::EMBLDB::Common
|
45
|
+
|
46
|
+
# returns contents in the ID line.
|
47
|
+
# * Bio::EMBL#id_line -> <ID Hash>
|
48
|
+
# where <ID Hash> is:
|
49
|
+
# {'ENTRY_NAME' => String, 'MOLECULE_TYPE' => String, 'DIVISION' => String,
|
50
|
+
# 'SEQUENCE_LENGTH' => Int, 'SEQUENCE_VERSION' => Int}
|
51
|
+
#
|
52
|
+
# ID Line
|
53
|
+
# "ID ENTRY_NAME DATA_CLASS; MOLECULE_TYPE; DIVISION; SEQUENCE_LENGTH BP."
|
54
|
+
#
|
55
|
+
# DATA_CLASS = ['standard']
|
56
|
+
#
|
57
|
+
# MOLECULE_TYPE: DNA RNA XXX
|
58
|
+
#
|
59
|
+
# Code ( DIVISION )
|
60
|
+
# EST (ESTs)
|
61
|
+
# PHG (Bacteriophage)
|
62
|
+
# FUN (Fungi)
|
63
|
+
# GSS (Genome survey)
|
64
|
+
# HTC (High Throughput cDNAs)
|
65
|
+
# HTG (HTGs)
|
66
|
+
# HUM (Human)
|
67
|
+
# INV (Invertebrates)
|
68
|
+
# ORG (Organelles)
|
69
|
+
# MAM (Other Mammals)
|
70
|
+
# VRT (Other Vertebrates)
|
71
|
+
# PLN (Plants)
|
72
|
+
# PRO (Prokaryotes)
|
73
|
+
# ROD (Rodents)
|
74
|
+
# SYN (Synthetic)
|
75
|
+
# STS (STSs)
|
76
|
+
# UNC (Unclassified)
|
77
|
+
# VRL (Viruses)
|
78
|
+
#
|
79
|
+
# Rel 89-
|
80
|
+
# ID CD789012; SV 4; linear; genomic DNA; HTG; MAM; 500 BP.
|
81
|
+
# ID <1>; SV <2>; <3>; <4>; <5>; <6>; <7> BP.
|
82
|
+
# 1. Primary accession number
|
83
|
+
# 2. Sequence version number
|
84
|
+
# 3. Topology: 'circular' or 'linear'
|
85
|
+
# 4. Molecule type (see note 1 below)
|
86
|
+
# 5. Data class (see section 3.1)
|
87
|
+
# 6. Taxonomic division (see section 3.2)
|
88
|
+
# 7. Sequence length (see note 2 below)
|
89
|
+
def id_line(key=nil)
|
90
|
+
unless @data['ID']
|
91
|
+
tmp = Hash.new
|
92
|
+
idline = fetch('ID').split(/; +/)
|
93
|
+
tmp['ENTRY_NAME'], tmp['DATA_CLASS'] = idline.shift.split(/ +/)
|
94
|
+
if idline.first =~ /^SV/
|
95
|
+
tmp['SEQUENCE_VERSION'] = idline.shift.split(' ').last
|
96
|
+
tmp['TOPOLOGY'] = idline.shift
|
97
|
+
tmp['MOLECULE_TYPE'] = idline.shift
|
98
|
+
tmp['DATA_CLASS'] = idline.shift
|
99
|
+
else
|
100
|
+
tmp['MOLECULE_TYPE'] = idline.shift
|
101
|
+
end
|
102
|
+
tmp['DIVISION'] = idline.shift
|
103
|
+
tmp['SEQUENCE_LENGTH'] = idline.shift.strip.split(' ').first.to_i
|
104
|
+
|
105
|
+
@data['ID'] = tmp
|
106
|
+
end
|
107
|
+
|
108
|
+
if key
|
109
|
+
@data['ID'][key]
|
110
|
+
else
|
111
|
+
@data['ID']
|
112
|
+
end
|
113
|
+
end
|
114
|
+
|
115
|
+
# returns ENTRY_NAME in the ID line.
|
116
|
+
# * Bio::EMBL#entry -> String
|
117
|
+
def entry
|
118
|
+
id_line('ENTRY_NAME')
|
119
|
+
end
|
120
|
+
alias entry_name entry
|
121
|
+
alias entry_id entry
|
122
|
+
|
123
|
+
# returns MOLECULE_TYPE in the ID line.
|
124
|
+
# * Bio::EMBL#molecule -> String
|
125
|
+
def molecule
|
126
|
+
id_line('MOLECULE_TYPE')
|
127
|
+
end
|
128
|
+
alias molecule_type molecule
|
129
|
+
|
130
|
+
def data_class
|
131
|
+
id_line('DATA_CLASS')
|
132
|
+
end
|
133
|
+
|
134
|
+
def topology
|
135
|
+
id_line('TOPOLOGY')
|
136
|
+
end
|
137
|
+
|
138
|
+
# returns DIVISION in the ID line.
|
139
|
+
# * Bio::EMBL#division -> String
|
140
|
+
def division
|
141
|
+
id_line('DIVISION')
|
142
|
+
end
|
143
|
+
|
144
|
+
# returns SEQUENCE_LENGTH in the ID line.
|
145
|
+
# * Bio::EMBL#sequencelength -> String
|
146
|
+
def sequence_length
|
147
|
+
id_line('SEQUENCE_LENGTH')
|
148
|
+
end
|
149
|
+
alias seqlen sequence_length
|
150
|
+
|
151
|
+
|
152
|
+
# AC Line
|
153
|
+
# "AC A12345; B23456;"
|
154
|
+
|
155
|
+
|
156
|
+
# returns the version information in the sequence version (SV) line.
|
157
|
+
# * Bio::EMBL#sv -> Accession.Version in String
|
158
|
+
# * Bio::EMBL#version -> accession in Int
|
159
|
+
#
|
160
|
+
# SV Line; sequence version (1/entry)
|
161
|
+
# SV Accession.Version
|
162
|
+
def sv
|
163
|
+
if (v = field_fetch('SV').sub(/;/,'')) == ""
|
164
|
+
[id_line['ENTRY_NAME'], id_line['SEQUENCE_VERSION']].join('.')
|
165
|
+
else
|
166
|
+
v
|
167
|
+
end
|
168
|
+
end
|
169
|
+
def version
|
170
|
+
(sv.split(".")[1] || id_line['SEQUENCE_VERSION']).to_i
|
171
|
+
end
|
172
|
+
|
173
|
+
|
174
|
+
# returns contents in the date (DT) line.
|
175
|
+
# * Bio::EMBL#dt -> <DT Hash>
|
176
|
+
# where <DT Hash> is:
|
177
|
+
# {}
|
178
|
+
# * Bio::EMBL#dt(key) -> String
|
179
|
+
# keys: 'created' and 'updated'
|
180
|
+
#
|
181
|
+
# DT Line; date (2/entry)
|
182
|
+
def dt(key=nil)
|
183
|
+
unless @data['DT']
|
184
|
+
tmp = Hash.new
|
185
|
+
dt_line = self.get('DT').split(/\n/)
|
186
|
+
tmp['created'] = dt_line[0].sub(/\w{2} /,'').strip
|
187
|
+
tmp['updated'] = dt_line[1].sub(/\w{2} /,'').strip
|
188
|
+
@data['DT'] = tmp
|
189
|
+
end
|
190
|
+
if key
|
191
|
+
@data['DT'][key]
|
192
|
+
else
|
193
|
+
@data['DT']
|
194
|
+
end
|
195
|
+
end
|
196
|
+
|
197
|
+
|
198
|
+
|
199
|
+
##
|
200
|
+
# DE Line; description (>=1)
|
201
|
+
#
|
202
|
+
|
203
|
+
|
204
|
+
##
|
205
|
+
# KW Line; keyword (>=1)
|
206
|
+
# KW [Keyword;]+
|
207
|
+
#
|
208
|
+
# Bio::EMBLDB#kw -> Array
|
209
|
+
# #keywords -> Array
|
210
|
+
|
211
|
+
|
212
|
+
##
|
213
|
+
# OS Line; organism species (>=1)
|
214
|
+
# OS Genus species (name)
|
215
|
+
# "OS Trifolium repens (white clover)"
|
216
|
+
#
|
217
|
+
# Bio::EMBLDB#os -> Array
|
218
|
+
|
219
|
+
|
220
|
+
##
|
221
|
+
# OC Line; organism classification (>=1)
|
222
|
+
#
|
223
|
+
# Bio::EMBLDB#oc -> Array
|
224
|
+
|
225
|
+
|
226
|
+
##
|
227
|
+
# OG Line; organella (0 or 1/entry)
|
228
|
+
# ["Mitochondrion", "Chloroplast","Kinetoplast", "Cyanelle", "Plastid"]
|
229
|
+
# or a plasmid name (e.g. "Plasmid pBR322").
|
230
|
+
#
|
231
|
+
# Bio::EMBLDB#og -> String
|
232
|
+
|
233
|
+
|
234
|
+
##
|
235
|
+
# R Lines
|
236
|
+
# RN RC RP RX RA RT RL
|
237
|
+
#
|
238
|
+
# Bio::EMBLDB#ref
|
239
|
+
|
240
|
+
|
241
|
+
##
|
242
|
+
# DR Line; defabases cross-regerence (>=0)
|
243
|
+
# "DR database_identifier; primary_identifier; secondary_identifier."
|
244
|
+
#
|
245
|
+
# Bio::EMBLDB#dr
|
246
|
+
|
247
|
+
|
248
|
+
# returns feature table header (String) in the feature header (FH) line.
|
249
|
+
#
|
250
|
+
# FH Line; feature table header (0 or 2)
|
251
|
+
def fh
|
252
|
+
fetch('FH')
|
253
|
+
end
|
254
|
+
|
255
|
+
# returns contents in the feature table (FT) lines.
|
256
|
+
# * Bio::EMBL#ft -> Bio::Features
|
257
|
+
# * Bio::EMBL#ft {} -> {|Bio::Feature| }
|
258
|
+
#
|
259
|
+
# same as features method in bio/db/genbank.rb
|
260
|
+
#
|
261
|
+
# FT Line; feature table data (>=0)
|
262
|
+
def ft
|
263
|
+
unless @data['FT']
|
264
|
+
ary = Array.new
|
265
|
+
in_quote = false
|
266
|
+
@orig['FT'].each_line do |line|
|
267
|
+
next if line =~ /^FEATURES/
|
268
|
+
|
269
|
+
head = line[0,20].strip # feature key (source, CDS, ...)
|
270
|
+
body = line[20,60].chomp # feature value (position, /qualifier=)
|
271
|
+
if line =~ /^FT {3}(\S+)/
|
272
|
+
ary.push([ $1, body ]) # [ feature, position, /q="data", ... ]
|
273
|
+
elsif body =~ /^ \// and not in_quote
|
274
|
+
ary.last.push(body) # /q="data..., /q=data, /q
|
275
|
+
|
276
|
+
if body =~ /=" / and body !~ /"$/
|
277
|
+
in_quote = true
|
278
|
+
end
|
279
|
+
|
280
|
+
else
|
281
|
+
ary.last.last << body # ...data..., ...data..."
|
282
|
+
|
283
|
+
if body =~ /"$/
|
284
|
+
in_quote = false
|
285
|
+
end
|
286
|
+
end
|
287
|
+
end
|
288
|
+
|
289
|
+
ary.map! do |subary|
|
290
|
+
parse_qualifiers(subary)
|
291
|
+
end
|
292
|
+
|
293
|
+
@data['FT'] = ary.extend(Bio::Features::BackwardCompatibility)
|
294
|
+
end
|
295
|
+
if block_given?
|
296
|
+
@data['FT'].each do |feature|
|
297
|
+
yield feature
|
298
|
+
end
|
299
|
+
else
|
300
|
+
@data['FT']
|
301
|
+
end
|
302
|
+
end
|
303
|
+
alias features ft
|
304
|
+
|
305
|
+
# iterates on CDS features in the FT lines.
|
306
|
+
def each_cds
|
307
|
+
ft.each do |cds_feature|
|
308
|
+
if cds_feature.feature == 'CDS'
|
309
|
+
yield cds_feature
|
310
|
+
end
|
311
|
+
end
|
312
|
+
end
|
313
|
+
|
314
|
+
# iterates on gene features in the FT lines.
|
315
|
+
def each_gene
|
316
|
+
ft.each do |gene_feature|
|
317
|
+
if gene_feature.feature == 'gene'
|
318
|
+
yield gene_feature
|
319
|
+
end
|
320
|
+
end
|
321
|
+
end
|
322
|
+
|
323
|
+
|
324
|
+
# returns comment text in the comments (CC) line.
|
325
|
+
#
|
326
|
+
# CC Line; comments of notes (>=0)
|
327
|
+
def cc
|
328
|
+
get('CC').to_s.gsub(/^CC /, '')
|
329
|
+
end
|
330
|
+
alias comment cc
|
331
|
+
|
332
|
+
##
|
333
|
+
# XX Line; spacer line (many)
|
334
|
+
# def nxx
|
335
|
+
# end
|
336
|
+
|
337
|
+
|
338
|
+
# returns sequence header information in the sequence header (SQ) line.
|
339
|
+
# * Bio::EMBL#sq -> <SQ Hash>
|
340
|
+
# where <SQ Hash> is:
|
341
|
+
# {'ntlen' => Int, 'other' => Int,
|
342
|
+
# 'a' => Int, 'c' => Int, 'g' => Int, 't' => Int}
|
343
|
+
# * Bio::EMBL#sq(base) -> <base content in Int>
|
344
|
+
# * Bio::EMBL#sq[base] -> <base content in Int>
|
345
|
+
#
|
346
|
+
# SQ Line; sequence header (1/entry)
|
347
|
+
# SQ Sequence 1859 BP; 609 A; 314 C; 355 G; 581 T; 0 other;
|
348
|
+
def sq(base = nil)
|
349
|
+
unless @data['SQ']
|
350
|
+
fetch('SQ') =~ \
|
351
|
+
/(\d+) BP\; (\d+) A; (\d+) C; (\d+) G; (\d+) T; (\d+) other;/
|
352
|
+
@data['SQ'] = {'ntlen' => $1.to_i, 'other' => $6.to_i,
|
353
|
+
'a' => $2.to_i, 'c' => $3.to_i , 'g' => $4.to_i, 't' => $5.to_i}
|
354
|
+
else
|
355
|
+
@data['SQ']
|
356
|
+
end
|
357
|
+
|
358
|
+
if base
|
359
|
+
@data['SQ'][base.downcase]
|
360
|
+
else
|
361
|
+
@data['SQ']
|
362
|
+
end
|
363
|
+
end
|
364
|
+
|
365
|
+
|
366
|
+
# returns the nucleotie sequence in this entry.
|
367
|
+
# * Bio::EMBL#seq -> Bio::Sequence::NA
|
368
|
+
#
|
369
|
+
# @orig[''] as sequence
|
370
|
+
# bb Line; (blanks) sequence data (>=1)
|
371
|
+
def seq
|
372
|
+
Bio::Sequence::NA.new( fetch('').gsub(/ /,'').gsub(/\d+/,'') )
|
373
|
+
end
|
374
|
+
alias naseq seq
|
375
|
+
alias ntseq seq
|
376
|
+
|
377
|
+
#--
|
378
|
+
# // Line; termination line (end; 1/entry)
|
379
|
+
#++
|
380
|
+
|
381
|
+
# modified date. Returns Date object, String or nil.
|
382
|
+
def date_modified
|
383
|
+
parse_date(self.dt['updated'])
|
384
|
+
end
|
385
|
+
|
386
|
+
# created date. Returns Date object, String or nil.
|
387
|
+
def date_created
|
388
|
+
parse_date(self.dt['created'])
|
389
|
+
end
|
390
|
+
|
391
|
+
# release number when last updated
|
392
|
+
def release_modified
|
393
|
+
parse_release_version(self.dt['updated'])[0]
|
394
|
+
end
|
395
|
+
|
396
|
+
# release number when created
|
397
|
+
def release_created
|
398
|
+
parse_release_version(self.dt['created'])[0]
|
399
|
+
end
|
400
|
+
|
401
|
+
# entry version number numbered by EMBL
|
402
|
+
def entry_version
|
403
|
+
parse_release_version(self.dt['updated'])[1]
|
404
|
+
end
|
405
|
+
|
406
|
+
# parse date string. Returns Date object.
|
407
|
+
def parse_date(str)
|
408
|
+
begin
|
409
|
+
Date.parse(str)
|
410
|
+
rescue ArgumentError, TypeError, NoMethodError, NameError
|
411
|
+
str
|
412
|
+
end
|
413
|
+
end
|
414
|
+
private :parse_date
|
415
|
+
|
416
|
+
# extracts release and version numbers from DT line
|
417
|
+
def parse_release_version(str)
|
418
|
+
return [ nil, nil ] unless str
|
419
|
+
a = str.split(/[\(\,\)]/)
|
420
|
+
dstr = a.shift
|
421
|
+
rel = nil
|
422
|
+
ver = nil
|
423
|
+
a.each do |x|
|
424
|
+
case x
|
425
|
+
when /Rel\.\s*(.+)/
|
426
|
+
rel = $1.strip
|
427
|
+
when /Version\s*(.+)/
|
428
|
+
ver = $1.strip
|
429
|
+
end
|
430
|
+
end
|
431
|
+
[ rel, ver ]
|
432
|
+
end
|
433
|
+
private :parse_release_version
|
434
|
+
|
435
|
+
# database references (DR).
|
436
|
+
# Returns an array of Bio::Sequence::DBLink objects.
|
437
|
+
def dblinks
|
438
|
+
get('DR').split(/\n/).collect { |x|
|
439
|
+
Bio::Sequence::DBLink.parse_embl_DR_line(x)
|
440
|
+
}
|
441
|
+
end
|
442
|
+
|
443
|
+
# species
|
444
|
+
def species
|
445
|
+
self.fetch('OS')
|
446
|
+
end
|
447
|
+
|
448
|
+
# taxonomy classfication
|
449
|
+
alias classification oc
|
450
|
+
|
451
|
+
# features
|
452
|
+
alias features ft
|
453
|
+
|
454
|
+
|
455
|
+
# converts the entry to Bio::Sequence object
|
456
|
+
# ---
|
457
|
+
# *Arguments*::
|
458
|
+
# *Returns*:: Bio::Sequence object
|
459
|
+
def to_biosequence
|
460
|
+
Bio::Sequence.adapter(self, Bio::Sequence::Adapter::EMBL)
|
461
|
+
end
|
462
|
+
|
463
|
+
### private methods
|
464
|
+
|
465
|
+
private
|
466
|
+
|
467
|
+
##
|
468
|
+
# same as Bio::GenBank#parse_qualifiers(feature)
|
469
|
+
def parse_qualifiers(ary)
|
470
|
+
feature = Feature.new
|
471
|
+
|
472
|
+
feature.feature = ary.shift
|
473
|
+
feature.position = ary.shift.gsub(/\s/, '')
|
474
|
+
|
475
|
+
ary.each do |f|
|
476
|
+
if f =~ %r{/([^=]+)=?"?([^"]*)"?}
|
477
|
+
qualifier, value = $1, $2
|
478
|
+
|
479
|
+
if value.empty?
|
480
|
+
value = true
|
481
|
+
end
|
482
|
+
|
483
|
+
case qualifier
|
484
|
+
when 'translation'
|
485
|
+
value = Sequence::AA.new(value.gsub(/\s/, ''))
|
486
|
+
when 'codon_start'
|
487
|
+
value = value.to_i
|
488
|
+
end
|
489
|
+
|
490
|
+
feature.append(Feature::Qualifier.new(qualifier, value))
|
491
|
+
end
|
492
|
+
end
|
493
|
+
|
494
|
+
return feature
|
495
|
+
end
|
496
|
+
|
497
|
+
end # class EMBL
|
498
|
+
|
499
|
+
end # module Bio
|
500
|
+
|