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
data/lib/bio/io/fetch.rb
ADDED
@@ -0,0 +1,195 @@
|
|
1
|
+
#
|
2
|
+
# = bio/io/biofetch.rb - BioFetch access module
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2002, 2005 Toshiaki Katayama <k@bioruby.org>,
|
5
|
+
# Copyright (C) 2006 Jan Aerts <jan.aerts@bbsrc.ac.uk>
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id:$
|
9
|
+
#
|
10
|
+
# == DESCRIPTION
|
11
|
+
#
|
12
|
+
# Using BioRuby BioFetch server
|
13
|
+
#
|
14
|
+
# br_server = Bio::Fetch.new()
|
15
|
+
# puts br_server.databases
|
16
|
+
# puts br_server.formats('embl')
|
17
|
+
# puts br_server.maxids
|
18
|
+
#
|
19
|
+
# Using EBI BioFetch server
|
20
|
+
#
|
21
|
+
# ebi_server = Bio::Fetch.new('http://www.ebi.ac.uk/cgi-bin/dbfetch')
|
22
|
+
# puts ebi_server.fetch('embl', 'J00231', 'raw')
|
23
|
+
# puts ebi_server.fetch('embl', 'J00231', 'html')
|
24
|
+
# puts Bio::Fetch.query('genbank', 'J00231')
|
25
|
+
# puts Bio::Fetch.query('genbank', 'J00231', 'raw', 'fasta')
|
26
|
+
#
|
27
|
+
|
28
|
+
require 'uri'
|
29
|
+
require 'cgi'
|
30
|
+
require 'bio/command'
|
31
|
+
|
32
|
+
module Bio
|
33
|
+
# = DESCRIPTION
|
34
|
+
# The Bio::Fetch class provides an interface to dbfetch servers. Given
|
35
|
+
# a database name and an accession number, these servers return the associated
|
36
|
+
# record. For example, for the embl database on the EBI, that would be a
|
37
|
+
# nucleic or amino acid sequence.
|
38
|
+
#
|
39
|
+
# Possible dbfetch servers include:
|
40
|
+
# * http://bioruby.org/cgi-bin/biofetch.rb (default)
|
41
|
+
# * http://www.ebi.ac.uk/cgi-bin/dbfetch
|
42
|
+
#
|
43
|
+
# If you're behind a proxy server, be sure to set your HTTP_PROXY
|
44
|
+
# environment variable accordingly.
|
45
|
+
#
|
46
|
+
# = USAGE
|
47
|
+
# require 'bio'
|
48
|
+
#
|
49
|
+
# # Retrieve the sequence of accession number M33388 from the EMBL
|
50
|
+
# # database.
|
51
|
+
# server = Bio::Fetch.new() #uses default server
|
52
|
+
# puts server.fetch('embl','M33388')
|
53
|
+
#
|
54
|
+
# # Do the same thing without creating a Bio::Fetch object. This method always
|
55
|
+
# # uses the default dbfetch server: http://bioruby.org/cgi-bin/biofetch.rb
|
56
|
+
# puts Bio::Fetch.query('embl','M33388')
|
57
|
+
#
|
58
|
+
# # To know what databases are available on the bioruby dbfetch server:
|
59
|
+
# server = Bio::Fetch.new()
|
60
|
+
# puts server.databases
|
61
|
+
#
|
62
|
+
# # Some databases provide their data in different formats (e.g. 'fasta',
|
63
|
+
# # 'genbank' or 'embl'). To check which formats are supported by a given
|
64
|
+
# # database:
|
65
|
+
# puts server.formats('embl')
|
66
|
+
#
|
67
|
+
class Fetch
|
68
|
+
|
69
|
+
# Create a new Bio::Fetch server object that can subsequently be queried
|
70
|
+
# using the Bio::Fetch#fetch method
|
71
|
+
# ---
|
72
|
+
# *Arguments*:
|
73
|
+
# * _url_: URL of dbfetch server (default = 'http://bioruby.org/cgi-bin/biofetch.rb')
|
74
|
+
# *Returns*:: Bio::Fetch object
|
75
|
+
def initialize(url = 'http://bioruby.org/cgi-bin/biofetch.rb')
|
76
|
+
@url = url
|
77
|
+
schema, user, @host, @port, reg, @path, = URI.split(@url)
|
78
|
+
end
|
79
|
+
|
80
|
+
# The default database to query
|
81
|
+
#--
|
82
|
+
# This will be used by the get_by_id method
|
83
|
+
#++
|
84
|
+
attr_accessor :database
|
85
|
+
|
86
|
+
# Get raw database entry by id. This method lets the Bio::Registry class
|
87
|
+
# use Bio::Fetch objects.
|
88
|
+
def get_by_id(id)
|
89
|
+
fetch(@database, id)
|
90
|
+
end
|
91
|
+
|
92
|
+
# Fetch a database entry as specified by database (db), entry id (id),
|
93
|
+
# 'raw' text or 'html' (style), and format. When using BioRuby's
|
94
|
+
# BioFetch server, value for the format should not be set.
|
95
|
+
# Examples:
|
96
|
+
# server = Bio::Fetch.new('http://www.ebi.ac.uk/cgi-bin/dbfetch')
|
97
|
+
# puts server.fetch('embl','M33388','raw','fasta')
|
98
|
+
# puts server.fetch('refseq','NM_12345','html','embl')
|
99
|
+
# ---
|
100
|
+
# *Arguments*:
|
101
|
+
# * _database_: name of database to query (see Bio::Fetch#databases to get list of supported databases)
|
102
|
+
# * _id_: single ID or ID list separated by commas or white space
|
103
|
+
# * _style_: [raw|html] (default = 'raw')
|
104
|
+
# * _format_: name of output format (see Bio::Fetch#formats)
|
105
|
+
def fetch(db, id, style = 'raw', format = nil)
|
106
|
+
query = [ [ 'db', db ],
|
107
|
+
[ 'id', id ],
|
108
|
+
[ 'style', style ] ]
|
109
|
+
query.push([ 'format', format ]) if format
|
110
|
+
|
111
|
+
_get(query)
|
112
|
+
end
|
113
|
+
|
114
|
+
# Shortcut for using BioRuby's BioFetch server. You can fetch an entry
|
115
|
+
# without creating an instance of BioFetch server. This method uses the
|
116
|
+
# default dbfetch server, which is http://bioruby.org/cgi-bin/biofetch.rb
|
117
|
+
#
|
118
|
+
# Example:
|
119
|
+
# puts Bio::Fetch.query('refseq','NM_12345')
|
120
|
+
#
|
121
|
+
# ---
|
122
|
+
# *Arguments*:
|
123
|
+
# * _database_: name of database to query (see Bio::Fetch#databases to get list of supported databases)
|
124
|
+
# * _id_: single ID or ID list separated by commas or white space
|
125
|
+
# * _style_: [raw|html] (default = 'raw')
|
126
|
+
# * _format_: name of output format (see Bio::Fetch#formats)
|
127
|
+
def self.query(*args)
|
128
|
+
self.new.fetch(*args)
|
129
|
+
end
|
130
|
+
|
131
|
+
# Using this method, the user can ask a dbfetch server what databases
|
132
|
+
# it supports. This would normally be the first step you'd take when
|
133
|
+
# you use a dbfetch server for the first time.
|
134
|
+
# Example:
|
135
|
+
# server = Bio::Fetch.new()
|
136
|
+
# puts server.databases # returns "aa aax bl cpd dgenes dr ec eg emb ..."
|
137
|
+
#
|
138
|
+
# This method only works for the bioruby dbfetch server. For a list
|
139
|
+
# of databases available from the EBI, see the EBI website at
|
140
|
+
# http://www.ebi.ac.uk/cgi-bin/dbfetch/
|
141
|
+
# ---
|
142
|
+
# *Returns*:: array of database names
|
143
|
+
def databases
|
144
|
+
_get_single('info', 'dbs').strip.split(/\s+/)
|
145
|
+
end
|
146
|
+
|
147
|
+
# Lists the formats that are available for a given database. Like the
|
148
|
+
# Bio::Fetch#databases method, this method is only available on
|
149
|
+
# the bioruby dbfetch server.
|
150
|
+
# Example:
|
151
|
+
# server = Bio::Fetch.new()
|
152
|
+
# puts server.formats('embl') # returns "default fasta"
|
153
|
+
# ---
|
154
|
+
# *Arguments*:
|
155
|
+
# * _database_:: name of database you want the supported formats for
|
156
|
+
# *Returns*:: array of formats
|
157
|
+
def formats(database = @database)
|
158
|
+
if database
|
159
|
+
query = [ [ 'info', 'formats' ],
|
160
|
+
[ 'db', database ] ]
|
161
|
+
_get(query).strip.split(/\s+/)
|
162
|
+
end
|
163
|
+
end
|
164
|
+
|
165
|
+
# A dbfetch server will only return entries up to a given maximum number.
|
166
|
+
# This method retrieves that number from the server. As for the databases
|
167
|
+
# and formats methods, the maxids method only works for the bioruby
|
168
|
+
# dbfetch server.
|
169
|
+
# ---
|
170
|
+
# *Arguments*: none
|
171
|
+
# *Returns*:: number
|
172
|
+
def maxids
|
173
|
+
_get_single('info', 'maxids').to_i
|
174
|
+
end
|
175
|
+
|
176
|
+
private
|
177
|
+
# (private) query to the server.
|
178
|
+
# ary must be nested array, e.g. [ [ key0, val0 ], [ key1, val1 ], ... ]
|
179
|
+
def _get(ary)
|
180
|
+
query = ary.collect do |a|
|
181
|
+
"#{CGI.escape(a[0])}=#{CGI.escape(a[1])}"
|
182
|
+
end.join('&')
|
183
|
+
Bio::Command.read_uri(@url + '?' + query)
|
184
|
+
end
|
185
|
+
|
186
|
+
# (private) query with single parameter
|
187
|
+
def _get_single(key, val)
|
188
|
+
query = "#{CGI.escape(key)}=#{CGI.escape(val)}"
|
189
|
+
Bio::Command.read_uri(@url + '?' + query)
|
190
|
+
end
|
191
|
+
|
192
|
+
end
|
193
|
+
|
194
|
+
end # module Bio
|
195
|
+
|
@@ -0,0 +1,482 @@
|
|
1
|
+
#
|
2
|
+
# = bio/io/flatfile.rb - flatfile access wrapper class
|
3
|
+
#
|
4
|
+
# Copyright (C) 2001-2006 Naohisa Goto <ng@bioruby.org>
|
5
|
+
#
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: flatfile.rb,v 1.61 2007/11/15 07:07:16 k Exp $
|
9
|
+
#
|
10
|
+
#
|
11
|
+
# Bio::FlatFile is a helper and wrapper class to read a biological data file.
|
12
|
+
# It acts like a IO object.
|
13
|
+
# It can automatically detect data format, and users do not need to tell
|
14
|
+
# the class what the data is.
|
15
|
+
#
|
16
|
+
|
17
|
+
module Bio
|
18
|
+
|
19
|
+
# Bio::FlatFile is a helper and wrapper class to read a biological data file.
|
20
|
+
# It acts like a IO object.
|
21
|
+
# It can automatically detect data format, and users do not need to tell
|
22
|
+
# the class what the data is.
|
23
|
+
class FlatFile
|
24
|
+
|
25
|
+
autoload :AutoDetect, 'bio/io/flatfile/autodetection'
|
26
|
+
autoload :Splitter, 'bio/io/flatfile/splitter'
|
27
|
+
autoload :BufferedInputStream, 'bio/io/flatfile/buffer'
|
28
|
+
|
29
|
+
include Enumerable
|
30
|
+
|
31
|
+
#
|
32
|
+
# Bio::FlatFile.open(file, *arg)
|
33
|
+
# Bio::FlatFile.open(dbclass, file, *arg)
|
34
|
+
#
|
35
|
+
# Creates a new Bio::FlatFile object to read a file or a stream
|
36
|
+
# which contains _dbclass_ data.
|
37
|
+
#
|
38
|
+
# _dbclass_ should be a class (or module) or nil.
|
39
|
+
# e.g. Bio::GenBank, Bio::FastaFormat.
|
40
|
+
#
|
41
|
+
# If _file_ is a filename (which doesn't have gets method),
|
42
|
+
# the method opens a local file named _file_
|
43
|
+
# with <code>File.open(filename, *arg)</code>.
|
44
|
+
#
|
45
|
+
# When _dbclass_ is omitted or nil is given to _dbclass_,
|
46
|
+
# the method tries to determine database class
|
47
|
+
# (file format) automatically.
|
48
|
+
# When it fails to determine, dbclass is set to nil
|
49
|
+
# and FlatFile#next_entry would fail.
|
50
|
+
# You can still set dbclass using FlatFile#dbclass= method.
|
51
|
+
#
|
52
|
+
# * Example 1
|
53
|
+
# Bio::FlatFile.open(Bio::GenBank, "genbank/gbest40.seq")
|
54
|
+
# * Example 2
|
55
|
+
# Bio::FlatFile.open(nil, "embl/est_hum17.dat")
|
56
|
+
# * Example 3
|
57
|
+
# Bio::FlatFile.open("genbank/gbest40.seq")
|
58
|
+
#
|
59
|
+
# * Example 4
|
60
|
+
# Bio::FlatFile.open(Bio::GenBank, $stdin)
|
61
|
+
#
|
62
|
+
# If it is called with a block, the block will be executed with
|
63
|
+
# a new Bio::FlatFile object. If filename is given,
|
64
|
+
# the file is automatically closed when leaving the block.
|
65
|
+
#
|
66
|
+
# * Example 5
|
67
|
+
# Bio::FlatFile.open(nil, 'test4.fst') do |ff|
|
68
|
+
# ff.each { |e| print e.definition, "\n" }
|
69
|
+
# end
|
70
|
+
#
|
71
|
+
# * Example 6
|
72
|
+
# Bio::FlatFile.open('test4.fst') do |ff|
|
73
|
+
# ff.each { |e| print e.definition, "\n" }
|
74
|
+
# end
|
75
|
+
#
|
76
|
+
# Compatibility Note:
|
77
|
+
# <em>*arg</em> is completely passed to the <code>File.open</code>
|
78
|
+
# and you cannot specify ":raw => true" or ":raw => false".
|
79
|
+
#
|
80
|
+
def self.open(*arg, &block)
|
81
|
+
# FlatFile.open(dbclass, file, mode, perm)
|
82
|
+
# FlatFile.open(file, mode, perm)
|
83
|
+
if arg.size <= 0
|
84
|
+
raise ArgumentError, 'wrong number of arguments (0 for 1)'
|
85
|
+
end
|
86
|
+
x = arg.shift
|
87
|
+
if x.is_a?(Module) then
|
88
|
+
# FlatFile.open(dbclass, filename_or_io, ...)
|
89
|
+
dbclass = x
|
90
|
+
elsif x.nil? then
|
91
|
+
# FlatFile.open(nil, filename_or_io, ...)
|
92
|
+
dbclass = nil
|
93
|
+
else
|
94
|
+
# FlatFile.open(filename, ...)
|
95
|
+
dbclass = nil
|
96
|
+
arg.unshift(x)
|
97
|
+
end
|
98
|
+
if arg.size <= 0
|
99
|
+
raise ArgumentError, 'wrong number of arguments (1 for 2)'
|
100
|
+
end
|
101
|
+
file = arg.shift
|
102
|
+
# check if file is filename or IO object
|
103
|
+
unless file.respond_to?(:gets)
|
104
|
+
# 'file' is a filename
|
105
|
+
_open_file(dbclass, file, *arg, &block)
|
106
|
+
else
|
107
|
+
# 'file' is a IO object
|
108
|
+
ff = self.new(dbclass, file)
|
109
|
+
block_given? ? (yield ff) : ff
|
110
|
+
end
|
111
|
+
end
|
112
|
+
|
113
|
+
# Same as Bio::FlatFile.open(nil, filename_or_stream, mode, perm, options).
|
114
|
+
#
|
115
|
+
# * Example 1
|
116
|
+
# Bio::FlatFile.auto(ARGF)
|
117
|
+
# * Example 2
|
118
|
+
# Bio::FlatFile.auto("embl/est_hum17.dat")
|
119
|
+
# * Example 3
|
120
|
+
# Bio::FlatFile.auto(IO.popen("gzip -dc nc1101.flat.gz"))
|
121
|
+
#
|
122
|
+
def self.auto(*arg, &block)
|
123
|
+
self.open(nil, *arg, &block)
|
124
|
+
end
|
125
|
+
|
126
|
+
# Same as FlatFile.auto(filename_or_stream, *arg).to_a
|
127
|
+
#
|
128
|
+
# (This method might be OBSOLETED in the future.)
|
129
|
+
def self.to_a(*arg)
|
130
|
+
self.auto(*arg) do |ff|
|
131
|
+
raise 'cannot determine file format' unless ff.dbclass
|
132
|
+
ff.to_a
|
133
|
+
end
|
134
|
+
end
|
135
|
+
|
136
|
+
# Same as FlatFile.auto(filename, *arg),
|
137
|
+
# except that it only accept filename and doesn't accept IO object.
|
138
|
+
# File format is automatically determined.
|
139
|
+
#
|
140
|
+
# It can accept a block.
|
141
|
+
# If a block is given, it returns the block's return value.
|
142
|
+
# Otherwise, it returns a new FlatFile object.
|
143
|
+
#
|
144
|
+
def self.open_file(filename, *arg)
|
145
|
+
_open_file(nil, filename, *arg)
|
146
|
+
end
|
147
|
+
|
148
|
+
# Same as FlatFile.open(dbclass, filename, *arg),
|
149
|
+
# except that it only accept filename and doesn't accept IO object.
|
150
|
+
#
|
151
|
+
# It can accept a block.
|
152
|
+
# If a block is given, it returns the block's return value.
|
153
|
+
# Otherwise, it returns a new FlatFile object.
|
154
|
+
#
|
155
|
+
def self._open_file(dbclass, filename, *arg)
|
156
|
+
if block_given? then
|
157
|
+
BufferedInputStream.open_file(filename, *arg) do |stream|
|
158
|
+
yield self.new(dbclass, stream)
|
159
|
+
end
|
160
|
+
else
|
161
|
+
stream = BufferedInputStream.open_file(filename, *arg)
|
162
|
+
self.new(dbclass, stream)
|
163
|
+
end
|
164
|
+
end
|
165
|
+
private_class_method :_open_file
|
166
|
+
|
167
|
+
# Opens URI specified as _uri_.
|
168
|
+
# _uri_ must be a String or URI object.
|
169
|
+
# *arg is passed to OpenURI.open_uri or URI#open.
|
170
|
+
#
|
171
|
+
# Like FlatFile#open, it can accept a block.
|
172
|
+
#
|
173
|
+
# Note that you MUST explicitly require 'open-uri'.
|
174
|
+
# Because open-uri.rb modifies existing class,
|
175
|
+
# it isn't required by default.
|
176
|
+
#
|
177
|
+
def self.open_uri(uri, *arg)
|
178
|
+
if block_given? then
|
179
|
+
BufferedInputStream.open_uri(uri, *arg) do |stream|
|
180
|
+
yield self.new(nil, stream)
|
181
|
+
end
|
182
|
+
else
|
183
|
+
stream = BufferedInputStream.open_uri(uri, *arg)
|
184
|
+
self.new(nil, stream)
|
185
|
+
end
|
186
|
+
end
|
187
|
+
|
188
|
+
# Executes the block for every entry in the stream.
|
189
|
+
# Same as FlatFile.open(*arg) { |ff| ff.each { |entry| ... }}.
|
190
|
+
#
|
191
|
+
# * Example
|
192
|
+
# Bio::FlatFile.foreach('test.fst') { |e| puts e.definition }
|
193
|
+
#
|
194
|
+
def self.foreach(*arg)
|
195
|
+
self.open(*arg) do |flatfileobj|
|
196
|
+
flatfileobj.each do |entry|
|
197
|
+
yield entry
|
198
|
+
end
|
199
|
+
end
|
200
|
+
end
|
201
|
+
|
202
|
+
# Same as FlatFile.open, except that 'stream' should be a opened
|
203
|
+
# stream object (IO, File, ..., who have the 'gets' method).
|
204
|
+
#
|
205
|
+
# * Example 1
|
206
|
+
# Bio::FlatFile.new(Bio::GenBank, ARGF)
|
207
|
+
# * Example 2
|
208
|
+
# Bio::FlatFile.new(Bio::GenBank, IO.popen("gzip -dc nc1101.flat.gz"))
|
209
|
+
#
|
210
|
+
# Compatibility Note:
|
211
|
+
# Now, you cannot specify ":raw => true" or ":raw => false".
|
212
|
+
# Below styles are DEPRECATED.
|
213
|
+
#
|
214
|
+
# * Example 3 (deprecated)
|
215
|
+
# # Bio::FlatFile.new(nil, $stdin, :raw=>true) # => ERROR
|
216
|
+
# # Please rewrite as below.
|
217
|
+
# ff = Bio::FlatFile.new(nil, $stdin)
|
218
|
+
# ff.raw = true
|
219
|
+
# * Example 3 in old style (deprecated)
|
220
|
+
# # Bio::FlatFile.new(nil, $stdin, true) # => ERROR
|
221
|
+
# # Please rewrite as below.
|
222
|
+
# ff = Bio::FlatFile.new(nil, $stdin)
|
223
|
+
# ff.raw = true
|
224
|
+
#
|
225
|
+
def initialize(dbclass, stream)
|
226
|
+
# 2nd arg: IO object
|
227
|
+
if stream.kind_of?(BufferedInputStream)
|
228
|
+
@stream = stream
|
229
|
+
else
|
230
|
+
@stream = BufferedInputStream.for_io(stream)
|
231
|
+
end
|
232
|
+
# 1st arg: database class (or file format autodetection)
|
233
|
+
if dbclass then
|
234
|
+
self.dbclass = dbclass
|
235
|
+
else
|
236
|
+
autodetect
|
237
|
+
end
|
238
|
+
#
|
239
|
+
@skip_leader_mode = :firsttime
|
240
|
+
@firsttime_flag = true
|
241
|
+
# default raw mode is false
|
242
|
+
self.raw = false
|
243
|
+
end
|
244
|
+
|
245
|
+
# The mode how to skip leader of the data.
|
246
|
+
# :firsttime :: (DEFAULT) only head of file (= first time to read)
|
247
|
+
# :everytime :: everytime to read entry
|
248
|
+
# nil :: never skip
|
249
|
+
attr_accessor :skip_leader_mode
|
250
|
+
|
251
|
+
# (DEPRECATED) IO object in the flatfile object.
|
252
|
+
#
|
253
|
+
# Compatibility Note: Bio::FlatFile#io is deprecated.
|
254
|
+
# Please use Bio::FlatFile#to_io instead.
|
255
|
+
def io
|
256
|
+
warn "Bio::FlatFile#io is deprecated."
|
257
|
+
@stream.to_io
|
258
|
+
end
|
259
|
+
|
260
|
+
# IO object in the flatfile object.
|
261
|
+
#
|
262
|
+
# Compatibility Note: Bio::FlatFile#io is deprecated.
|
263
|
+
def to_io
|
264
|
+
@stream.to_io
|
265
|
+
end
|
266
|
+
|
267
|
+
# Pathname, filename or URI (or nil).
|
268
|
+
def path
|
269
|
+
@stream.path
|
270
|
+
end
|
271
|
+
|
272
|
+
# Exception class to be raised when data format hasn't been specified.
|
273
|
+
class UnknownDataFormatError < IOError
|
274
|
+
end
|
275
|
+
|
276
|
+
# Get next entry.
|
277
|
+
def next_entry
|
278
|
+
raise UnknownDataFormatError,
|
279
|
+
'file format auto-detection failed?' unless @dbclass
|
280
|
+
if @skip_leader_mode and
|
281
|
+
((@firsttime_flag and @skip_leader_mode == :firsttime) or
|
282
|
+
@skip_leader_mode == :everytime)
|
283
|
+
@splitter.skip_leader
|
284
|
+
end
|
285
|
+
if raw then
|
286
|
+
r = @splitter.get_entry
|
287
|
+
else
|
288
|
+
r = @splitter.get_parsed_entry
|
289
|
+
end
|
290
|
+
@firsttime_flag = false
|
291
|
+
return nil unless r
|
292
|
+
if raw then
|
293
|
+
r
|
294
|
+
else
|
295
|
+
@entry = r
|
296
|
+
@entry
|
297
|
+
end
|
298
|
+
end
|
299
|
+
attr_reader :entry
|
300
|
+
|
301
|
+
# Returns the last raw entry as a string.
|
302
|
+
def entry_raw
|
303
|
+
@splitter.entry
|
304
|
+
end
|
305
|
+
|
306
|
+
# a flag to write down entry start and end positions
|
307
|
+
def entry_pos_flag
|
308
|
+
@splitter.entry_pos_flag
|
309
|
+
end
|
310
|
+
|
311
|
+
# Sets flag to write down entry start and end positions
|
312
|
+
def entry_pos_flag=(x)
|
313
|
+
@splitter.entry_pos_flag = x
|
314
|
+
end
|
315
|
+
|
316
|
+
# start position of the last entry
|
317
|
+
def entry_start_pos
|
318
|
+
@splitter.entry_start_pos
|
319
|
+
end
|
320
|
+
|
321
|
+
# (end position of the last entry) + 1
|
322
|
+
def entry_ended_pos
|
323
|
+
@splitter.entry_ended_pos
|
324
|
+
end
|
325
|
+
|
326
|
+
# Iterates over each entry in the flatfile.
|
327
|
+
#
|
328
|
+
# * Example
|
329
|
+
# include Bio
|
330
|
+
# ff = FlatFile.open(GenBank, "genbank/gbhtg14.seq")
|
331
|
+
# ff.each_entry do |x|
|
332
|
+
# puts x.definition
|
333
|
+
# end
|
334
|
+
def each_entry
|
335
|
+
while e = self.next_entry
|
336
|
+
yield e
|
337
|
+
end
|
338
|
+
end
|
339
|
+
alias :each :each_entry
|
340
|
+
|
341
|
+
# Resets file pointer to the start of the flatfile.
|
342
|
+
# (similar to IO#rewind)
|
343
|
+
def rewind
|
344
|
+
r = (@splitter || @stream).rewind
|
345
|
+
@firsttime_flag = true
|
346
|
+
r
|
347
|
+
end
|
348
|
+
|
349
|
+
# Closes input stream.
|
350
|
+
# (similar to IO#close)
|
351
|
+
def close
|
352
|
+
@stream.close
|
353
|
+
end
|
354
|
+
|
355
|
+
# Returns current position of input stream.
|
356
|
+
# If the input stream is not a normal file,
|
357
|
+
# the result is not guaranteed.
|
358
|
+
# It is similar to IO#pos.
|
359
|
+
# Note that it will not be equal to io.pos,
|
360
|
+
# because FlatFile has its own internal buffer.
|
361
|
+
def pos
|
362
|
+
@stream.pos
|
363
|
+
end
|
364
|
+
|
365
|
+
# (Not recommended to use it.)
|
366
|
+
# Sets position of input stream.
|
367
|
+
# If the input stream is not a normal file,
|
368
|
+
# the result is not guaranteed.
|
369
|
+
# It is similar to IO#pos=.
|
370
|
+
# Note that it will not be equal to io.pos=,
|
371
|
+
# because FlatFile has its own internal buffer.
|
372
|
+
def pos=(p)
|
373
|
+
@stream.pos=(p)
|
374
|
+
end
|
375
|
+
|
376
|
+
# Returns true if input stream is end-of-file.
|
377
|
+
# Otherwise, returns false.
|
378
|
+
# (Similar to IO#eof?, but may not be equal to io.eof?,
|
379
|
+
# because FlatFile has its own internal buffer.)
|
380
|
+
def eof?
|
381
|
+
@stream.eof?
|
382
|
+
end
|
383
|
+
|
384
|
+
# If true is given, the next_entry method returns
|
385
|
+
# a entry as a text, whereas if false, returns as a parsed object.
|
386
|
+
def raw=(bool)
|
387
|
+
@raw = (bool ? true : false)
|
388
|
+
end
|
389
|
+
|
390
|
+
# If true, raw mode.
|
391
|
+
attr_reader :raw
|
392
|
+
|
393
|
+
# Similar to IO#gets.
|
394
|
+
# Internal use only. Users should not call it directly.
|
395
|
+
def gets(*arg)
|
396
|
+
@stream.gets(*arg)
|
397
|
+
end
|
398
|
+
|
399
|
+
# Sets database class. Plese use only if autodetect fails.
|
400
|
+
def dbclass=(klass)
|
401
|
+
if klass then
|
402
|
+
@dbclass = klass
|
403
|
+
begin
|
404
|
+
@splitter = @dbclass.flatfile_splitter(@dbclass, @stream)
|
405
|
+
rescue NameError, NoMethodError
|
406
|
+
begin
|
407
|
+
splitter_class = @dbclass::FLATFILE_SPLITTER
|
408
|
+
rescue NameError
|
409
|
+
splitter_class = Splitter::Default
|
410
|
+
end
|
411
|
+
@splitter = splitter_class.new(klass, @stream)
|
412
|
+
end
|
413
|
+
else
|
414
|
+
@dbclass = nil
|
415
|
+
@splitter = nil
|
416
|
+
end
|
417
|
+
end
|
418
|
+
|
419
|
+
# Returns database class which is automatically detected or
|
420
|
+
# given in FlatFile#initialize.
|
421
|
+
attr_reader :dbclass
|
422
|
+
|
423
|
+
# Performs determination of database class (file format).
|
424
|
+
# Pre-reads +lines+ lines for format determination (default 31 lines).
|
425
|
+
# If fails, returns nil or false. Otherwise, returns database class.
|
426
|
+
#
|
427
|
+
# The method can be called anytime if you want (but not recommended).
|
428
|
+
# This might be useful if input file is a mixture of muitiple format data.
|
429
|
+
def autodetect(lines = 31, ad = AutoDetect.default)
|
430
|
+
if r = ad.autodetect_flatfile(self, lines)
|
431
|
+
self.dbclass = r
|
432
|
+
else
|
433
|
+
self.dbclass = nil unless self.dbclass
|
434
|
+
end
|
435
|
+
r
|
436
|
+
end
|
437
|
+
|
438
|
+
# Detects database class (== file format) of given file.
|
439
|
+
# If fails to determine, returns nil.
|
440
|
+
def self.autodetect_file(filename)
|
441
|
+
self.open_file(filename).dbclass
|
442
|
+
end
|
443
|
+
|
444
|
+
# Detects database class (== file format) of given input stream.
|
445
|
+
# If fails to determine, returns nil.
|
446
|
+
# Caution: the method reads some data from the input stream,
|
447
|
+
# and the data will be lost.
|
448
|
+
def self.autodetect_io(io)
|
449
|
+
self.new(nil, io).dbclass
|
450
|
+
end
|
451
|
+
|
452
|
+
# This is OBSOLETED. Please use autodetect_io(io) instead.
|
453
|
+
def self.autodetect_stream(io)
|
454
|
+
$stderr.print "Bio::FlatFile.autodetect_stream will be deprecated." if $VERBOSE
|
455
|
+
self.autodetect_io(io)
|
456
|
+
end
|
457
|
+
|
458
|
+
# Detects database class (== file format) of given string.
|
459
|
+
# If fails to determine, returns false or nil.
|
460
|
+
def self.autodetect(text)
|
461
|
+
AutoDetect.default.autodetect(text)
|
462
|
+
end
|
463
|
+
|
464
|
+
# Return a hash of the entries, where the names of the
|
465
|
+
def to_hash(indexing_method = :entry_id)
|
466
|
+
hash = {}
|
467
|
+
each do |entry|
|
468
|
+
hash[entry.send(indexing_method)] = entry
|
469
|
+
end
|
470
|
+
hash
|
471
|
+
end
|
472
|
+
|
473
|
+
end #class FlatFile
|
474
|
+
|
475
|
+
end #module Bio
|
476
|
+
|
477
|
+
if __FILE__ == $0
|
478
|
+
if ARGV.size == 2
|
479
|
+
require 'bio'
|
480
|
+
p Bio::FlatFile.open(eval(ARGV.shift), ARGV.shift).next_entry
|
481
|
+
end
|
482
|
+
end
|