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 +1,226 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/mafft/report.rb - MAFFT report class
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2003, 2007 Naohisa Goto <ng@bioruby.org>
|
5
|
+
# License:: The Ruby License
|
6
|
+
#
|
7
|
+
# $Id: report.rb,v 1.13 2007/07/16 12:21:39 ngoto Exp $
|
8
|
+
#
|
9
|
+
# MAFFT result parser class.
|
10
|
+
# MAFFT is a very fast multiple sequence alignment software.
|
11
|
+
#
|
12
|
+
# Since a result of MAFFT is simply a multiple-fasta format,
|
13
|
+
# the significance of this class is to keep standard form and
|
14
|
+
# interface between Bio::ClustalW::Report.
|
15
|
+
#
|
16
|
+
# Bio::Alignment::MultiFastaFormat is a generic data class for
|
17
|
+
# fasta-formatted multiple sequence alignment data.
|
18
|
+
# Bio::MAFFT::Report inherits Bio::Alignment::MultiFastaFormat.
|
19
|
+
#
|
20
|
+
# == References
|
21
|
+
#
|
22
|
+
# * K. Katoh, K. Misawa, K. Kuma and T. Miyata.
|
23
|
+
# MAFFT: a novel method for rapid multiple sequence alignment based
|
24
|
+
# on fast Fourier transform. Nucleic Acids Res. 30: 3059-3066, 2002.
|
25
|
+
# http://nar.oupjournals.org/cgi/content/abstract/30/14/3059
|
26
|
+
# * http://www.biophys.kyoto-u.ac.jp/~katoh/programs/align/mafft/
|
27
|
+
#
|
28
|
+
|
29
|
+
require 'stringio'
|
30
|
+
require 'bio/db/fasta'
|
31
|
+
require 'bio/io/flatfile'
|
32
|
+
require 'bio/alignment'
|
33
|
+
require 'bio/appl/mafft'
|
34
|
+
|
35
|
+
module Bio
|
36
|
+
module Alignment
|
37
|
+
# Data class for fasta-formatted multiple sequence alignment data,
|
38
|
+
# which is simply multiple entiries of fasta formatted sequences.
|
39
|
+
class MultiFastaFormat
|
40
|
+
|
41
|
+
# delimiter for flatfile
|
42
|
+
DELIMITER = RS = nil
|
43
|
+
|
44
|
+
# Creates a new data object.
|
45
|
+
# +str+ should be a (multi-)fasta formatted string.
|
46
|
+
def initialize(str)
|
47
|
+
ff = Bio::FlatFile.new(Bio::FastaFormat, StringIO.new(str))
|
48
|
+
@data = ff.to_a
|
49
|
+
@alignment = nil
|
50
|
+
@seq_method = nil
|
51
|
+
end
|
52
|
+
|
53
|
+
# Gets an multiple alignment.
|
54
|
+
# Returns a Bio::Alignment object.
|
55
|
+
# +method+ should be one of :naseq, :aaseq, :seq, or nil (default).
|
56
|
+
# nil means to automatically determine nucleotide or amino acid.
|
57
|
+
#
|
58
|
+
# This method returns previously parsed object
|
59
|
+
# if the same method is given (or guessed method is the same).
|
60
|
+
def alignment(method = nil)
|
61
|
+
m = determine_seq_method(@data, method)
|
62
|
+
if !@alignment or m != @seq_method then
|
63
|
+
@seq_method = m
|
64
|
+
@alignment = do_parse(@data, @seq_method)
|
65
|
+
end
|
66
|
+
@alignment
|
67
|
+
end
|
68
|
+
|
69
|
+
# Gets an array of the fasta formatted sequence objects.
|
70
|
+
# Returns an array of Bio::FastaFormat objects.
|
71
|
+
def entries
|
72
|
+
@data
|
73
|
+
end
|
74
|
+
|
75
|
+
private
|
76
|
+
# determines seqtype.
|
77
|
+
# if nil is given, try to guess DNA or protein.
|
78
|
+
def determine_seq_method(data, m = nil)
|
79
|
+
case m
|
80
|
+
when :aaseq
|
81
|
+
:aaseq
|
82
|
+
when :naseq
|
83
|
+
:naseq
|
84
|
+
when :seq
|
85
|
+
:seq
|
86
|
+
when nil
|
87
|
+
# auto-detection
|
88
|
+
score = 0
|
89
|
+
data[0, 3].each do |e|
|
90
|
+
k = e.to_seq.guess
|
91
|
+
if k == Bio::Sequence::NA then
|
92
|
+
score += 1
|
93
|
+
elsif k == Bio::Sequence::AA then
|
94
|
+
score -= 1
|
95
|
+
end
|
96
|
+
end
|
97
|
+
if score > 0 then
|
98
|
+
:naseq
|
99
|
+
elsif score < 0 then
|
100
|
+
:aaseq
|
101
|
+
else
|
102
|
+
:seq
|
103
|
+
end
|
104
|
+
else
|
105
|
+
raise 'one of :naseq, :aaseq, :seq, or nil should be given'
|
106
|
+
end
|
107
|
+
end
|
108
|
+
|
109
|
+
# Parses a result.
|
110
|
+
def do_parse(ary, seqmethod)
|
111
|
+
a = Bio::Alignment.new
|
112
|
+
a.add_sequences(ary) do |x|
|
113
|
+
[ x.__send__(seqmethod), x.definition ]
|
114
|
+
end
|
115
|
+
a
|
116
|
+
end
|
117
|
+
end #class MultiFastaFormat
|
118
|
+
end #module Alignment
|
119
|
+
|
120
|
+
class MAFFT
|
121
|
+
|
122
|
+
# MAFFT result parser class.
|
123
|
+
# MAFFT is a very fast multiple sequence alignment software.
|
124
|
+
#
|
125
|
+
# Since a result of MAFFT is simply a multiple-fasta format,
|
126
|
+
# the significance of this class is to keep standard form and
|
127
|
+
# interface between Bio::ClustalW::Report.
|
128
|
+
class Report < Bio::Alignment::MultiFastaFormat
|
129
|
+
|
130
|
+
# Creates a new Report object.
|
131
|
+
# +str+ should be multi-fasta formatted text as a string.
|
132
|
+
#
|
133
|
+
# Compatibility Note: the old usage (to get array of Bio::FastaFormat
|
134
|
+
# objects) is deprecated.
|
135
|
+
#
|
136
|
+
# Compatibility Note 2: the argument +seqclass+ is deprecated.
|
137
|
+
#
|
138
|
+
# +seqclass+ should be one of following:
|
139
|
+
# Class: Bio::Sequence::AA, Bio::Sequence::NA, ...
|
140
|
+
# String: 'PROTEIN', 'DNA', ...
|
141
|
+
#
|
142
|
+
def initialize(str, seqclass = nil)
|
143
|
+
if str.is_a?(Array) then
|
144
|
+
warn "Array of Bio::FastaFormat objects will be no longer accepted."
|
145
|
+
@data = str
|
146
|
+
else
|
147
|
+
super(str)
|
148
|
+
end
|
149
|
+
|
150
|
+
if seqclass then
|
151
|
+
warn "the 2nd argument (seqclass) will be no deprecated."
|
152
|
+
case seqclass
|
153
|
+
when /PROTEIN/i
|
154
|
+
@seqclass = Bio::Sequence::AA
|
155
|
+
when /[DR]NA/i
|
156
|
+
@seqclass = Bio::Sequence::NA
|
157
|
+
else
|
158
|
+
if seqclass.is_a?(Module) then
|
159
|
+
@seqclass = seqclass
|
160
|
+
else
|
161
|
+
@seqclass = nil
|
162
|
+
end
|
163
|
+
end
|
164
|
+
end
|
165
|
+
end
|
166
|
+
|
167
|
+
# sequence data. Returns an array of Bio::FastaFormat.
|
168
|
+
attr_reader :data
|
169
|
+
|
170
|
+
# Sequence class (Bio::Sequence::AA, Bio::Sequence::NA, ...)
|
171
|
+
#
|
172
|
+
# Compatibility note: This method will be removed in the tufure.
|
173
|
+
attr_reader :seqclass
|
174
|
+
|
175
|
+
# Gets an multiple alignment.
|
176
|
+
# Returns a Bio::Alignment object.
|
177
|
+
def alignment(method = nil)
|
178
|
+
super
|
179
|
+
end
|
180
|
+
|
181
|
+
# This method will be deprecated. Instead, please use alignment.
|
182
|
+
#
|
183
|
+
# Gets an multiple alignment.
|
184
|
+
# Returns a Bio::Alignment object.
|
185
|
+
def align
|
186
|
+
warn "Bio::MAFFT::Report#align is deprecated. Please use \'alignment\'."
|
187
|
+
alignment
|
188
|
+
end
|
189
|
+
|
190
|
+
# This will be deprecated. Instead, please use alignment.output_fasta.
|
191
|
+
#
|
192
|
+
# Gets an fasta-format string of the sequences.
|
193
|
+
# Returns a string.
|
194
|
+
# Same as align.to_fasta.
|
195
|
+
# Please refer to Bio::Alignment#output_fasta for arguments.
|
196
|
+
def to_fasta(*arg)
|
197
|
+
warn "Bio::MAFFT::report#to_fasta is deprecated. Please use \'alignment.output_fasta\'"
|
198
|
+
alignment.output_fasta(*arg)
|
199
|
+
end
|
200
|
+
|
201
|
+
# Compatibility note: Behavior of the method will be changed
|
202
|
+
# in the future.
|
203
|
+
#
|
204
|
+
# Gets an array of the sequences.
|
205
|
+
# Returns an array of Bio::FastaFormat instances.
|
206
|
+
def to_a
|
207
|
+
@data
|
208
|
+
end
|
209
|
+
|
210
|
+
private
|
211
|
+
# Parsing a result.
|
212
|
+
def do_parse(ary, seqmethod)
|
213
|
+
if @seqclass then
|
214
|
+
a = Bio::Alignment.new
|
215
|
+
a.add_sequences(ary) do |x|
|
216
|
+
[ @seqclass.new(x.seq), x.definition ]
|
217
|
+
end
|
218
|
+
else
|
219
|
+
super(ary, seqmethod)
|
220
|
+
end
|
221
|
+
end
|
222
|
+
|
223
|
+
end #class Report
|
224
|
+
end #class MAFFT
|
225
|
+
end #module Bio
|
226
|
+
|
@@ -0,0 +1,52 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/muscle.rb - MUSCLE application wrapper class
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2006-2007
|
5
|
+
# Jeffrey Blakeslee and John Conery University of Oregon <jeffb@uoregon.edu>
|
6
|
+
# Naohisa Goto <ng@bioruby.org>
|
7
|
+
# License:: The Ruby License
|
8
|
+
#
|
9
|
+
# $Id: muscle.rb,v 1.1 2007/07/16 12:25:50 ngoto Exp $
|
10
|
+
#
|
11
|
+
#
|
12
|
+
# Bio::Muscle is a wrapper class to execute MUSCLE.
|
13
|
+
#
|
14
|
+
# == References
|
15
|
+
#
|
16
|
+
# * http://www.drive5.com/muscle/
|
17
|
+
# * Edgar R.C.
|
18
|
+
# MUSCLE: multiple sequence alignment with high accuracy and
|
19
|
+
# high throughput. Nucleic Acids Res. 32: 1792-1797, 2004.
|
20
|
+
# * Edgar, R.C.
|
21
|
+
# MUSCLE: a multiple sequence alignment method with reduced time
|
22
|
+
# and space complexity. BMC Bioinformatics 5: 113, 2004.
|
23
|
+
#
|
24
|
+
|
25
|
+
module Bio
|
26
|
+
|
27
|
+
# Bio::Muscle is a wrapper class to execute MUSCLE.
|
28
|
+
#
|
29
|
+
# Please refer documents in bio/apple/muscle.rb for references.
|
30
|
+
class Muscle < Bio::Alignment::FactoryTemplate::StdinInFileOut
|
31
|
+
|
32
|
+
# default program name
|
33
|
+
DEFAULT_PROGRAM = 'muscle'.freeze
|
34
|
+
|
35
|
+
# default report parser
|
36
|
+
DEFAULT_PARSER = Bio::Alignment::MultiFastaFormat
|
37
|
+
|
38
|
+
private
|
39
|
+
# generates options specifying input filename.
|
40
|
+
# returns an array of string
|
41
|
+
def _option_input_file(fn)
|
42
|
+
[ '-in', fn ]
|
43
|
+
end
|
44
|
+
|
45
|
+
# generates options specifying output filename.
|
46
|
+
# returns an array of string
|
47
|
+
def _option_output_file(fn)
|
48
|
+
[ '-out', fn ]
|
49
|
+
end
|
50
|
+
end #class Muscle
|
51
|
+
|
52
|
+
end #module Bio
|
@@ -0,0 +1,95 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/paml/baseml.rb - Wrapper for running PAML program baseml
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2008
|
5
|
+
# Naohisa Goto <ng@bioruby.org>
|
6
|
+
#
|
7
|
+
# License:: The Ruby License
|
8
|
+
#
|
9
|
+
# == Description
|
10
|
+
#
|
11
|
+
# This file contains Bio::PAML::Baseml, a wrapper class running baseml.
|
12
|
+
#
|
13
|
+
# == References
|
14
|
+
#
|
15
|
+
# * http://abacus.gene.ucl.ac.uk/software/paml.html
|
16
|
+
#
|
17
|
+
|
18
|
+
require 'bio/appl/paml/common'
|
19
|
+
|
20
|
+
module Bio::PAML
|
21
|
+
|
22
|
+
# == Description
|
23
|
+
#
|
24
|
+
# Bio::PAML::Baseml is a wrapper for running PAML baseml program.
|
25
|
+
#
|
26
|
+
# Because most of the methods in this class are inherited from
|
27
|
+
# Bio::PAML::Common, see documents of Bio::PAML::Common for details.
|
28
|
+
#
|
29
|
+
# == Examples
|
30
|
+
#
|
31
|
+
# Example 1:
|
32
|
+
#
|
33
|
+
# require 'bio'
|
34
|
+
# # Reads multi-fasta formatted file and gets a Bio::Alignment object.
|
35
|
+
# alignment = Bio::FlatFile.open(Bio::Alignment::MultiFastaFormat,
|
36
|
+
# 'example.fst').alignment
|
37
|
+
# # Reads newick tree from a file
|
38
|
+
# tree = Bio::FlatFile.open(Bio::Newick, 'example.tree').tree
|
39
|
+
# # Creates a Baseml object
|
40
|
+
# baseml = Bio::PAML::Baseml.new
|
41
|
+
# # Sets parameters
|
42
|
+
# baseml.parameters[:runmode] = 0
|
43
|
+
# baseml.parameters[:RateAncestor] = 1
|
44
|
+
# # You can also set many parameters at a time.
|
45
|
+
# baseml.parameters.update({ :alpha => 0.5, :fix_alpha => 0 })
|
46
|
+
# # Executes baseml with the alignment and the tree
|
47
|
+
# report = baseml.query(alignment, tree)
|
48
|
+
#
|
49
|
+
class Baseml < Common
|
50
|
+
|
51
|
+
autoload :Report, 'bio/appl/paml/baseml/report'
|
52
|
+
|
53
|
+
# Default program name
|
54
|
+
DEFAULT_PROGRAM = 'baseml'.freeze
|
55
|
+
|
56
|
+
# Default parameters when running baseml.
|
57
|
+
#
|
58
|
+
# The parameters whose values are different from the baseml defalut
|
59
|
+
# value (described in pamlDOC.pdf) in PAML 4.1 are:
|
60
|
+
# seqfile, outfile, treefile, ndata, noisy, verbose
|
61
|
+
#
|
62
|
+
DEFAULT_PARAMETERS = {
|
63
|
+
# Essential argumemts
|
64
|
+
:seqfile => nil,
|
65
|
+
:outfile => nil,
|
66
|
+
# Optional arguments
|
67
|
+
:treefile => nil,
|
68
|
+
:noisy => 0,
|
69
|
+
:verbose => 1,
|
70
|
+
:runmode => 0,
|
71
|
+
:model => 5,
|
72
|
+
:Mgene => 0,
|
73
|
+
:ndata => 1,
|
74
|
+
:clock => 0,
|
75
|
+
:fix_kappa => 0,
|
76
|
+
:kappa => 2.5,
|
77
|
+
:fix_alpha => 1,
|
78
|
+
:alpha => 0.0,
|
79
|
+
:Malpha => 0,
|
80
|
+
:ncatG => 5,
|
81
|
+
:fix_rho => 1,
|
82
|
+
:rho => 0.0,
|
83
|
+
:nparK => 0,
|
84
|
+
:nhomo => 0,
|
85
|
+
:getSE => 0,
|
86
|
+
:RateAncestor => 0,
|
87
|
+
:Small_Diff => 1e-6,
|
88
|
+
:cleandata => 1,
|
89
|
+
:fix_blength => 0,
|
90
|
+
:method => 0
|
91
|
+
}
|
92
|
+
|
93
|
+
end #class Baseml
|
94
|
+
end #module Bio::PAML
|
95
|
+
|
@@ -0,0 +1,32 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/paml/baseml/report.rb - parser class for PAML program baseml
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2008
|
5
|
+
# Naohisa Goto <ng@bioruby.org>
|
6
|
+
#
|
7
|
+
# License:: The Ruby License
|
8
|
+
#
|
9
|
+
# == Description
|
10
|
+
#
|
11
|
+
# This file contains Bio::PAML::Baseml::Report, a parser class for a result
|
12
|
+
# of baseml.
|
13
|
+
#
|
14
|
+
# == References
|
15
|
+
#
|
16
|
+
# * http://abacus.gene.ucl.ac.uk/software/paml.html
|
17
|
+
#
|
18
|
+
|
19
|
+
require 'bio/appl/paml/baseml'
|
20
|
+
|
21
|
+
module Bio::PAML
|
22
|
+
class Baseml
|
23
|
+
|
24
|
+
# UNDER CONSTRUCTION.
|
25
|
+
#
|
26
|
+
# Bio::PAML::Baseml::Report is a parser class for a baseml result.
|
27
|
+
#
|
28
|
+
class Report < Bio::PAML::Common::Report
|
29
|
+
end #class Report
|
30
|
+
|
31
|
+
end #class Baseml
|
32
|
+
end #module Bio::PAML
|
@@ -0,0 +1,242 @@
|
|
1
|
+
#
|
2
|
+
# = bio/appl/paml/codeml.rb - Wrapper for running PAML program codeml
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2008
|
5
|
+
# Michael D. Barton <mail@michaelbarton.me.uk>,
|
6
|
+
# Naohisa Goto <ng@bioruby.org>
|
7
|
+
#
|
8
|
+
# License:: The Ruby License
|
9
|
+
#
|
10
|
+
# == Description
|
11
|
+
#
|
12
|
+
# This file contains a wrapper for running the CODEML tool for estimating evolutionary rate
|
13
|
+
#
|
14
|
+
# == References
|
15
|
+
#
|
16
|
+
# * http://abacus.gene.ucl.ac.uk/software/paml.html
|
17
|
+
#
|
18
|
+
|
19
|
+
require 'bio/appl/paml/common'
|
20
|
+
|
21
|
+
module Bio
|
22
|
+
module PAML
|
23
|
+
|
24
|
+
# == Description
|
25
|
+
#
|
26
|
+
# Bio::PAML::Codeml is a wrapper for estimating evolutionary rate using the CODEML
|
27
|
+
# tool. The class provides methods for generating the necessary configuration
|
28
|
+
# file, and running codeml with the specified binary. Codeml output is
|
29
|
+
# returned when codeml is run. Bio::PAML::Codeml::Report and Bio::PAML::Codeml::Rates
|
30
|
+
# provide simple classes for parsing and accessing the Codeml report and
|
31
|
+
# rates files respectively.
|
32
|
+
#
|
33
|
+
# == Examples
|
34
|
+
#
|
35
|
+
# Example 1:
|
36
|
+
#
|
37
|
+
# require 'bio'
|
38
|
+
# # Reads multi-fasta formatted file and gets a Bio::Alignment object.
|
39
|
+
# alignment = Bio::FlatFile.open(Bio::Alignment::MultiFastaFormat,
|
40
|
+
# 'example.fst').alignment
|
41
|
+
# # Reads newick tree from a file
|
42
|
+
# tree = Bio::FlatFile.open(Bio::Newick, 'example.tree').tree
|
43
|
+
# # Creates a Codeml object
|
44
|
+
# codeml = Bio::PAML::Codeml.new
|
45
|
+
# # Sets parameters
|
46
|
+
# codeml.parameters[:runmode] = 0
|
47
|
+
# codeml.parameters[:RateAncestor] = 1
|
48
|
+
# # You can also set many parameters at a time.
|
49
|
+
# codeml.parameters.update({ :alpha => 0.5, :fix_alpha => 0 })
|
50
|
+
# # Executes codeml with the alignment and the tree
|
51
|
+
# report = codeml.query(alignment, tree)
|
52
|
+
#
|
53
|
+
# Example 2 (Obsolete usage):
|
54
|
+
#
|
55
|
+
# # Create a control file, setting some Codeml options
|
56
|
+
# # Default parameters are used otherwise, see RDoc for defaults
|
57
|
+
# # The names of the parameters correspond to those specified
|
58
|
+
# # in the Codeml documentation
|
59
|
+
# control_file = Tempfile.new('codeml_ctl')
|
60
|
+
# control_file.close(false)
|
61
|
+
# # Prepare output file as a temporary file
|
62
|
+
# output_file = Tempfile.new('codeml_test')
|
63
|
+
# output_file.close(false)
|
64
|
+
# Bio::PAML::Codeml.create_control_file(config_file.path, {
|
65
|
+
# :model => 1,
|
66
|
+
# :fix_kappa => 1,
|
67
|
+
# :aaRatefile => TEST_DATA + '/wag.dat',
|
68
|
+
# :seqfile => TEST_DATA + '/abglobin.aa',
|
69
|
+
# :treefile => TEST_DATA + '/abglobin.trees',
|
70
|
+
# :outfile => output_file.path,
|
71
|
+
# })
|
72
|
+
#
|
73
|
+
# # Create an instance of Codeml specifying where the codeml binary is
|
74
|
+
# codeml = Bio::PAML::Codeml.new('/path/to/codeml')
|
75
|
+
#
|
76
|
+
# # Run codeml using a control file
|
77
|
+
# # Returns the command line output
|
78
|
+
# codeml_output = codeml.run(control_file)
|
79
|
+
#
|
80
|
+
class Codeml < Common
|
81
|
+
|
82
|
+
autoload :Report, 'bio/appl/paml/codeml/report'
|
83
|
+
autoload :Rates, 'bio/appl/paml/codeml/rates'
|
84
|
+
|
85
|
+
# Default program name
|
86
|
+
DEFAULT_PROGRAM = 'codeml'.freeze
|
87
|
+
|
88
|
+
# Default parameters when running codeml.
|
89
|
+
#
|
90
|
+
# The parameters whose values are different from the codeml defalut
|
91
|
+
# value (described in pamlDOC.pdf) in PAML 4.1 are:
|
92
|
+
# seqfile, outfile, treefile, ndata, noisy, verbose, cleandata
|
93
|
+
#
|
94
|
+
DEFAULT_PARAMETERS = {
|
95
|
+
# Essential argumemts
|
96
|
+
:seqfile => nil,
|
97
|
+
:outfile => nil,
|
98
|
+
# Optional arguments
|
99
|
+
:treefile => nil,
|
100
|
+
:noisy => 0,
|
101
|
+
:verbose => 1,
|
102
|
+
:runmode => 0,
|
103
|
+
:seqtype => 2,
|
104
|
+
:CodonFreq => 2,
|
105
|
+
:ndata => 1,
|
106
|
+
:clock => 0,
|
107
|
+
:aaDist => 0,
|
108
|
+
:aaRatefile => 'wag.dat',
|
109
|
+
:model => 2,
|
110
|
+
:NSsites => 0,
|
111
|
+
:icode => 0,
|
112
|
+
:Mgene => 0,
|
113
|
+
:fix_kappa => 0,
|
114
|
+
:kappa => 2,
|
115
|
+
:fix_omega => 0,
|
116
|
+
:omega => 0.4,
|
117
|
+
:fix_alpha => 0,
|
118
|
+
:alpha => 0.0,
|
119
|
+
:Malpha => 0,
|
120
|
+
:ncatG => 3,
|
121
|
+
:fix_rho => 1,
|
122
|
+
:rho => 0.0,
|
123
|
+
:getSE => 0,
|
124
|
+
:RateAncestor => 0,
|
125
|
+
:Small_Diff => 0.5e-6,
|
126
|
+
:cleandata => 1,
|
127
|
+
:fix_blength => 0,
|
128
|
+
:method => 0
|
129
|
+
}
|
130
|
+
|
131
|
+
# OBSOLETE. This method should not be used.
|
132
|
+
# Instead, use parameters.
|
133
|
+
def options
|
134
|
+
warn 'The method Codeml#options will be changed to be used for command line arguments in the future. Instead, use Codeml#parameters.'
|
135
|
+
parameters
|
136
|
+
end
|
137
|
+
|
138
|
+
# OBSOLETE. This method should not be used.
|
139
|
+
# Instead, use parameters=(hash).
|
140
|
+
def options=(hash)
|
141
|
+
warn 'The method Codeml#options=() will be changed to be used for command line arguments in the future. Instead, use Codeml#parameters=().'
|
142
|
+
self.parameters=(hash)
|
143
|
+
end
|
144
|
+
|
145
|
+
# Obsolete. This method will be removed in the future.
|
146
|
+
# Helper method for creating a codeml control file.
|
147
|
+
# Note that default parameters are automatically merged.
|
148
|
+
def self.create_control_file(parameters, filename)
|
149
|
+
parameters = DEFAULT_PARAMETERS.merge(parameters)
|
150
|
+
File.open(filename, 'w') do |file|
|
151
|
+
parameters.each do |key, value|
|
152
|
+
file.puts "#{key.to_s} = #{value.to_s}" if value
|
153
|
+
end
|
154
|
+
end
|
155
|
+
filename
|
156
|
+
end
|
157
|
+
|
158
|
+
# OBSOLETE. This method will soon be removed.
|
159
|
+
# Instead, use create_control_file(parameters, filename).
|
160
|
+
def self.create_config_file(parameters, filename)
|
161
|
+
warn "The method Codeml.create_config_file(parameters, filename) will soon be removed. Instead, use Codeml.create_control_file(filename, parameters)."
|
162
|
+
create_control_file(parameters, filename)
|
163
|
+
end
|
164
|
+
|
165
|
+
|
166
|
+
# Runs the program on the internal parameters with the specified
|
167
|
+
# sequence alignment and tree.
|
168
|
+
#
|
169
|
+
# Note that parameters[:seqfile] and parameters[:outfile]
|
170
|
+
# are always modified, and parameters[:treefile] and
|
171
|
+
# parameters[:aaRatefile] are modified when tree and aarate are
|
172
|
+
# specified respectively.
|
173
|
+
#
|
174
|
+
# For other important information, see the document of
|
175
|
+
# Bio::PAML::Common#query.
|
176
|
+
#
|
177
|
+
# ---
|
178
|
+
# *Arguments*:
|
179
|
+
# * (required) _alignment_: Bio::Alignment object or similar object
|
180
|
+
# * (optional) _tree_: Bio::Tree object
|
181
|
+
# * (optional) _aarate_: String or nil
|
182
|
+
# *Returns*:: Report object
|
183
|
+
def query(alignment, tree = nil, aarate = nil)
|
184
|
+
begin
|
185
|
+
aaratefile = prepare_aaratefile(aarate)
|
186
|
+
ret = super(alignment, tree)
|
187
|
+
ensure
|
188
|
+
finalize_aaratefile(aaratefile)
|
189
|
+
end
|
190
|
+
ret
|
191
|
+
end
|
192
|
+
|
193
|
+
# Runs the program on the internal parameters with the specified
|
194
|
+
# sequence alignment data string and tree data string.
|
195
|
+
#
|
196
|
+
# Note that parameters[:outfile] is always modified, and
|
197
|
+
# parameters[:seqfile], parameters[:treefile], and
|
198
|
+
# parameters[:aaRatefile] are modified when
|
199
|
+
# alignment, tree, and aarate are specified respectively.
|
200
|
+
#
|
201
|
+
# It raises RuntimeError if seqfile is not specified in the argument
|
202
|
+
# or in the parameter.
|
203
|
+
#
|
204
|
+
# For other important information, see the document of query method.
|
205
|
+
#
|
206
|
+
# ---
|
207
|
+
# *Arguments*:
|
208
|
+
# * (optional) _alignment_: String
|
209
|
+
# * (optional) _tree_: String or nil
|
210
|
+
# * (optional) _aarate_: String or nil
|
211
|
+
# *Returns*:: contents of output file (String)
|
212
|
+
def query_by_string(alignment = nil, tree = nil, aarate = nil)
|
213
|
+
begin
|
214
|
+
aaratefile = prepare_aaratefile(aarate)
|
215
|
+
ret = super(alignment, tree)
|
216
|
+
ensure
|
217
|
+
finalize_aaratefile(aaratefile)
|
218
|
+
end
|
219
|
+
ret
|
220
|
+
end
|
221
|
+
|
222
|
+
private
|
223
|
+
|
224
|
+
# (private) prepares temporary file for aaRatefile if needed
|
225
|
+
def prepare_aaratefile(aarate)
|
226
|
+
if aarate then
|
227
|
+
aaratefile = Tempfile.new('codeml_aarate')
|
228
|
+
aaratefile.print aarate
|
229
|
+
aaratefile.close(false)
|
230
|
+
self.parameters[:aaRatefile] = aaratefile.path
|
231
|
+
end
|
232
|
+
aaratefile
|
233
|
+
end
|
234
|
+
|
235
|
+
# (private) removes temporary file for aaRatefile if needed
|
236
|
+
def finalize_aaratefile(aaratefile)
|
237
|
+
aaratefile.close(true) if aaratefile
|
238
|
+
end
|
239
|
+
|
240
|
+
end # End class Codeml
|
241
|
+
end # End module PAML
|
242
|
+
end # End module Bio
|