ngoto-bio 1.2.9.9001
Sign up to get free protection for your applications and to get access to all the features.
- data/ChangeLog +2046 -0
- data/README.rdoc +223 -0
- data/README_DEV.rdoc +285 -0
- data/Rakefile +70 -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/bioruby.gemspec +463 -0
- data/bioruby.gemspec.erb +79 -0
- data/doc/Changes-0.7.rd +369 -0
- data/doc/Changes-1.3.rdoc +195 -0
- data/doc/KEGG_API.rd +1843 -0
- data/doc/KEGG_API.rd.ja +1834 -0
- data/doc/Tutorial.rd +1296 -0
- data/doc/Tutorial.rd.ja +2640 -0
- data/etc/bioinformatics/seqdatabase.ini +210 -0
- data/extconf.rb +2 -0
- data/lib/bio/alignment.rb +2518 -0
- data/lib/bio/appl/bl2seq/report.rb +334 -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 +671 -0
- data/lib/bio/appl/blast/xmlparser.rb +236 -0
- data/lib/bio/appl/blast.rb +505 -0
- data/lib/bio/appl/blat/report.rb +530 -0
- data/lib/bio/appl/clustalw/report.rb +152 -0
- data/lib/bio/appl/clustalw.rb +219 -0
- data/lib/bio/appl/emboss.rb +203 -0
- data/lib/bio/appl/fasta/format10.rb +325 -0
- data/lib/bio/appl/fasta.rb +235 -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/report.rb +683 -0
- data/lib/bio/appl/hmmer.rb +126 -0
- data/lib/bio/appl/iprscan/report.rb +374 -0
- data/lib/bio/appl/mafft/report.rb +226 -0
- data/lib/bio/appl/mafft.rb +259 -0
- data/lib/bio/appl/muscle.rb +52 -0
- data/lib/bio/appl/paml/baseml/report.rb +32 -0
- data/lib/bio/appl/paml/baseml.rb +95 -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/codeml.rb +242 -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/report.rb +32 -0
- data/lib/bio/appl/paml/yn00.rb +103 -0
- data/lib/bio/appl/phylip/alignment.rb +129 -0
- data/lib/bio/appl/phylip/distance_matrix.rb +96 -0
- data/lib/bio/appl/probcons.rb +41 -0
- data/lib/bio/appl/psort/report.rb +457 -0
- data/lib/bio/appl/psort.rb +548 -0
- data/lib/bio/appl/pts1.rb +263 -0
- data/lib/bio/appl/sim4/report.rb +485 -0
- data/lib/bio/appl/sim4.rb +124 -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 +349 -0
- data/lib/bio/data/codontable.rb +722 -0
- data/lib/bio/data/na.rb +223 -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/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/fasta.rb +410 -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/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/pdb.rb +29 -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.rb +329 -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/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/flatfile.rb +473 -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/aa.rb +125 -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/sequence.rb +456 -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/shell.rb +44 -0
- data/lib/bio/tree.rb +852 -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/color_scheme.rb +191 -0
- data/lib/bio/util/contingency_table.rb +370 -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/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/double_stranded.rb +321 -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/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/sequence_range.rb +257 -0
- data/lib/bio/util/restriction_enzyme/range/vertical_cut_range.rb +77 -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.rb +200 -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/restriction_enzyme.rb +228 -0
- data/lib/bio/util/sirna.rb +288 -0
- data/lib/bio.rb +300 -0
- data/rdoc.zsh +8 -0
- data/sample/any2fasta.rb +59 -0
- data/sample/biofetch.rb +475 -0
- data/sample/color_scheme_na.rb +91 -0
- data/sample/dbget +37 -0
- data/sample/demo_sequence.rb +158 -0
- data/sample/enzymes.rb +78 -0
- data/sample/fasta2tab.rb +99 -0
- data/sample/fastagrep.rb +72 -0
- data/sample/fastasort.rb +54 -0
- data/sample/fsplit.rb +51 -0
- data/sample/gb2fasta.rb +30 -0
- data/sample/gb2tab.rb +325 -0
- data/sample/gbtab2mysql.rb +161 -0
- data/sample/genes2nuc.rb +33 -0
- data/sample/genes2pep.rb +33 -0
- data/sample/genes2tab.rb +81 -0
- data/sample/genome2rb.rb +29 -0
- data/sample/genome2tab.rb +76 -0
- data/sample/goslim.rb +303 -0
- data/sample/gt2fasta.rb +47 -0
- data/sample/na2aa.rb +34 -0
- data/sample/pmfetch.rb +42 -0
- data/sample/pmsearch.rb +42 -0
- data/sample/psortplot_html.rb +214 -0
- data/sample/ssearch2tab.rb +96 -0
- data/sample/tdiary.rb +158 -0
- data/sample/tfastx2tab.rb +100 -0
- data/sample/vs-genes.rb +212 -0
- data/setup.rb +1596 -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 +23 -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_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/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 +488 -0
- data/test/unit/bio/io/test_soapwsdl.rb +32 -0
- data/test/unit/bio/sequence/test_aa.rb +103 -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 +484 -0
@@ -0,0 +1,111 @@
|
|
1
|
+
#
|
2
|
+
# bio/util/restriction_enzyme/string_formatting.rb - Useful functions for string manipulation
|
3
|
+
#
|
4
|
+
# Author:: Trevor Wennblom <mailto:trevor@corevx.com>
|
5
|
+
# Copyright:: Copyright (c) 2005-2007 Midwinter Laboratories, LLC (http://midwinterlabs.com)
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: string_formatting.rb,v 1.6 2007/07/16 19:28:48 k Exp $
|
9
|
+
#
|
10
|
+
|
11
|
+
require 'bio/util/restriction_enzyme'
|
12
|
+
|
13
|
+
module Bio
|
14
|
+
class RestrictionEnzyme
|
15
|
+
|
16
|
+
module StringFormatting
|
17
|
+
include CutSymbol
|
18
|
+
extend CutSymbol
|
19
|
+
|
20
|
+
# Return the sequence with spacing for alignment. Does not add whitespace
|
21
|
+
# around cut symbols.
|
22
|
+
#
|
23
|
+
# Example:
|
24
|
+
# pattern = 'n^ng^arraxt^n'
|
25
|
+
# add_spacing( pattern ) # => "n^n g^a r r a x t^n"
|
26
|
+
#
|
27
|
+
# ---
|
28
|
+
# *Arguments*
|
29
|
+
# * +seq+: sequence with cut symbols
|
30
|
+
# * +cs+: (_optional_) Cut symbol along the string. The reason this is
|
31
|
+
# definable outside of CutSymbol is that this is a utility function used
|
32
|
+
# to form vertical and horizontal cuts such as:
|
33
|
+
#
|
34
|
+
# a|t g c
|
35
|
+
# +---+
|
36
|
+
# t a c|g
|
37
|
+
# *Returns*:: +String+ sequence with single character distance between bases
|
38
|
+
def add_spacing( seq, cs = cut_symbol )
|
39
|
+
str = ''
|
40
|
+
flag = false
|
41
|
+
seq.each_byte do |c|
|
42
|
+
c = c.chr
|
43
|
+
if c == cs
|
44
|
+
str += c
|
45
|
+
flag = false
|
46
|
+
elsif flag
|
47
|
+
str += ' ' + c
|
48
|
+
else
|
49
|
+
str += c
|
50
|
+
flag = true
|
51
|
+
end
|
52
|
+
end
|
53
|
+
str
|
54
|
+
end
|
55
|
+
|
56
|
+
# Remove extraneous nucleic acid wildcards ('n' padding) from the
|
57
|
+
# left and right sides
|
58
|
+
#
|
59
|
+
# ---
|
60
|
+
# *Arguments*
|
61
|
+
# * +s+: sequence with extraneous 'n' padding
|
62
|
+
# *Returns*:: +String+ sequence without 'n' padding on the sides
|
63
|
+
def strip_padding( s )
|
64
|
+
if s[0].chr == 'n'
|
65
|
+
s =~ %r{(n+)(.+)}
|
66
|
+
s = $2
|
67
|
+
end
|
68
|
+
if s[-1].chr == 'n'
|
69
|
+
s =~ %r{(.+?)(n+)$}
|
70
|
+
s = $1
|
71
|
+
end
|
72
|
+
s
|
73
|
+
end
|
74
|
+
|
75
|
+
# Remove extraneous nucleic acid wildcards ('n' padding) from the
|
76
|
+
# left and right sides and remove cut symbols
|
77
|
+
#
|
78
|
+
# ---
|
79
|
+
# *Arguments*
|
80
|
+
# * +s+: sequence with extraneous 'n' padding and cut symbols
|
81
|
+
# *Returns*:: +String+ sequence without 'n' padding on the sides or cut symbols
|
82
|
+
def strip_cuts_and_padding( s )
|
83
|
+
strip_padding( s.tr(cut_symbol, '') )
|
84
|
+
end
|
85
|
+
|
86
|
+
# Return the 'n' padding on the left side of the strand
|
87
|
+
#
|
88
|
+
# ---
|
89
|
+
# *Arguments*
|
90
|
+
# * +s+: sequence with extraneous 'n' padding on the left side of the strand
|
91
|
+
# *Returns*:: +String+ the 'n' padding from the left side
|
92
|
+
def left_padding( s )
|
93
|
+
s =~ %r{^n+}
|
94
|
+
ret = $&
|
95
|
+
ret ? ret : '' # Don't pass nil values
|
96
|
+
end
|
97
|
+
|
98
|
+
# Return the 'n' padding on the right side of the strand
|
99
|
+
#
|
100
|
+
# ---
|
101
|
+
# *Arguments*
|
102
|
+
# * +s+: sequence with extraneous 'n' padding on the right side of the strand
|
103
|
+
# *Returns*:: +String+ the 'n' padding from the right side
|
104
|
+
def right_padding( s )
|
105
|
+
s =~ %r{n+$}
|
106
|
+
ret = $&
|
107
|
+
ret ? ret : '' # Don't pass nil values
|
108
|
+
end
|
109
|
+
end # StringFormatting
|
110
|
+
end # RestrictionEnzyme
|
111
|
+
end # Bio
|
@@ -0,0 +1,228 @@
|
|
1
|
+
#
|
2
|
+
# bio/util/restriction_enzyme.rb - Digests DNA based on restriction enzyme cut patterns
|
3
|
+
#
|
4
|
+
# Author:: Trevor Wennblom <mailto:trevor@corevx.com>
|
5
|
+
# Copyright:: Copyright (c) 2005-2007 Midwinter Laboratories, LLC (http://midwinterlabs.com)
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: restriction_enzyme.rb,v 1.16 2007/07/16 19:28:48 k Exp $
|
9
|
+
#
|
10
|
+
|
11
|
+
module Bio #:nodoc:
|
12
|
+
|
13
|
+
autoload :REBASE, 'bio/db/rebase'
|
14
|
+
|
15
|
+
# = Description
|
16
|
+
#
|
17
|
+
# Bio::RestrictionEnzyme allows you to fragment a DNA strand using one
|
18
|
+
# or more restriction enzymes. Bio::RestrictionEnzyme is aware that
|
19
|
+
# multiple enzymes may be competing for the same recognition site and
|
20
|
+
# returns the various possible fragmentation patterns that result in
|
21
|
+
# such circumstances.
|
22
|
+
#
|
23
|
+
# When using Bio::RestrictionEnzyme you may simply use the name of common
|
24
|
+
# enzymes to cut your sequence or you may construct your own unique enzymes
|
25
|
+
# to use.
|
26
|
+
#
|
27
|
+
# Visit the documentaion for individual classes for more information.
|
28
|
+
#
|
29
|
+
# An examination of the unit tests will also reveal several interesting uses
|
30
|
+
# for the curious programmer.
|
31
|
+
#
|
32
|
+
# = Usage
|
33
|
+
#
|
34
|
+
# == Basic
|
35
|
+
#
|
36
|
+
# EcoRI cut pattern:
|
37
|
+
# G|A A T T C
|
38
|
+
# +-------+
|
39
|
+
# C T T A A|G
|
40
|
+
#
|
41
|
+
# This can also be written as:
|
42
|
+
# G^AATTC
|
43
|
+
#
|
44
|
+
# Note that to use the method +cut_with_enzyme+ from a Bio::Sequence object
|
45
|
+
# you currently must +require+ +bio/util/restriction_enzyme+ directly. If
|
46
|
+
# instead you're going to directly call Bio::RestrictionEnzyme::Analysis
|
47
|
+
# then only +bio+ needs to be +required+.
|
48
|
+
#
|
49
|
+
# require 'bio'
|
50
|
+
# require 'bio/util/restriction_enzyme'
|
51
|
+
#
|
52
|
+
# seq = Bio::Sequence::NA.new('gaattc')
|
53
|
+
# cuts = seq.cut_with_enzyme('EcoRI')
|
54
|
+
# cuts.primary # => ["aattc", "g"]
|
55
|
+
# cuts.complement # => ["cttaa", "g"]
|
56
|
+
# cuts.inspect # => "[#<struct Bio::RestrictionEnzyme::Fragment primary=\"g \", complement=\"cttaa\">, #<struct Bio::RestrictionEnzyme::Fragment primary=\"aattc\", complement=\" g\">]"
|
57
|
+
#
|
58
|
+
# seq = Bio::Sequence::NA.new('gaattc')
|
59
|
+
# cuts = seq.cut_with_enzyme('g^aattc')
|
60
|
+
# cuts.primary # => ["aattc", "g"]
|
61
|
+
# cuts.complement # => ["cttaa", "g"]
|
62
|
+
#
|
63
|
+
# seq = Bio::Sequence::NA.new('gaattc')
|
64
|
+
# cuts = seq.cut_with_enzyme('g^aattc', 'gaatt^c')
|
65
|
+
# cuts.primary # => ["aattc", "c", "g", "gaatt"]
|
66
|
+
# cuts.complement # => ["c", "cttaa", "g", "ttaag"]
|
67
|
+
#
|
68
|
+
# seq = Bio::Sequence::NA.new('gaattcgaattc')
|
69
|
+
# cuts = seq.cut_with_enzyme('EcoRI')
|
70
|
+
# cuts.primary # => ["aattc", "aattcg", "g"]
|
71
|
+
# cuts.complement # => ["cttaa", "g", "gcttaa"]
|
72
|
+
#
|
73
|
+
# seq = Bio::Sequence::NA.new('gaattcgggaattc')
|
74
|
+
# cuts = seq.cut_with_enzyme('EcoRI')
|
75
|
+
# cuts.primary # => ["aattc", "aattcggg", "g"]
|
76
|
+
# cuts.complement # => ["cttaa", "g", "gcccttaa"]
|
77
|
+
#
|
78
|
+
# cuts[0].inspect # => "#<struct Bio::RestrictionEnzyme::Fragment primary=\"g \", complement=\"cttaa\">"
|
79
|
+
#
|
80
|
+
# cuts[0].primary # => "g "
|
81
|
+
# cuts[0].complement # => "cttaa"
|
82
|
+
#
|
83
|
+
# cuts[1].primary # => "aattcggg "
|
84
|
+
# cuts[1].complement # => " gcccttaa"
|
85
|
+
#
|
86
|
+
# cuts[2].primary # => "aattc"
|
87
|
+
# cuts[2].complement # => " g"
|
88
|
+
#
|
89
|
+
# == Advanced
|
90
|
+
#
|
91
|
+
# require 'bio'
|
92
|
+
#
|
93
|
+
# enzyme_1 = Bio::RestrictionEnzyme.new('anna', [1,1], [3,3])
|
94
|
+
# enzyme_2 = Bio::RestrictionEnzyme.new('gg', [1,1])
|
95
|
+
# a = Bio::RestrictionEnzyme::Analysis.cut('agga', enzyme_1, enzyme_2)
|
96
|
+
# a.primary # => ["a", "ag", "g", "ga"]
|
97
|
+
# a.complement # => ["c", "ct", "t", "tc"]
|
98
|
+
#
|
99
|
+
# a[0].primary # => "ag"
|
100
|
+
# a[0].complement # => "tc"
|
101
|
+
#
|
102
|
+
# a[1].primary # => "ga"
|
103
|
+
# a[1].complement # => "ct"
|
104
|
+
#
|
105
|
+
# a[2].primary # => "a"
|
106
|
+
# a[2].complement # => "t"
|
107
|
+
#
|
108
|
+
# a[3].primary # => "g"
|
109
|
+
# a[3].complement # => "c"
|
110
|
+
#
|
111
|
+
# = Todo / under development
|
112
|
+
#
|
113
|
+
# * Circular DNA cutting
|
114
|
+
#
|
115
|
+
|
116
|
+
class RestrictionEnzyme
|
117
|
+
|
118
|
+
#require 'bio/util/restriction_enzyme/cut_symbol'
|
119
|
+
|
120
|
+
autoload :CutSymbol, 'bio/util/restriction_enzyme/cut_symbol'
|
121
|
+
autoload :StringFormatting, 'bio/util/restriction_enzyme/string_formatting'
|
122
|
+
autoload :SingleStrand, 'bio/util/restriction_enzyme/single_strand'
|
123
|
+
autoload :SingleStrandComplement, 'bio/util/restriction_enzyme/single_strand_complement'
|
124
|
+
autoload :DoubleStranded, 'bio/util/restriction_enzyme/double_stranded'
|
125
|
+
autoload :Analysis, 'bio/util/restriction_enzyme/analysis'
|
126
|
+
autoload :Range, 'bio/util/restriction_enzyme/range/sequence_range'
|
127
|
+
|
128
|
+
include CutSymbol
|
129
|
+
extend CutSymbol
|
130
|
+
|
131
|
+
# See Bio::RestrictionEnzyme::DoubleStranded.new for more information.
|
132
|
+
#
|
133
|
+
# ---
|
134
|
+
# *Arguments*
|
135
|
+
# * +users_enzyme_or_rebase_or_pattern+: One of three possible parameters: The name of an enzyme, a REBASE::EnzymeEntry object, or a nucleotide pattern with a cut mark.
|
136
|
+
# * +cut_locations+: The cut locations in enzyme index notation.
|
137
|
+
# *Returns*:: Bio::RestrictionEnzyme::DoubleStranded
|
138
|
+
#--
|
139
|
+
# Factory for DoubleStranded
|
140
|
+
#++
|
141
|
+
def self.new(users_enzyme_or_rebase_or_pattern, *cut_locations)
|
142
|
+
DoubleStranded.new(users_enzyme_or_rebase_or_pattern, *cut_locations)
|
143
|
+
end
|
144
|
+
|
145
|
+
# REBASE enzyme data information
|
146
|
+
#
|
147
|
+
# Returns a Bio::REBASE object loaded with all of the enzyme data on file.
|
148
|
+
#
|
149
|
+
# ---
|
150
|
+
# *Arguments*
|
151
|
+
# * _none_
|
152
|
+
# *Returns*:: Bio::REBASE
|
153
|
+
def self.rebase
|
154
|
+
enzymes_yaml_file = File.join(File.dirname(File.expand_path(__FILE__)), 'restriction_enzyme', 'enzymes.yaml')
|
155
|
+
@@rebase_enzymes ||= Bio::REBASE.load_yaml(enzymes_yaml_file)
|
156
|
+
@@rebase_enzymes
|
157
|
+
end
|
158
|
+
|
159
|
+
# Check if supplied name is the name of an available enzyme
|
160
|
+
#
|
161
|
+
# See Bio::REBASE.enzyme_name?
|
162
|
+
#
|
163
|
+
# ---
|
164
|
+
# *Arguments*
|
165
|
+
# * +name+: Enzyme name
|
166
|
+
# *Returns*:: +true+ _or_ +false+
|
167
|
+
def self.enzyme_name?( name )
|
168
|
+
self.rebase.enzyme_name?(name)
|
169
|
+
end
|
170
|
+
|
171
|
+
# See Bio::RestrictionEnzyme::Analysis.cut
|
172
|
+
def self.cut( sequence, enzymes )
|
173
|
+
Bio::RestrictionEnzyme::Analysis.cut( sequence, enzymes )
|
174
|
+
end
|
175
|
+
|
176
|
+
# A Bio::RestrictionEnzyme::Fragment is a DNA fragment composed of fused primary and
|
177
|
+
# complementary strands that would be found floating in solution after a full
|
178
|
+
# sequence is digested by one or more RestrictionEnzymes.
|
179
|
+
#
|
180
|
+
# You will notice that either the primary or complement strand will be
|
181
|
+
# padded with spaces to make them line up according to the original DNA
|
182
|
+
# configuration before they were cut.
|
183
|
+
#
|
184
|
+
# Example:
|
185
|
+
#
|
186
|
+
# Fragment 1:
|
187
|
+
# primary = "attaca"
|
188
|
+
# complement = " atga"
|
189
|
+
#
|
190
|
+
# Fragment 2:
|
191
|
+
# primary = "g "
|
192
|
+
# complement = "cta"
|
193
|
+
#
|
194
|
+
# View these with the +primary+ and +complement+ methods.
|
195
|
+
#
|
196
|
+
# Bio::RestrictionEnzyme::Fragment is a simple +Struct+ object.
|
197
|
+
#
|
198
|
+
# Note: unrelated to Bio::RestrictionEnzyme::Range::SequenceRange::Fragment
|
199
|
+
Fragment = Struct.new(:primary, :complement, :p_left, :p_right, :c_left, :c_right)
|
200
|
+
|
201
|
+
# Bio::RestrictionEnzyme::Fragments inherits from +Array+.
|
202
|
+
#
|
203
|
+
# Bio::RestrictionEnzyme::Fragments is a container for Fragment objects. It adds the
|
204
|
+
# methods +primary+ and +complement+ which returns an +Array+ of all
|
205
|
+
# respective strands from it's Fragment members in alphabetically sorted
|
206
|
+
# order. Note that it will
|
207
|
+
# not return duplicate items and does not return the spacing/padding
|
208
|
+
# that you would
|
209
|
+
# find by accessing the members directly.
|
210
|
+
#
|
211
|
+
# Example:
|
212
|
+
#
|
213
|
+
# primary = ['attaca', 'g']
|
214
|
+
# complement = ['atga', 'cta']
|
215
|
+
#
|
216
|
+
# Note: unrelated to Bio::RestrictionEnzyme::Range::SequenceRange::Fragments
|
217
|
+
class Fragments < Array
|
218
|
+
def primary; strip_and_sort(:primary); end
|
219
|
+
def complement; strip_and_sort(:complement); end
|
220
|
+
|
221
|
+
protected
|
222
|
+
|
223
|
+
def strip_and_sort( sym_strand )
|
224
|
+
self.map {|uf| uf.send( sym_strand ).tr(' ', '') }.sort
|
225
|
+
end
|
226
|
+
end
|
227
|
+
end # RestrictionEnzyme
|
228
|
+
end # Bio
|
@@ -0,0 +1,288 @@
|
|
1
|
+
#
|
2
|
+
# = bio/util/sirna.rb - Class for designing small inhibitory RNAs
|
3
|
+
#
|
4
|
+
# Copyright:: Copyright (C) 2004, 2005
|
5
|
+
# Itoshi NIKAIDO <dritoshi@gmail.com>
|
6
|
+
# License:: The Ruby License
|
7
|
+
#
|
8
|
+
# $Id: sirna.rb,v 1.11 2007/04/05 23:35:41 trevor Exp $
|
9
|
+
#
|
10
|
+
# == Bio::SiRNA - Designing siRNA.
|
11
|
+
#
|
12
|
+
# This class implements the selection rules described by Kumiko Ui-Tei
|
13
|
+
# et al. (2004) and Reynolds et al. (2004).
|
14
|
+
#
|
15
|
+
# == Example
|
16
|
+
#
|
17
|
+
# seq = Bio::Sequence::NA.new(ARGF.read)
|
18
|
+
#
|
19
|
+
# sirna = Bio::SiRNA.new(seq)
|
20
|
+
# pairs = sirna.design
|
21
|
+
#
|
22
|
+
# pairs.each do |pair|
|
23
|
+
# puts pair.report
|
24
|
+
# shrna = Bio::SiRNA::ShRNA.new(pair)
|
25
|
+
# shrna.design
|
26
|
+
# puts shrna.report
|
27
|
+
#
|
28
|
+
# puts shrna.top_strand.dna
|
29
|
+
# puts shrna.bottom_strand.dna
|
30
|
+
# end
|
31
|
+
#
|
32
|
+
# == References
|
33
|
+
#
|
34
|
+
# * Kumiko Ui-Tei et al. Guidelines for the selection of highly effective
|
35
|
+
# siRNA sequences for mammalian and chick RNA interference.
|
36
|
+
# Nucl. Acids. Res. 2004 32: 936-948.
|
37
|
+
#
|
38
|
+
# * Angela Reynolds et al. Rational siRNA design for RNA interference.
|
39
|
+
# Nature Biotech. 2004 22: 326-330.
|
40
|
+
#
|
41
|
+
|
42
|
+
require 'bio/sequence'
|
43
|
+
|
44
|
+
module Bio
|
45
|
+
|
46
|
+
# = Bio::SiRNA
|
47
|
+
# Designing siRNA.
|
48
|
+
#
|
49
|
+
# This class implements the selection rules described by Kumiko Ui-Tei
|
50
|
+
# et al. (2004) and Reynolds et al. (2004).
|
51
|
+
class SiRNA
|
52
|
+
|
53
|
+
# A parameter of size of antisense.
|
54
|
+
attr_accessor :antisense_size
|
55
|
+
|
56
|
+
# A parameter of maximal %GC.
|
57
|
+
attr_accessor :max_gc_percent
|
58
|
+
|
59
|
+
# A parameter of minimum %GC.
|
60
|
+
attr_accessor :min_gc_percent
|
61
|
+
|
62
|
+
# Input is a Bio::Sequence::NA object (the target sequence).
|
63
|
+
# Output is a list of Bio::SiRNA::Pair object.
|
64
|
+
def initialize(seq, antisense_size = 21, max_gc_percent = 60.0, min_gc_percent = 40.0)
|
65
|
+
@seq = seq.rna!
|
66
|
+
@pairs = Array.new
|
67
|
+
@antisense_size = antisense_size
|
68
|
+
@max_gc_percent = max_gc_percent
|
69
|
+
@min_gc_percent = min_gc_percent
|
70
|
+
end
|
71
|
+
|
72
|
+
# Ui-Tei's rule.
|
73
|
+
def uitei?(target)
|
74
|
+
return false unless /^.{2}[GC]/i =~ target
|
75
|
+
return false unless /[AU].{2}$/i =~ target
|
76
|
+
return false if /[GC]{9}/i =~ target
|
77
|
+
|
78
|
+
one_third = target.size * 1 / 3
|
79
|
+
start_pos = @target_size - one_third - 1
|
80
|
+
remain_seq = target.subseq(start_pos, @target_size - 2)
|
81
|
+
au_number = remain_seq.scan(/[AU]/i).size
|
82
|
+
return false if au_number < 5
|
83
|
+
|
84
|
+
return true
|
85
|
+
end
|
86
|
+
|
87
|
+
# Reynolds' rule.
|
88
|
+
def reynolds?(target)
|
89
|
+
return false if /[GC]{9}/i =~ target
|
90
|
+
return false unless /^.{4}A.{6}U.{2}[AUC].{5}[AU].{2}$/i =~ target
|
91
|
+
return true
|
92
|
+
end
|
93
|
+
|
94
|
+
# same as design('uitei').
|
95
|
+
def uitei
|
96
|
+
design('uitei')
|
97
|
+
end
|
98
|
+
|
99
|
+
# same as design('reynolds').
|
100
|
+
def reynolds
|
101
|
+
design('reynolds')
|
102
|
+
end
|
103
|
+
|
104
|
+
# rule can be one of 'uitei' (default) and 'reynolds'.
|
105
|
+
def design(rule = 'uitei')
|
106
|
+
@target_size = @antisense_size + 2
|
107
|
+
|
108
|
+
target_start = 0
|
109
|
+
@seq.window_search(@target_size) do |target|
|
110
|
+
antisense = target.subseq(1, @target_size - 2).complement.rna
|
111
|
+
sense = target.subseq(3, @target_size)
|
112
|
+
|
113
|
+
target_start += 1
|
114
|
+
target_stop = target_start + @target_size
|
115
|
+
|
116
|
+
antisense_gc_percent = antisense.gc_percent
|
117
|
+
next if antisense_gc_percent > @max_gc_percent
|
118
|
+
next if antisense_gc_percent < @min_gc_percent
|
119
|
+
|
120
|
+
case rule
|
121
|
+
when 'uitei'
|
122
|
+
next unless uitei?(target)
|
123
|
+
when 'reynolds'
|
124
|
+
next unless reynolds?(target)
|
125
|
+
else
|
126
|
+
raise NotImplementedError
|
127
|
+
end
|
128
|
+
|
129
|
+
pair = Bio::SiRNA::Pair.new(target, sense, antisense, target_start, target_stop, rule, antisense_gc_percent)
|
130
|
+
@pairs.push(pair)
|
131
|
+
end
|
132
|
+
return @pairs
|
133
|
+
end
|
134
|
+
|
135
|
+
# = Bio::SiRNA::Pair
|
136
|
+
class Pair
|
137
|
+
|
138
|
+
attr_accessor :target
|
139
|
+
|
140
|
+
attr_accessor :sense
|
141
|
+
|
142
|
+
attr_accessor :antisense
|
143
|
+
|
144
|
+
attr_accessor :start
|
145
|
+
|
146
|
+
attr_accessor :stop
|
147
|
+
|
148
|
+
attr_accessor :rule
|
149
|
+
|
150
|
+
attr_accessor :gc_percent
|
151
|
+
|
152
|
+
def initialize(target, sense, antisense, start, stop, rule, gc_percent)
|
153
|
+
@target = target
|
154
|
+
@sense = sense
|
155
|
+
@antisense = antisense
|
156
|
+
@start = start
|
157
|
+
@stop = stop
|
158
|
+
@rule = rule
|
159
|
+
@gc_percent = gc_percent
|
160
|
+
end
|
161
|
+
|
162
|
+
# human readable report
|
163
|
+
def report
|
164
|
+
report = "### siRNA\n"
|
165
|
+
report << 'Start: ' + @start.to_s + "\n"
|
166
|
+
report << 'Stop: ' + @stop.to_s + "\n"
|
167
|
+
report << 'Rule: ' + @rule.to_s + "\n"
|
168
|
+
report << 'GC %: ' + @gc_percent.to_s + "\n"
|
169
|
+
report << 'Target: ' + @target.upcase + "\n"
|
170
|
+
report << 'Sense: ' + ' ' + @sense.upcase + "\n"
|
171
|
+
report << 'Antisense: ' + @antisense.reverse.upcase + "\n"
|
172
|
+
end
|
173
|
+
|
174
|
+
# computer parsable report
|
175
|
+
#def to_s
|
176
|
+
# [ @antisense, @start, @stop ].join("\t")
|
177
|
+
#end
|
178
|
+
|
179
|
+
end # class Pair
|
180
|
+
|
181
|
+
|
182
|
+
# = Bio::SiRNA::ShRNA
|
183
|
+
# Designing shRNA.
|
184
|
+
class ShRNA
|
185
|
+
|
186
|
+
# Bio::Sequence::NA
|
187
|
+
attr_accessor :top_strand
|
188
|
+
|
189
|
+
# Bio::Sequence::NA
|
190
|
+
attr_accessor :bottom_strand
|
191
|
+
|
192
|
+
# Input is a Bio::SiRNA::Pair object (the target sequence).
|
193
|
+
def initialize(pair)
|
194
|
+
@pair = pair
|
195
|
+
end
|
196
|
+
|
197
|
+
# only the 'BLOCK-iT' rule is implemented for now.
|
198
|
+
def design(method = 'BLOCK-iT')
|
199
|
+
case method
|
200
|
+
when 'BLOCK-iT'
|
201
|
+
block_it
|
202
|
+
else
|
203
|
+
raise NotImplementedError
|
204
|
+
end
|
205
|
+
end
|
206
|
+
|
207
|
+
|
208
|
+
# same as design('BLOCK-iT').
|
209
|
+
# method can be one of 'piGENE' (default) and 'BLOCK-iT'.
|
210
|
+
def block_it(method = 'piGENE')
|
211
|
+
top = Bio::Sequence::NA.new('CACC') # top_strand_shrna_overhang
|
212
|
+
bot = Bio::Sequence::NA.new('AAAA') # bottom_strand_shrna_overhang
|
213
|
+
fwd = @pair.sense
|
214
|
+
rev = @pair.sense.complement
|
215
|
+
|
216
|
+
case method
|
217
|
+
when 'BLOCK-iT'
|
218
|
+
# From BLOCK-iT's manual
|
219
|
+
loop_fwd = Bio::Sequence::NA.new('CGAA')
|
220
|
+
loop_rev = loop_fwd.complement
|
221
|
+
when 'piGENE'
|
222
|
+
# From piGENE document
|
223
|
+
loop_fwd = Bio::Sequence::NA.new('GTGTGCTGTCC')
|
224
|
+
loop_rev = loop_fwd.complement
|
225
|
+
else
|
226
|
+
raise NotImplementedError
|
227
|
+
end
|
228
|
+
|
229
|
+
if /^G/i =~ fwd
|
230
|
+
@top_strand = top + fwd + loop_fwd + rev
|
231
|
+
@bottom_strand = bot + fwd + loop_rev + rev
|
232
|
+
else
|
233
|
+
@top_strand = top + 'G' + fwd + loop_fwd + rev
|
234
|
+
@bottom_strand = bot + fwd + loop_rev + rev + 'C'
|
235
|
+
end
|
236
|
+
end
|
237
|
+
|
238
|
+
# human readable report
|
239
|
+
def report
|
240
|
+
report = "### shRNA\n"
|
241
|
+
report << "Top strand shRNA (#{@top_strand.length} nt):\n"
|
242
|
+
report << " 5'-#{@top_strand.upcase}-3'\n"
|
243
|
+
report << "Bottom strand shRNA (#{@bottom_strand.length} nt):\n"
|
244
|
+
report << " 3'-#{@bottom_strand.reverse.upcase}-5'\n"
|
245
|
+
end
|
246
|
+
|
247
|
+
end # class ShRNA
|
248
|
+
|
249
|
+
end # class SiRNA
|
250
|
+
|
251
|
+
end # module Bio
|
252
|
+
|
253
|
+
|
254
|
+
if __FILE__ == $0
|
255
|
+
|
256
|
+
seq = Bio::Sequence::NA.new(ARGF.read)
|
257
|
+
|
258
|
+
sirna = Bio::SiRNA.new(seq)
|
259
|
+
pairs = sirna.design # or .design('uitei') or .uitei or .reynolds
|
260
|
+
|
261
|
+
pairs.each do |pair|
|
262
|
+
puts pair.report
|
263
|
+
|
264
|
+
shrna = Bio::SiRNA::ShRNA.new(pair)
|
265
|
+
shrna.design # or .design('BLOCK-iT') or .block_it
|
266
|
+
puts shrna.report
|
267
|
+
|
268
|
+
puts "# as DNA"
|
269
|
+
puts shrna.top_strand.dna
|
270
|
+
puts shrna.bottom_strand.dna
|
271
|
+
end
|
272
|
+
|
273
|
+
end
|
274
|
+
|
275
|
+
=begin
|
276
|
+
|
277
|
+
= ChangeLog
|
278
|
+
|
279
|
+
2005/03/21 Itoshi NIKAIDO <itoshi.nikaido@nifty.com>
|
280
|
+
Bio::SiRNA#ShRNA_designer method was changed design method.
|
281
|
+
|
282
|
+
2004/06/25
|
283
|
+
Bio::ShRNA class was added.
|
284
|
+
|
285
|
+
2004/06/17 Itoshi NIKAIDO <itoshi.nikaido@nifty.com>
|
286
|
+
We can use shRNA loop sequence from piGene document.
|
287
|
+
|
288
|
+
=end
|