bio 1.0.0 → 1.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/bin/bioruby +14 -122
- data/bin/br_biofetch.rb +2 -2
- data/bin/br_bioflat.rb +2 -2
- data/bin/br_biogetseq.rb +2 -2
- data/bin/br_pmfetch.rb +3 -3
- data/doc/Changes-0.7.rd +77 -0
- data/doc/KEGG_API.rd +523 -232
- data/doc/KEGG_API.rd.ja +529 -207
- data/doc/Tutorial.rd +48 -11
- data/lib/bio.rb +59 -6
- data/lib/bio/alignment.rb +713 -103
- data/lib/bio/appl/bl2seq/report.rb +2 -18
- data/lib/bio/appl/blast.rb +108 -91
- data/lib/bio/appl/blast/format0.rb +33 -18
- data/lib/bio/appl/blast/format8.rb +6 -20
- data/lib/bio/appl/blast/report.rb +293 -429
- data/lib/bio/appl/blast/rexml.rb +8 -22
- data/lib/bio/appl/blast/wublast.rb +21 -12
- data/lib/bio/appl/blast/xmlparser.rb +180 -183
- data/lib/bio/appl/blat/report.rb +127 -30
- data/lib/bio/appl/clustalw.rb +87 -59
- data/lib/bio/appl/clustalw/report.rb +20 -22
- data/lib/bio/appl/emboss.rb +113 -20
- data/lib/bio/appl/fasta.rb +173 -198
- data/lib/bio/appl/fasta/format10.rb +244 -347
- 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 +5 -23
- data/lib/bio/appl/hmmer.rb +8 -45
- data/lib/bio/appl/hmmer/report.rb +2 -20
- data/lib/bio/appl/iprscan/report.rb +374 -0
- data/lib/bio/appl/mafft.rb +87 -50
- data/lib/bio/appl/mafft/report.rb +151 -44
- data/lib/bio/appl/muscle.rb +52 -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.rb +89 -96
- data/lib/bio/appl/psort/report.rb +6 -22
- data/lib/bio/appl/pts1.rb +263 -0
- data/lib/bio/appl/sim4.rb +26 -36
- data/lib/bio/appl/sim4/report.rb +2 -18
- data/lib/bio/appl/sosui/report.rb +5 -20
- data/lib/bio/appl/spidey/report.rb +2 -2
- data/lib/bio/appl/targetp/report.rb +4 -20
- data/lib/bio/appl/tcoffee.rb +55 -0
- data/lib/bio/appl/tmhmm/report.rb +4 -20
- data/lib/bio/command.rb +235 -64
- data/lib/bio/data/aa.rb +21 -26
- data/lib/bio/data/codontable.rb +2 -20
- data/lib/bio/data/na.rb +19 -4
- data/lib/bio/db.rb +27 -12
- data/lib/bio/db/aaindex.rb +2 -20
- data/lib/bio/db/embl/common.rb +4 -21
- data/lib/bio/db/embl/embl.rb +33 -85
- data/lib/bio/db/embl/sptr.rb +612 -302
- data/lib/bio/db/embl/swissprot.rb +10 -29
- data/lib/bio/db/embl/trembl.rb +10 -29
- data/lib/bio/db/embl/uniprot.rb +10 -29
- data/lib/bio/db/fantom.rb +15 -20
- data/lib/bio/db/fasta.rb +3 -3
- data/lib/bio/db/genbank/common.rb +37 -46
- data/lib/bio/db/genbank/ddbj.rb +6 -18
- data/lib/bio/db/genbank/genbank.rb +47 -186
- data/lib/bio/db/genbank/genpept.rb +4 -17
- data/lib/bio/db/genbank/refseq.rb +4 -17
- data/lib/bio/db/gff.rb +103 -35
- data/lib/bio/db/go.rb +4 -20
- data/lib/bio/db/kegg/brite.rb +26 -36
- data/lib/bio/db/kegg/compound.rb +81 -85
- data/lib/bio/db/kegg/drug.rb +98 -0
- data/lib/bio/db/kegg/enzyme.rb +133 -110
- data/lib/bio/db/kegg/expression.rb +2 -20
- data/lib/bio/db/kegg/genes.rb +208 -238
- data/lib/bio/db/kegg/genome.rb +164 -285
- data/lib/bio/db/kegg/glycan.rb +114 -157
- data/lib/bio/db/kegg/keggtab.rb +242 -303
- data/lib/bio/db/kegg/kgml.rb +117 -160
- data/lib/bio/db/kegg/orthology.rb +112 -0
- data/lib/bio/db/kegg/reaction.rb +54 -69
- data/lib/bio/db/kegg/taxonomy.rb +331 -0
- data/lib/bio/db/lasergene.rb +209 -0
- data/lib/bio/db/litdb.rb +3 -27
- data/lib/bio/db/medline.rb +228 -249
- data/lib/bio/db/nbrf.rb +3 -3
- data/lib/bio/db/newick.rb +510 -0
- data/lib/bio/db/nexus.rb +1854 -0
- data/lib/bio/db/pdb.rb +5 -17
- data/lib/bio/db/pdb/atom.rb +2 -18
- data/lib/bio/db/pdb/chain.rb +2 -18
- data/lib/bio/db/pdb/chemicalcomponent.rb +2 -18
- data/lib/bio/db/pdb/model.rb +2 -18
- data/lib/bio/db/pdb/pdb.rb +73 -34
- data/lib/bio/db/pdb/residue.rb +4 -20
- data/lib/bio/db/pdb/utils.rb +2 -18
- data/lib/bio/db/prosite.rb +403 -422
- data/lib/bio/db/rebase.rb +84 -40
- data/lib/bio/db/soft.rb +404 -0
- data/lib/bio/db/transfac.rb +5 -17
- data/lib/bio/feature.rb +106 -52
- data/lib/bio/io/das.rb +32 -42
- data/lib/bio/io/dbget.rb +2 -20
- data/lib/bio/io/ddbjxml.rb +77 -138
- data/lib/bio/io/ebisoap.rb +158 -0
- data/lib/bio/io/ensembl.rb +229 -0
- data/lib/bio/io/fastacmd.rb +89 -82
- data/lib/bio/io/fetch.rb +163 -96
- data/lib/bio/io/flatfile.rb +170 -73
- data/lib/bio/io/flatfile/bdb.rb +3 -16
- data/lib/bio/io/flatfile/index.rb +2 -2
- data/lib/bio/io/flatfile/indexer.rb +3 -2
- data/lib/bio/io/higet.rb +12 -31
- data/lib/bio/io/keggapi.rb +210 -269
- data/lib/bio/io/ncbisoap.rb +155 -0
- data/lib/bio/io/pubmed.rb +169 -147
- data/lib/bio/io/registry.rb +4 -20
- data/lib/bio/io/soapwsdl.rb +43 -38
- data/lib/bio/io/sql.rb +242 -305
- data/lib/bio/location.rb +407 -285
- data/lib/bio/map.rb +410 -0
- data/lib/bio/pathway.rb +558 -695
- data/lib/bio/reference.rb +272 -75
- data/lib/bio/sequence.rb +255 -13
- data/lib/bio/sequence/aa.rb +71 -10
- data/lib/bio/sequence/common.rb +187 -33
- data/lib/bio/sequence/compat.rb +59 -4
- data/lib/bio/sequence/format.rb +54 -7
- data/lib/bio/sequence/generic.rb +3 -3
- data/lib/bio/sequence/na.rb +328 -26
- data/lib/bio/shell.rb +11 -4
- data/lib/bio/shell/core.rb +221 -160
- data/lib/bio/shell/demo.rb +18 -15
- data/lib/bio/shell/interface.rb +14 -12
- data/lib/bio/shell/irb.rb +95 -0
- data/lib/bio/shell/object.rb +45 -26
- data/lib/bio/shell/plugin/blast.rb +42 -0
- data/lib/bio/shell/plugin/codon.rb +22 -14
- data/lib/bio/shell/plugin/das.rb +58 -0
- data/lib/bio/shell/plugin/emboss.rb +2 -2
- data/lib/bio/shell/plugin/entry.rb +22 -11
- data/lib/bio/shell/plugin/flatfile.rb +2 -2
- data/lib/bio/shell/plugin/keggapi.rb +13 -6
- data/lib/bio/shell/plugin/midi.rb +4 -4
- data/lib/bio/shell/plugin/obda.rb +2 -2
- data/lib/bio/shell/plugin/psort.rb +56 -0
- data/lib/bio/shell/plugin/seq.rb +35 -8
- data/lib/bio/shell/plugin/soap.rb +87 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/bioruby_generator.rb +29 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/_classes.rhtml +4 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/_log.rhtml +27 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/_methods.rhtml +11 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/_modules.rhtml +4 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/_variables.rhtml +7 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby-bg.gif +0 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby-console.png +0 -0
- data/lib/bio/shell/rails/{public/images/icon.png → vendor/plugins/generators/bioruby/templates/bioruby-gem.png} +0 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby-link.gif +0 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby.css +369 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby.rhtml +47 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby_controller.rb +144 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/bioruby_helper.rb +47 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/commands.rhtml +8 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/history.rhtml +10 -0
- data/lib/bio/shell/rails/vendor/plugins/generators/bioruby/templates/index.rhtml +22 -0
- data/lib/bio/shell/script.rb +25 -0
- data/lib/bio/shell/setup.rb +109 -0
- data/lib/bio/shell/web.rb +70 -58
- data/lib/bio/tree.rb +850 -0
- data/lib/bio/util/color_scheme.rb +84 -107
- data/lib/bio/util/color_scheme/buried.rb +5 -24
- data/lib/bio/util/color_scheme/helix.rb +5 -24
- data/lib/bio/util/color_scheme/hydropathy.rb +5 -24
- data/lib/bio/util/color_scheme/nucleotide.rb +5 -24
- data/lib/bio/util/color_scheme/strand.rb +5 -24
- data/lib/bio/util/color_scheme/taylor.rb +5 -24
- data/lib/bio/util/color_scheme/turn.rb +5 -24
- data/lib/bio/util/color_scheme/zappo.rb +5 -24
- data/lib/bio/util/contingency_table.rb +70 -43
- 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 +199 -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 +4 -22
- data/sample/color_scheme_na.rb +4 -12
- data/sample/enzymes.rb +78 -0
- data/sample/goslim.rb +5 -13
- data/sample/psortplot_html.rb +4 -12
- data/test/data/blast/2.2.15.blastp.m7 +876 -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/iprscan/merged.raw +32 -0
- data/test/data/iprscan/merged.txt +74 -0
- data/test/data/soft/GDS100_partial.soft +92 -0
- data/test/data/soft/GSE3457_family_partial.soft +874 -0
- data/test/functional/bio/io/test_ensembl.rb +103 -0
- data/test/functional/bio/io/test_soapwsdl.rb +5 -17
- data/test/unit/bio/appl/bl2seq/test_report.rb +2 -2
- data/test/unit/bio/appl/blast/test_report.rb +3 -16
- data/test/unit/bio/appl/blast/test_xmlparser.rb +4 -16
- data/test/unit/bio/appl/genscan/test_report.rb +3 -16
- data/test/unit/bio/appl/hmmer/test_report.rb +3 -16
- 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/sosui/test_report.rb +3 -16
- data/test/unit/bio/appl/targetp/test_report.rb +3 -16
- data/test/unit/bio/appl/test_blast.rb +3 -16
- data/test/unit/bio/appl/test_fasta.rb +4 -16
- data/test/unit/bio/appl/test_pts1.rb +140 -0
- data/test/unit/bio/appl/tmhmm/test_report.rb +3 -16
- data/test/unit/bio/data/test_aa.rb +4 -17
- data/test/unit/bio/data/test_codontable.rb +3 -16
- data/test/unit/bio/data/test_na.rb +3 -3
- data/test/unit/bio/db/embl/test_common.rb +3 -16
- data/test/unit/bio/db/embl/test_embl.rb +3 -16
- data/test/unit/bio/db/embl/test_embl_rel89.rb +219 -0
- data/test/unit/bio/db/embl/test_sptr.rb +1548 -41
- data/test/unit/bio/db/embl/test_uniprot.rb +3 -16
- data/test/unit/bio/db/kegg/test_genes.rb +3 -16
- data/test/unit/bio/db/pdb/test_pdb.rb +7 -24
- data/test/unit/bio/db/test_aaindex.rb +2 -2
- data/test/unit/bio/db/test_fasta.rb +3 -16
- data/test/unit/bio/db/test_gff.rb +3 -16
- data/test/unit/bio/db/test_lasergene.rb +95 -0
- data/test/unit/bio/db/test_newick.rb +56 -0
- data/test/unit/bio/db/test_nexus.rb +360 -0
- data/test/unit/bio/db/test_prosite.rb +5 -18
- data/test/unit/bio/db/test_rebase.rb +11 -25
- data/test/unit/bio/db/test_soft.rb +138 -0
- data/test/unit/bio/io/test_ddbjxml.rb +5 -17
- data/test/unit/bio/io/test_ensembl.rb +109 -0
- data/test/unit/bio/io/test_fastacmd.rb +3 -16
- data/test/unit/bio/io/test_flatfile.rb +237 -0
- data/test/unit/bio/io/test_soapwsdl.rb +4 -17
- data/test/unit/bio/sequence/test_aa.rb +3 -3
- data/test/unit/bio/sequence/test_common.rb +3 -16
- data/test/unit/bio/sequence/test_compat.rb +3 -16
- data/test/unit/bio/sequence/test_na.rb +29 -3
- data/test/unit/bio/shell/plugin/test_seq.rb +8 -8
- data/test/unit/bio/test_alignment.rb +16 -27
- data/test/unit/bio/test_command.rb +242 -25
- data/test/unit/bio/test_db.rb +3 -16
- data/test/unit/bio/test_feature.rb +4 -16
- data/test/unit/bio/test_location.rb +4 -16
- data/test/unit/bio/test_map.rb +230 -0
- data/test/unit/bio/test_pathway.rb +4 -16
- data/test/unit/bio/test_reference.rb +2 -2
- data/test/unit/bio/test_sequence.rb +7 -19
- data/test/unit/bio/test_shell.rb +3 -16
- 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 +100 -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 +6 -18
- data/test/unit/bio/util/test_contingency_table.rb +6 -18
- data/test/unit/bio/util/test_restriction_enzyme.rb +42 -0
- data/test/unit/bio/util/test_sirna.rb +3 -16
- metadata +228 -169
- data/doc/BioRuby.rd.ja +0 -225
- data/doc/Design.rd.ja +0 -341
- data/doc/TODO.rd.ja +0 -138
- data/lib/bio/appl/fasta/format6.rb +0 -37
- data/lib/bio/db/kegg/cell.rb +0 -88
- data/lib/bio/db/kegg/ko.rb +0 -178
- data/lib/bio/shell/rails/Rakefile +0 -10
- data/lib/bio/shell/rails/app/controllers/application.rb +0 -4
- data/lib/bio/shell/rails/app/controllers/shell_controller.rb +0 -94
- data/lib/bio/shell/rails/app/helpers/application_helper.rb +0 -3
- data/lib/bio/shell/rails/app/models/shell_connection.rb +0 -30
- data/lib/bio/shell/rails/app/views/layouts/shell.rhtml +0 -37
- data/lib/bio/shell/rails/app/views/shell/history.rhtml +0 -5
- data/lib/bio/shell/rails/app/views/shell/index.rhtml +0 -2
- data/lib/bio/shell/rails/app/views/shell/show.rhtml +0 -13
- data/lib/bio/shell/rails/config/boot.rb +0 -19
- data/lib/bio/shell/rails/config/database.yml +0 -85
- data/lib/bio/shell/rails/config/environment.rb +0 -53
- data/lib/bio/shell/rails/config/environments/development.rb +0 -19
- data/lib/bio/shell/rails/config/environments/production.rb +0 -19
- data/lib/bio/shell/rails/config/environments/test.rb +0 -19
- data/lib/bio/shell/rails/config/routes.rb +0 -19
- data/lib/bio/shell/rails/doc/README_FOR_APP +0 -2
- data/lib/bio/shell/rails/public/404.html +0 -8
- data/lib/bio/shell/rails/public/500.html +0 -8
- data/lib/bio/shell/rails/public/dispatch.cgi +0 -10
- data/lib/bio/shell/rails/public/dispatch.fcgi +0 -24
- data/lib/bio/shell/rails/public/dispatch.rb +0 -10
- data/lib/bio/shell/rails/public/favicon.ico +0 -0
- data/lib/bio/shell/rails/public/images/rails.png +0 -0
- data/lib/bio/shell/rails/public/index.html +0 -277
- data/lib/bio/shell/rails/public/javascripts/controls.js +0 -750
- data/lib/bio/shell/rails/public/javascripts/dragdrop.js +0 -584
- data/lib/bio/shell/rails/public/javascripts/effects.js +0 -854
- data/lib/bio/shell/rails/public/javascripts/prototype.js +0 -1785
- data/lib/bio/shell/rails/public/robots.txt +0 -1
- data/lib/bio/shell/rails/public/stylesheets/main.css +0 -187
- data/lib/bio/shell/rails/script/about +0 -3
- data/lib/bio/shell/rails/script/breakpointer +0 -3
- data/lib/bio/shell/rails/script/console +0 -3
- data/lib/bio/shell/rails/script/destroy +0 -3
- data/lib/bio/shell/rails/script/generate +0 -3
- data/lib/bio/shell/rails/script/performance/benchmarker +0 -3
- data/lib/bio/shell/rails/script/performance/profiler +0 -3
- data/lib/bio/shell/rails/script/plugin +0 -3
- data/lib/bio/shell/rails/script/process/reaper +0 -3
- data/lib/bio/shell/rails/script/process/spawner +0 -3
- data/lib/bio/shell/rails/script/process/spinner +0 -3
- data/lib/bio/shell/rails/script/runner +0 -3
- data/lib/bio/shell/rails/script/server +0 -42
- data/lib/bio/shell/rails/test/test_helper.rb +0 -28
data/lib/bio/tree.rb
ADDED
|
@@ -0,0 +1,850 @@
|
|
|
1
|
+
#
|
|
2
|
+
# = bio/tree.rb - phylogenetic tree data structure class
|
|
3
|
+
#
|
|
4
|
+
# Copyright:: Copyright (C) 2006
|
|
5
|
+
# Naohisa Goto <ng@bioruby.org>
|
|
6
|
+
# License:: The Ruby License
|
|
7
|
+
#
|
|
8
|
+
# $Id: tree.rb,v 1.8 2007/04/05 23:35:39 trevor Exp $
|
|
9
|
+
#
|
|
10
|
+
|
|
11
|
+
require 'matrix'
|
|
12
|
+
require 'bio/pathway'
|
|
13
|
+
|
|
14
|
+
module Bio
|
|
15
|
+
|
|
16
|
+
# This is the class for phylogenetic tree.
|
|
17
|
+
# It stores a phylogenetic tree.
|
|
18
|
+
#
|
|
19
|
+
# Internally, it is based on Bio::Pathway class.
|
|
20
|
+
# However, users cannot handle Bio::Pathway object directly.
|
|
21
|
+
#
|
|
22
|
+
# This is alpha version. Incompatible changes may be made frequently.
|
|
23
|
+
class Tree
|
|
24
|
+
|
|
25
|
+
# Error when there are no path between specified nodes
|
|
26
|
+
class NoPathError < RuntimeError; end
|
|
27
|
+
|
|
28
|
+
# Edge object of each node.
|
|
29
|
+
# By default, the object doesn't contain any node information.
|
|
30
|
+
class Edge
|
|
31
|
+
|
|
32
|
+
# creates a new edge.
|
|
33
|
+
def initialize(distance = nil)
|
|
34
|
+
if distance.kind_of?(Numeric)
|
|
35
|
+
self.distance = distance
|
|
36
|
+
elsif distance
|
|
37
|
+
self.distance_string = distance
|
|
38
|
+
end
|
|
39
|
+
end
|
|
40
|
+
|
|
41
|
+
# evolutionary distance
|
|
42
|
+
attr_reader :distance
|
|
43
|
+
|
|
44
|
+
# evolutionary distance represented as a string
|
|
45
|
+
attr_reader :distance_string
|
|
46
|
+
|
|
47
|
+
# set evolutionary distance value
|
|
48
|
+
def distance=(num)
|
|
49
|
+
@distance = num
|
|
50
|
+
@distance_string = (num ? num.to_s : num)
|
|
51
|
+
end
|
|
52
|
+
|
|
53
|
+
# set evolutionary distance value from a string
|
|
54
|
+
def distance_string=(str)
|
|
55
|
+
if str.to_s.strip.empty?
|
|
56
|
+
@distance = nil
|
|
57
|
+
@distance_string = str
|
|
58
|
+
else
|
|
59
|
+
@distance = str.to_f
|
|
60
|
+
@distance_string = str
|
|
61
|
+
end
|
|
62
|
+
end
|
|
63
|
+
|
|
64
|
+
# visualization of this object
|
|
65
|
+
def inspect
|
|
66
|
+
"<Edge distance=#{@distance.inspect}>"
|
|
67
|
+
end
|
|
68
|
+
|
|
69
|
+
# string representation of this object
|
|
70
|
+
def to_s
|
|
71
|
+
@distance_string.to_s
|
|
72
|
+
end
|
|
73
|
+
|
|
74
|
+
#---
|
|
75
|
+
# methods for NHX (New Hampshire eXtended) and/or PhyloXML
|
|
76
|
+
#+++
|
|
77
|
+
|
|
78
|
+
# log likelihood value (:L in NHX)
|
|
79
|
+
attr_accessor :log_likelihood
|
|
80
|
+
|
|
81
|
+
# width of the edge
|
|
82
|
+
# (<branch width="w"> of PhyloXML, or :W="w" in NHX)
|
|
83
|
+
attr_accessor :width
|
|
84
|
+
|
|
85
|
+
# Other NHX parameters. Returns a Hash.
|
|
86
|
+
# Note that :L and :W
|
|
87
|
+
# are not stored here but stored in the proper attributes in this class.
|
|
88
|
+
# However, if you force to set these parameters in this hash,
|
|
89
|
+
# the parameters in this hash are preferred when generating NHX.
|
|
90
|
+
# In addition, If the same parameters are defined at Node object,
|
|
91
|
+
# the parameters in the node are preferred.
|
|
92
|
+
def nhx_parameters
|
|
93
|
+
@nhx_parameters ||= {}
|
|
94
|
+
@nhx_parameters
|
|
95
|
+
end
|
|
96
|
+
|
|
97
|
+
end #class Edge
|
|
98
|
+
|
|
99
|
+
# Gets distance value from the given edge.
|
|
100
|
+
# Returns float or any other numeric value or nil.
|
|
101
|
+
def get_edge_distance(edge)
|
|
102
|
+
begin
|
|
103
|
+
dist = edge.distance
|
|
104
|
+
rescue NoMethodError
|
|
105
|
+
dist = edge
|
|
106
|
+
end
|
|
107
|
+
dist
|
|
108
|
+
end
|
|
109
|
+
|
|
110
|
+
# Gets distance string from the given edge.
|
|
111
|
+
# Returns a string or nil.
|
|
112
|
+
def get_edge_distance_string(edge)
|
|
113
|
+
begin
|
|
114
|
+
dist = edge.distance_string
|
|
115
|
+
rescue NoMethodError
|
|
116
|
+
dist = (edge ? edge.to_s : nil)
|
|
117
|
+
end
|
|
118
|
+
dist
|
|
119
|
+
end
|
|
120
|
+
|
|
121
|
+
# Returns edge1 + edge2
|
|
122
|
+
def get_edge_merged(edge1, edge2)
|
|
123
|
+
dist1 = get_edge_distance(edge1)
|
|
124
|
+
dist2 = get_edge_distance(edge2)
|
|
125
|
+
if dist1 and dist2 then
|
|
126
|
+
Edge.new(dist1 + dist2)
|
|
127
|
+
elsif dist1 then
|
|
128
|
+
Edge.new(dist1)
|
|
129
|
+
elsif dist2 then
|
|
130
|
+
Edge.new(dist2)
|
|
131
|
+
else
|
|
132
|
+
Edge.new
|
|
133
|
+
end
|
|
134
|
+
end
|
|
135
|
+
|
|
136
|
+
# Node object.
|
|
137
|
+
class Node
|
|
138
|
+
|
|
139
|
+
# Creates a new node.
|
|
140
|
+
def initialize(name = nil)
|
|
141
|
+
@name = name if name
|
|
142
|
+
end
|
|
143
|
+
|
|
144
|
+
# name of the node
|
|
145
|
+
attr_accessor :name
|
|
146
|
+
|
|
147
|
+
# bootstrap value
|
|
148
|
+
attr_reader :bootstrap
|
|
149
|
+
|
|
150
|
+
# bootstrap value as a string
|
|
151
|
+
attr_reader :bootstrap_string
|
|
152
|
+
|
|
153
|
+
# sets a bootstrap value
|
|
154
|
+
def bootstrap=(num)
|
|
155
|
+
@bootstrap_string = (num ? num.to_s : num)
|
|
156
|
+
@bootstrap = num
|
|
157
|
+
end
|
|
158
|
+
|
|
159
|
+
# sets a bootstrap value from a string
|
|
160
|
+
def bootstrap_string=(str)
|
|
161
|
+
if str.to_s.strip.empty?
|
|
162
|
+
@bootstrap = nil
|
|
163
|
+
@bootstrap_string = str
|
|
164
|
+
else
|
|
165
|
+
i = str.to_i
|
|
166
|
+
f = str.to_f
|
|
167
|
+
@bootstrap = (i == f ? i : f)
|
|
168
|
+
@bootstrap_string = str
|
|
169
|
+
end
|
|
170
|
+
end
|
|
171
|
+
|
|
172
|
+
# visualization of this object
|
|
173
|
+
def inspect
|
|
174
|
+
if @name and !@name.empty? then
|
|
175
|
+
str = "(Node:#{@name.inspect}"
|
|
176
|
+
else
|
|
177
|
+
str = sprintf('(Node:%x', (self.__id__ << 1) & 0xffffffff)
|
|
178
|
+
end
|
|
179
|
+
str += " bootstrap=#{@bootstrap.inspect}" if @bootstrap
|
|
180
|
+
str += ")"
|
|
181
|
+
str
|
|
182
|
+
end
|
|
183
|
+
|
|
184
|
+
# string representation of this object
|
|
185
|
+
def to_s
|
|
186
|
+
@name.to_s
|
|
187
|
+
end
|
|
188
|
+
|
|
189
|
+
# the order of the node
|
|
190
|
+
# (lower value, high priority)
|
|
191
|
+
attr_accessor :order_number
|
|
192
|
+
|
|
193
|
+
#---
|
|
194
|
+
# methods for NHX (New Hampshire eXtended) and/or PhyloXML
|
|
195
|
+
#+++
|
|
196
|
+
|
|
197
|
+
# Phylogenetic events.
|
|
198
|
+
# Returns an Array of one (or more?) of the following symbols
|
|
199
|
+
# :gene_duplication
|
|
200
|
+
# :speciation
|
|
201
|
+
def events
|
|
202
|
+
@events ||= []
|
|
203
|
+
@events
|
|
204
|
+
end
|
|
205
|
+
|
|
206
|
+
# EC number (EC_number in PhyloXML, or :E in NHX)
|
|
207
|
+
attr_accessor :ec_number
|
|
208
|
+
|
|
209
|
+
# scientific name (scientific_name in PhyloXML, or :S in NHX)
|
|
210
|
+
attr_accessor :scientific_name
|
|
211
|
+
|
|
212
|
+
# taxonomy identifier (taxonomy_identifier in PhyloXML, or :T in NHX)
|
|
213
|
+
attr_accessor :taxonomy_id
|
|
214
|
+
|
|
215
|
+
# Other NHX parameters. Returns a Hash.
|
|
216
|
+
# Note that :D, :E, :S, and :T
|
|
217
|
+
# are not stored here but stored in the proper attributes in this class.
|
|
218
|
+
# However, if you force to set these parameters in this hash,
|
|
219
|
+
# the parameters in this hash are preferred when generating NHX.
|
|
220
|
+
def nhx_parameters
|
|
221
|
+
@nhx_parameters ||= {}
|
|
222
|
+
@nhx_parameters
|
|
223
|
+
end
|
|
224
|
+
|
|
225
|
+
end #class Node
|
|
226
|
+
|
|
227
|
+
# Gets node name
|
|
228
|
+
def get_node_name(node)
|
|
229
|
+
begin
|
|
230
|
+
node.name
|
|
231
|
+
rescue NoMethodError
|
|
232
|
+
node.to_s
|
|
233
|
+
end
|
|
234
|
+
end
|
|
235
|
+
|
|
236
|
+
def get_node_bootstrap(node)
|
|
237
|
+
begin
|
|
238
|
+
node.bootstrap
|
|
239
|
+
rescue NoMethodError
|
|
240
|
+
nil
|
|
241
|
+
end
|
|
242
|
+
end
|
|
243
|
+
|
|
244
|
+
def get_node_bootstrap_string(node)
|
|
245
|
+
begin
|
|
246
|
+
node.bootstrap_string
|
|
247
|
+
rescue NoMethodError
|
|
248
|
+
nil
|
|
249
|
+
end
|
|
250
|
+
end
|
|
251
|
+
|
|
252
|
+
# Creates a new phylogenetic tree.
|
|
253
|
+
# When no arguments are given, it creates a new empty tree.
|
|
254
|
+
# When a Tree object is given, it copies the tree.
|
|
255
|
+
# Note that the new tree shares Node and Edge objects
|
|
256
|
+
# with the given tree.
|
|
257
|
+
def initialize(tree = nil)
|
|
258
|
+
# creates an undirected adjacency list graph
|
|
259
|
+
@pathway = Bio::Pathway.new([], true)
|
|
260
|
+
@root = nil
|
|
261
|
+
@options = {}
|
|
262
|
+
self.concat(tree) if tree
|
|
263
|
+
end
|
|
264
|
+
|
|
265
|
+
# root node of this tree
|
|
266
|
+
# (even if unrooted tree, it is used by some methods)
|
|
267
|
+
attr_accessor :root
|
|
268
|
+
|
|
269
|
+
# tree options; mainly used for tree output
|
|
270
|
+
attr_accessor :options
|
|
271
|
+
|
|
272
|
+
# Clears all nodes and edges.
|
|
273
|
+
# Returns self.
|
|
274
|
+
# Note that options and root are also cleared.
|
|
275
|
+
def clear
|
|
276
|
+
initialize
|
|
277
|
+
self
|
|
278
|
+
end
|
|
279
|
+
|
|
280
|
+
# Returns all nodes as an array.
|
|
281
|
+
def nodes
|
|
282
|
+
@pathway.graph.keys
|
|
283
|
+
end
|
|
284
|
+
|
|
285
|
+
# Number of nodes.
|
|
286
|
+
def number_of_nodes
|
|
287
|
+
@pathway.nodes
|
|
288
|
+
end
|
|
289
|
+
|
|
290
|
+
# Iterates over each node of this tree.
|
|
291
|
+
def each_node(&x) #:yields: node
|
|
292
|
+
@pathway.graph.each_key(&x)
|
|
293
|
+
self
|
|
294
|
+
end
|
|
295
|
+
|
|
296
|
+
# Iterates over each edges of this tree.
|
|
297
|
+
def each_edge #:yields: source, target, edge
|
|
298
|
+
@pathway.relations.each do |rel|
|
|
299
|
+
yield rel.node[0], rel.node[1], rel.relation
|
|
300
|
+
end
|
|
301
|
+
self
|
|
302
|
+
end
|
|
303
|
+
|
|
304
|
+
# Returns all edges an array of [ node0, node1, edge ]
|
|
305
|
+
def edges
|
|
306
|
+
@pathway.relations.collect do |rel|
|
|
307
|
+
[ rel.node[0], rel.node[1], rel.relation ]
|
|
308
|
+
end
|
|
309
|
+
end
|
|
310
|
+
|
|
311
|
+
# Returns number of edges in the tree.
|
|
312
|
+
def number_of_edges
|
|
313
|
+
@pathway.relations.size
|
|
314
|
+
end
|
|
315
|
+
|
|
316
|
+
# Returns an array of adjacent nodes of the given node.
|
|
317
|
+
def adjacent_nodes(node)
|
|
318
|
+
h = @pathway.graph[node]
|
|
319
|
+
h ? h.keys : []
|
|
320
|
+
end
|
|
321
|
+
|
|
322
|
+
# Returns all connected edges with adjacent nodes.
|
|
323
|
+
# Returns an array of the array [ source, target, edge ].
|
|
324
|
+
#
|
|
325
|
+
# The reason why the method name is "out_edges" is that
|
|
326
|
+
# it comes from the Boost Graph Library.
|
|
327
|
+
def out_edges(source)
|
|
328
|
+
h = @pathway.graph[source]
|
|
329
|
+
if h
|
|
330
|
+
h.collect { |key, val| [ source, key, val ] }
|
|
331
|
+
else
|
|
332
|
+
[]
|
|
333
|
+
end
|
|
334
|
+
end
|
|
335
|
+
|
|
336
|
+
# Iterates over each connected edges of the given node.
|
|
337
|
+
# Returns self.
|
|
338
|
+
#
|
|
339
|
+
# The reason why the method name is "each_out_edge" is that
|
|
340
|
+
# it comes from the Boost Graph Library.
|
|
341
|
+
def each_out_edge(source) #:yields: source, target, edge
|
|
342
|
+
h = @pathway.graph[source]
|
|
343
|
+
h.each { |key, val| yield source, key, val } if h
|
|
344
|
+
self
|
|
345
|
+
end
|
|
346
|
+
|
|
347
|
+
# Returns number of edges in the given node.
|
|
348
|
+
#
|
|
349
|
+
# The reason why the method name is "out_degree" is that
|
|
350
|
+
# it comes from the Boost Graph Library.
|
|
351
|
+
def out_degree(source)
|
|
352
|
+
h = @pathway.graph[source]
|
|
353
|
+
h ? h.size : 0
|
|
354
|
+
end
|
|
355
|
+
|
|
356
|
+
# Returns an edge from source to target.
|
|
357
|
+
# If source and target are not adjacent nodes, returns nil.
|
|
358
|
+
def get_edge(source, target)
|
|
359
|
+
h = @pathway.graph[source]
|
|
360
|
+
h ? h[target] : nil
|
|
361
|
+
end
|
|
362
|
+
|
|
363
|
+
# Adds a new edge to the tree.
|
|
364
|
+
# Returns the newly added edge.
|
|
365
|
+
# If the edge already exists, it is overwritten with new one.
|
|
366
|
+
def add_edge(source, target, edge = Edge.new)
|
|
367
|
+
@pathway.append(Bio::Relation.new(source, target, edge))
|
|
368
|
+
edge
|
|
369
|
+
end
|
|
370
|
+
|
|
371
|
+
# Finds a node in the tree by given name and returns the node.
|
|
372
|
+
# If the node does not found, returns nil.
|
|
373
|
+
# If multiple nodes with the same name exist,
|
|
374
|
+
# the result would be one of those (unspecified).
|
|
375
|
+
def get_node_by_name(str)
|
|
376
|
+
self.each_node do |node|
|
|
377
|
+
if get_node_name(node) == str
|
|
378
|
+
return node
|
|
379
|
+
end
|
|
380
|
+
end
|
|
381
|
+
nil
|
|
382
|
+
end
|
|
383
|
+
|
|
384
|
+
# Adds a node to the tree.
|
|
385
|
+
# Returns self.
|
|
386
|
+
# If the node already exists, it does nothing.
|
|
387
|
+
def add_node(node)
|
|
388
|
+
@pathway.graph[node] ||= {}
|
|
389
|
+
self
|
|
390
|
+
end
|
|
391
|
+
|
|
392
|
+
# If the node exists, returns true.
|
|
393
|
+
# Otherwise, returns false.
|
|
394
|
+
def include?(node)
|
|
395
|
+
@pathway.graph[node] ? true : false
|
|
396
|
+
end
|
|
397
|
+
|
|
398
|
+
# Removes all edges connected with the node.
|
|
399
|
+
# Returns self.
|
|
400
|
+
# If the node does not exist, raises IndexError.
|
|
401
|
+
def clear_node(node)
|
|
402
|
+
unless self.include?(node)
|
|
403
|
+
raise IndexError, 'the node does not exist'
|
|
404
|
+
end
|
|
405
|
+
@pathway.relations.delete_if do |rel|
|
|
406
|
+
rel.node.include?(node)
|
|
407
|
+
end
|
|
408
|
+
@pathway.graph[node].each_key do |k|
|
|
409
|
+
@pathway.graph[k].delete(node)
|
|
410
|
+
end
|
|
411
|
+
@pathway.graph[node].clear
|
|
412
|
+
self
|
|
413
|
+
end
|
|
414
|
+
|
|
415
|
+
# Removes the given node from the tree.
|
|
416
|
+
# All edges connected with the node are also removed.
|
|
417
|
+
# Returns self.
|
|
418
|
+
# If the node does not exist, raises IndexError.
|
|
419
|
+
def remove_node(node)
|
|
420
|
+
self.clear_node(node)
|
|
421
|
+
@pathway.graph.delete(node)
|
|
422
|
+
self
|
|
423
|
+
end
|
|
424
|
+
|
|
425
|
+
# Removes each node if the block returns not nil.
|
|
426
|
+
# All edges connected with the removed nodes are also removed.
|
|
427
|
+
# Returns self.
|
|
428
|
+
def remove_node_if
|
|
429
|
+
all = self.nodes
|
|
430
|
+
all.each do |node|
|
|
431
|
+
if yield node then
|
|
432
|
+
self.clear_node(node)
|
|
433
|
+
@pathway.graph.delete(node)
|
|
434
|
+
end
|
|
435
|
+
end
|
|
436
|
+
self
|
|
437
|
+
end
|
|
438
|
+
|
|
439
|
+
# Removes an edge between source and target.
|
|
440
|
+
# Returns self.
|
|
441
|
+
# If the edge does not exist, raises IndexError.
|
|
442
|
+
#---
|
|
443
|
+
# If two or more edges exists between source and target,
|
|
444
|
+
# all of them are removed.
|
|
445
|
+
#+++
|
|
446
|
+
def remove_edge(source, target)
|
|
447
|
+
unless self.get_edge(source, target) then
|
|
448
|
+
raise IndexError, 'edge not found'
|
|
449
|
+
end
|
|
450
|
+
fwd = [ source, target ]
|
|
451
|
+
rev = [ target, source ]
|
|
452
|
+
@pathway.relations.delete_if do |rel|
|
|
453
|
+
rel.node == fwd or rel.node == rev
|
|
454
|
+
end
|
|
455
|
+
h = @pathway.graph[source]
|
|
456
|
+
h.delete(target) if h
|
|
457
|
+
h = @pathway.graph[target]
|
|
458
|
+
h.delete(source) if h
|
|
459
|
+
self
|
|
460
|
+
end
|
|
461
|
+
|
|
462
|
+
# Removes each edge if the block returns not nil.
|
|
463
|
+
# Returns self.
|
|
464
|
+
def remove_edge_if #:yields: source, target, edge
|
|
465
|
+
removed_rel = []
|
|
466
|
+
@pathway.relations.delete_if do |rel|
|
|
467
|
+
if yield rel.node[0], rel.node[1], edge then
|
|
468
|
+
removed_rel << rel
|
|
469
|
+
true
|
|
470
|
+
end
|
|
471
|
+
end
|
|
472
|
+
removed_rel.each do |rel|
|
|
473
|
+
source = rel[0]
|
|
474
|
+
target = rel[1]
|
|
475
|
+
h = @pathway.graph[source]
|
|
476
|
+
h.delete(target) if h
|
|
477
|
+
h = @pathway.graph[target]
|
|
478
|
+
h.delete(source) if h
|
|
479
|
+
end
|
|
480
|
+
self
|
|
481
|
+
end
|
|
482
|
+
|
|
483
|
+
# Replaces each node by each block's return value.
|
|
484
|
+
# Returns self.
|
|
485
|
+
def collect_node! #:yields: node
|
|
486
|
+
tr = {}
|
|
487
|
+
self.each_node do |node|
|
|
488
|
+
tr[node] = yield node
|
|
489
|
+
end
|
|
490
|
+
# replaces nodes in @pathway.relations
|
|
491
|
+
@pathway.relations.each do |rel|
|
|
492
|
+
rel.node.collect! { |node| tr[node] }
|
|
493
|
+
end
|
|
494
|
+
# re-generates @pathway from relations
|
|
495
|
+
@pathway.to_list
|
|
496
|
+
# adds orphan nodes
|
|
497
|
+
tr.each_value do |newnode|
|
|
498
|
+
@pathway.graph[newnode] ||= {}
|
|
499
|
+
end
|
|
500
|
+
self
|
|
501
|
+
end
|
|
502
|
+
|
|
503
|
+
# Replaces each edge by each block's return value.
|
|
504
|
+
# Returns self.
|
|
505
|
+
def collect_edge! #:yields: source, target, edge
|
|
506
|
+
@pathway.relations.each do |rel|
|
|
507
|
+
newedge = yield rel.node[0], rel.node[1], rel.relation
|
|
508
|
+
rel.relation = newedge
|
|
509
|
+
@pathway.append(rel, false)
|
|
510
|
+
end
|
|
511
|
+
self
|
|
512
|
+
end
|
|
513
|
+
|
|
514
|
+
# Gets the sub-tree consisted of given nodes.
|
|
515
|
+
# _nodes_ must be an array of nodes.
|
|
516
|
+
# Nodes that do not exist in the original tree are ignored.
|
|
517
|
+
# Returns a Tree object.
|
|
518
|
+
# Note that the sub-tree shares Node and Edge objects
|
|
519
|
+
# with the original tree.
|
|
520
|
+
def subtree(nodes)
|
|
521
|
+
nodes = nodes.find_all do |x|
|
|
522
|
+
@pathway.graph[x]
|
|
523
|
+
end
|
|
524
|
+
return self.class.new if nodes.empty?
|
|
525
|
+
# creates subtree
|
|
526
|
+
new_tree = self.class.new
|
|
527
|
+
nodes.each do |x|
|
|
528
|
+
new_tree.add_node(x)
|
|
529
|
+
end
|
|
530
|
+
self.each_edge do |node1, node2, edge|
|
|
531
|
+
if new_tree.include?(node1) and new_tree.include?(node2) then
|
|
532
|
+
new_tree.add_edge(node1, node2, edge)
|
|
533
|
+
end
|
|
534
|
+
end
|
|
535
|
+
return new_tree
|
|
536
|
+
end
|
|
537
|
+
|
|
538
|
+
# Gets the sub-tree consisted of given nodes and
|
|
539
|
+
# all internal nodes connected between given nodes.
|
|
540
|
+
# _nodes_ must be an array of nodes.
|
|
541
|
+
# Nodes that do not exist in the original tree are ignored.
|
|
542
|
+
# Returns a Tree object.
|
|
543
|
+
# The result is unspecified for cyclic trees.
|
|
544
|
+
# Note that the sub-tree shares Node and Edge objects
|
|
545
|
+
# with the original tree.
|
|
546
|
+
def subtree_with_all_paths(nodes)
|
|
547
|
+
hash = {}
|
|
548
|
+
nodes.each { |x| hash[x] = true }
|
|
549
|
+
nodes.each_index do |i|
|
|
550
|
+
node1 = nodes[i]
|
|
551
|
+
(0...i).each do |j|
|
|
552
|
+
node2 = nodes[j]
|
|
553
|
+
unless node1 == node2 then
|
|
554
|
+
begin
|
|
555
|
+
path = self.path(node1, node2)
|
|
556
|
+
rescue IndexError, NoPathError
|
|
557
|
+
path = []
|
|
558
|
+
end
|
|
559
|
+
path.each { |x| hash[x] = true }
|
|
560
|
+
end
|
|
561
|
+
end
|
|
562
|
+
end
|
|
563
|
+
self.subtree(hash.keys)
|
|
564
|
+
end
|
|
565
|
+
|
|
566
|
+
# Concatenates the other tree.
|
|
567
|
+
# If the same edge exists, the edge in _other_ is used.
|
|
568
|
+
# Returns self.
|
|
569
|
+
# The result is unspecified if _other_ isn't a Tree object.
|
|
570
|
+
# Note that the Node and Edge objects in the _other_ tree are
|
|
571
|
+
# shared in the concatinated tree.
|
|
572
|
+
def concat(other)
|
|
573
|
+
#raise TypeError unless other.kind_of?(self.class)
|
|
574
|
+
other.each_node do |node|
|
|
575
|
+
self.add_node(node)
|
|
576
|
+
end
|
|
577
|
+
other.each_edge do |node1, node2, edge|
|
|
578
|
+
self.add_edge(node1, node2, edge)
|
|
579
|
+
end
|
|
580
|
+
self
|
|
581
|
+
end
|
|
582
|
+
|
|
583
|
+
# Gets path from node1 to node2.
|
|
584
|
+
# Retruns an array of nodes, including node1 and node2.
|
|
585
|
+
# If node1 and/or node2 do not exist, IndexError is raised.
|
|
586
|
+
# If node1 and node2 are not connected, NoPathError is raised.
|
|
587
|
+
# The result is unspecified for cyclic trees.
|
|
588
|
+
def path(node1, node2)
|
|
589
|
+
raise IndexError, 'node1 not found' unless @pathway.graph[node1]
|
|
590
|
+
raise IndexError, 'node2 not found' unless @pathway.graph[node2]
|
|
591
|
+
return [ node1 ] if node1 == node2
|
|
592
|
+
step, path = @pathway.bfs_shortest_path(node1, node2)
|
|
593
|
+
unless path[0] == node1 and path[-1] == node2 then
|
|
594
|
+
raise NoPathError, 'node1 and node2 are not connected'
|
|
595
|
+
end
|
|
596
|
+
path
|
|
597
|
+
end
|
|
598
|
+
|
|
599
|
+
# Iterates over each edge from node1 to node2.
|
|
600
|
+
# The result is unspecified for cyclic trees.
|
|
601
|
+
def each_edge_in_path(node1, node2)
|
|
602
|
+
path = self.path(node1, node2)
|
|
603
|
+
source = path.shift
|
|
604
|
+
path.each do |target|
|
|
605
|
+
edge = self.get_edge(source, target)
|
|
606
|
+
yield source, target, edge
|
|
607
|
+
source = target
|
|
608
|
+
end
|
|
609
|
+
self
|
|
610
|
+
end
|
|
611
|
+
|
|
612
|
+
# Returns distance between node1 and node2.
|
|
613
|
+
# It would raise error if the edges didn't contain distance values.
|
|
614
|
+
# The result is unspecified for cyclic trees.
|
|
615
|
+
def distance(node1, node2)
|
|
616
|
+
distance = 0
|
|
617
|
+
self.each_edge_in_path(node1, node2) do |source, target, edge|
|
|
618
|
+
distance += get_edge_distance(edge)
|
|
619
|
+
end
|
|
620
|
+
distance
|
|
621
|
+
end
|
|
622
|
+
|
|
623
|
+
# Gets the parent node of the _node_.
|
|
624
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
625
|
+
# Returns an <code>Node</code> object or nil.
|
|
626
|
+
# The result is unspecified for cyclic trees.
|
|
627
|
+
def parent(node, root = nil)
|
|
628
|
+
root ||= @root
|
|
629
|
+
self.path(root, node)[-2]
|
|
630
|
+
end
|
|
631
|
+
|
|
632
|
+
# Gets the adjacent children nodes of the _node_.
|
|
633
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
634
|
+
# Returns an array of <code>Node</code>s.
|
|
635
|
+
# The result is unspecified for cyclic trees.
|
|
636
|
+
def children(node, root = nil)
|
|
637
|
+
root ||= @root
|
|
638
|
+
path = self.path(root, node)
|
|
639
|
+
result = self.adjacent_nodes(node)
|
|
640
|
+
result -= path
|
|
641
|
+
result
|
|
642
|
+
end
|
|
643
|
+
|
|
644
|
+
# Gets all descendent nodes of the _node_.
|
|
645
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
646
|
+
# Returns an array of <code>Node</code>s.
|
|
647
|
+
# The result is unspecified for cyclic trees.
|
|
648
|
+
def descendents(node, root = nil)
|
|
649
|
+
root ||= @root
|
|
650
|
+
distance, route = @pathway.breadth_first_search(root)
|
|
651
|
+
d = distance[node]
|
|
652
|
+
result = []
|
|
653
|
+
distance.each do |key, val|
|
|
654
|
+
if val > d then
|
|
655
|
+
x = key
|
|
656
|
+
while x = route[x]
|
|
657
|
+
if x == node then
|
|
658
|
+
result << key
|
|
659
|
+
break
|
|
660
|
+
end
|
|
661
|
+
break if distance[x] <= d
|
|
662
|
+
end
|
|
663
|
+
end
|
|
664
|
+
end
|
|
665
|
+
result
|
|
666
|
+
end
|
|
667
|
+
|
|
668
|
+
# If _node_ is nil, returns an array of
|
|
669
|
+
# all leaves (nodes connected with one edge).
|
|
670
|
+
# Otherwise, gets all descendent leaf nodes of the _node_.
|
|
671
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
672
|
+
# Returns an array of <code>Node</code>s.
|
|
673
|
+
# The result is unspecified for cyclic trees.
|
|
674
|
+
def leaves(node = nil, root = nil)
|
|
675
|
+
unless node then
|
|
676
|
+
nodes = []
|
|
677
|
+
self.each_node do |x|
|
|
678
|
+
nodes << x if self.out_degree(x) == 1
|
|
679
|
+
end
|
|
680
|
+
return nodes
|
|
681
|
+
else
|
|
682
|
+
root ||= @root
|
|
683
|
+
self.descendents(node, root).find_all do |x|
|
|
684
|
+
self.adjacent_nodes(x).size == 1
|
|
685
|
+
end
|
|
686
|
+
end
|
|
687
|
+
end
|
|
688
|
+
|
|
689
|
+
# Gets all ancestral nodes of the _node_.
|
|
690
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
691
|
+
# Returns an array of <code>Node</code>s.
|
|
692
|
+
# The result is unspecified for cyclic trees.
|
|
693
|
+
def ancestors(node, root = nil)
|
|
694
|
+
root ||= @root
|
|
695
|
+
(self.path(root, node) - [ node ]).reverse
|
|
696
|
+
end
|
|
697
|
+
|
|
698
|
+
# Gets the lowest common ancestor of the two nodes.
|
|
699
|
+
# If _root_ isn't specified or _root_ is <code>nil</code>, @root is used.
|
|
700
|
+
# Returns a <code>Node</code> object or nil.
|
|
701
|
+
# The result is unspecified for cyclic trees.
|
|
702
|
+
def lowest_common_ancestor(node1, node2, root = nil)
|
|
703
|
+
root ||= @root
|
|
704
|
+
distance, route = @pathway.breadth_first_search(root)
|
|
705
|
+
x = node1; r1 = []
|
|
706
|
+
begin; r1 << x; end while x = route[x]
|
|
707
|
+
x = node2; r2 = []
|
|
708
|
+
begin; r2 << x; end while x = route[x]
|
|
709
|
+
return (r1 & r2).first
|
|
710
|
+
end
|
|
711
|
+
|
|
712
|
+
# Returns total distance of all edges.
|
|
713
|
+
# It would raise error if some edges didn't contain distance values.
|
|
714
|
+
def total_distance
|
|
715
|
+
distance = 0
|
|
716
|
+
self.each_edge do |source, target, edge|
|
|
717
|
+
distance += get_edge_distance(edge)
|
|
718
|
+
end
|
|
719
|
+
distance
|
|
720
|
+
end
|
|
721
|
+
|
|
722
|
+
# Calculates distance matrix of given nodes.
|
|
723
|
+
# If _nodes_ is nil, or is ommited, it acts the same as
|
|
724
|
+
# tree.distance_matrix(tree.leaves).
|
|
725
|
+
# Returns a matrix object.
|
|
726
|
+
# The result is unspecified for cyclic trees.
|
|
727
|
+
# Note 1: The diagonal values of the matrix are 0.
|
|
728
|
+
# Note 2: If the distance cannot be calculated, nil will be set.
|
|
729
|
+
def distance_matrix(nodes = nil)
|
|
730
|
+
nodes ||= self.leaves
|
|
731
|
+
matrix = []
|
|
732
|
+
nodes.each_index do |i|
|
|
733
|
+
row = []
|
|
734
|
+
nodes.each_index do |j|
|
|
735
|
+
if i == j then
|
|
736
|
+
distance = 0
|
|
737
|
+
elsif r = matrix[j] and val = r[i] then
|
|
738
|
+
distance = val
|
|
739
|
+
else
|
|
740
|
+
distance = (self.distance(nodes[i], nodes[j]) rescue nil)
|
|
741
|
+
end
|
|
742
|
+
row << distance
|
|
743
|
+
end
|
|
744
|
+
matrix << row
|
|
745
|
+
end
|
|
746
|
+
Matrix.rows(matrix, false)
|
|
747
|
+
end
|
|
748
|
+
|
|
749
|
+
# Shows the adjacency matrix representation of the tree.
|
|
750
|
+
# It shows matrix only for given nodes.
|
|
751
|
+
# If _nodes_ is nil or is ommitted,
|
|
752
|
+
# it acts the same as tree.adjacency_matrix(tree.nodes).
|
|
753
|
+
# If a block is given, for each edge,
|
|
754
|
+
# it yields _source_, _target_, and _edge_, and
|
|
755
|
+
# uses the returned value of the block.
|
|
756
|
+
# Without blocks, it uses edge.
|
|
757
|
+
# Returns a matrix object.
|
|
758
|
+
def adjacency_matrix(nodes = nil,
|
|
759
|
+
default_value = nil,
|
|
760
|
+
diagonal_value = nil) #:yields: source, target, edge
|
|
761
|
+
nodes ||= self.nodes
|
|
762
|
+
size = nodes.size
|
|
763
|
+
hash = {}
|
|
764
|
+
nodes.each_with_index { |x, i| hash[x] = i }
|
|
765
|
+
# prepares an matrix
|
|
766
|
+
matrix = Array.new(size, nil)
|
|
767
|
+
matrix.collect! { |x| Array.new(size, default_value) }
|
|
768
|
+
(0...size).each { |i| matrix[i][i] = diagonal_value }
|
|
769
|
+
# fills the matrix from each edge
|
|
770
|
+
self.each_edge do |source, target, edge|
|
|
771
|
+
i_source = hash[source]
|
|
772
|
+
i_target = hash[target]
|
|
773
|
+
if i_source and i_target then
|
|
774
|
+
val = block_given? ? (yield source, target, edge) : edge
|
|
775
|
+
matrix[i_source][i_target] = val
|
|
776
|
+
matrix[i_target][i_source] = val
|
|
777
|
+
end
|
|
778
|
+
end
|
|
779
|
+
Matrix.rows(matrix, false)
|
|
780
|
+
end
|
|
781
|
+
|
|
782
|
+
# Removes all nodes that are not branches nor leaves.
|
|
783
|
+
# That is, removes nodes connected with exactly two edges.
|
|
784
|
+
# For each removed node, two adjacent edges are merged and
|
|
785
|
+
# a new edge are created.
|
|
786
|
+
# Returns removed nodes.
|
|
787
|
+
# Note that orphan nodes are still kept unchanged.
|
|
788
|
+
def remove_nonsense_nodes
|
|
789
|
+
hash = {}
|
|
790
|
+
self.each_node do |node|
|
|
791
|
+
hash[node] = true if @pathway.graph[node].size == 2
|
|
792
|
+
end
|
|
793
|
+
hash.each_key do |node|
|
|
794
|
+
adjs = @pathway.graph[node].keys
|
|
795
|
+
edges = @pathway.graph[node].values
|
|
796
|
+
new_edge = get_edge_merged(edges[0], edges[1])
|
|
797
|
+
@pathway.graph[adjs[0]].delete(node)
|
|
798
|
+
@pathway.graph[adjs[1]].delete(node)
|
|
799
|
+
@pathway.graph.delete(node)
|
|
800
|
+
@pathway.append(Bio::Relation.new(adjs[0], adjs[1], new_edge))
|
|
801
|
+
end
|
|
802
|
+
#@pathway.to_relations
|
|
803
|
+
@pathway.relations.reject! do |rel|
|
|
804
|
+
hash[rel.node[0]] or hash[rel.node[1]]
|
|
805
|
+
end
|
|
806
|
+
return hash.keys
|
|
807
|
+
end
|
|
808
|
+
|
|
809
|
+
# Insert a new node between adjacent nodes node1 and node2.
|
|
810
|
+
# The old edge between node1 and node2 are changed to the edge
|
|
811
|
+
# between new_node and node2.
|
|
812
|
+
# The edge between node1 and new_node is newly created.
|
|
813
|
+
#
|
|
814
|
+
# If new_distance is specified, the distance between
|
|
815
|
+
# node1 and new_node is set to new_distance, and
|
|
816
|
+
# distance between new_node and node2 is set to
|
|
817
|
+
# <code>tree.get_edge(node1, node2).distance - new_distance</code>.
|
|
818
|
+
#
|
|
819
|
+
# Returns self.
|
|
820
|
+
# If node1 and node2 are not adjacent, raises IndexError.
|
|
821
|
+
#
|
|
822
|
+
# If new_node already exists in the tree, the tree would become
|
|
823
|
+
# circular. In addition, if the edge between new_node and
|
|
824
|
+
# node1 (or node2) already exists, it will be erased.
|
|
825
|
+
def insert_node(node1, node2, new_node, new_distance = nil)
|
|
826
|
+
unless edge = self.get_edge(node1, node2) then
|
|
827
|
+
raise IndexError, 'nodes not found or two nodes are not adjacent'
|
|
828
|
+
end
|
|
829
|
+
new_edge = Edge.new(new_distance)
|
|
830
|
+
self.remove_edge(node1, node2)
|
|
831
|
+
self.add_edge(node1, new_node, new_edge)
|
|
832
|
+
if new_distance and old_distance = get_edge_distance(edge) then
|
|
833
|
+
old_distance -= new_distance
|
|
834
|
+
begin
|
|
835
|
+
edge.distance = old_distance
|
|
836
|
+
rescue NoMethodError
|
|
837
|
+
edge = old_distance
|
|
838
|
+
end
|
|
839
|
+
end
|
|
840
|
+
self.add_edge(new_node, node2, edge)
|
|
841
|
+
self
|
|
842
|
+
end
|
|
843
|
+
end #class Tree
|
|
844
|
+
end #module Bio
|
|
845
|
+
|
|
846
|
+
#---
|
|
847
|
+
# temporary added
|
|
848
|
+
#+++
|
|
849
|
+
require 'bio/db/newick'
|
|
850
|
+
|