jandot-bio 1.2.1

data/lib/bio/shell/object.rb

@@ -0,0 +1,71 @@
+#
+# = bio/shell/object.rb - Object extension for the BioRuby shell
+#
+# Copyright::   Copyright (C) 2006
+#               Nobuya Tanaka <t@chemruby.org>,
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: object.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
+#
+
+require 'pp'
+require 'cgi'
+require 'yaml'
+
+### Object extention
+
+class Object
+  # Couldn't work for Fixnum (Marshal)
+  attr_accessor :memo
+
+  def output(format = :yaml)
+    case format
+    when :yaml
+      self.to_yaml
+    when :html
+      format_html
+    when :inspect
+      format_pp
+    when :png
+      # *TODO*
+    when :svg
+      # *TODO*
+    when :graph
+      # *TODO* (Gruff, RSRuby etc.)
+    else
+      #self.inspect.to_s.fold(80)
+      self.to_s
+    end
+  end
+
+  private
+
+  def format_html
+    "<pre>#{CGI.escapeHTML(format_pp)}</pre>"
+  end
+
+  def format_pp
+    str = ""
+    PP.pp(self, str)
+    return str
+  end
+
+end
+
+class Hash
+
+  private
+
+  def format_html
+    html = ""
+    html += "<table>"
+    @data.each do |k, v|
+      html += "<tr><td>#{k}</td><td>#{v}</td></tr>"
+    end
+    html += "</table>"
+    return html
+  end
+
+end
+

data/lib/bio/shell/plugin/blast.rb

@@ -0,0 +1,42 @@
+#
+# = bio/shell/plugin/blast.rb - plugin for BLAST services
+#
+# Copyright::   Copyright (C) 2006
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: blast.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio::Shell
+
+  private
+
+  # GenomeNet
+
+  def keggblast(query)
+    server = Bio::Blast.remote("blastp", "genes", "", "genomenet_tab")
+
+    if query[/^>/]
+      data = Bio::FastaFormat.new(query)
+      desc = data.definition
+      tmp = getseq(data.seq)
+    else
+      desc = "query"
+      tmp = getseq(query)
+    end
+    
+    if tmp.respond_to?(:translate)
+      aaseq = tmp.translate
+    else
+      aaseq = tmp
+    end
+
+    fasta = aaseq.to_fasta(desc, 60)
+    result = server.query(fasta)
+    puts server.output
+    return result
+  end
+
+end
+

data/lib/bio/shell/plugin/codon.rb

@@ -0,0 +1,218 @@
+#
+# = bio/shell/plugin/codon.rb - plugin for the codon table
+#
+# Copyright::   Copyright (C) 2005
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: codon.rb,v 1.16 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio::Shell
+
+  class ColoredCodonTable
+
+    @@properties = {
+      :basic => %w( H K R ),
+      :polar => %w( S T Y Q N S ),
+      :acidic => %w( D E ),
+      :nonpolar => %w( F L I M V P A C W G ),
+      :stop => %w( * ),
+    }
+
+    def initialize(number, cuhash = nil)
+      @aacode = Bio::AminoAcid.names
+      @table  = Bio::CodonTable[number]
+      @number = number
+      @cuhash = cuhash
+      setup_colors
+      if Bio::Shell.config[:color]
+        generate_colored_text
+      else
+        generate_mono_text
+      end
+    end
+    attr_reader :table
+
+    def setup_colors
+      c = Bio::Shell.colors
+
+      @colors = {
+        :text		=> c[:none],
+        :aa		=> c[:green],
+        :start		=> c[:red],
+        :stop		=> c[:red],
+        :basic		=> c[:cyan],
+        :polar		=> c[:blue],
+        :acidic		=> c[:magenta],
+        :nonpolar	=> c[:yellow],
+      }
+    end
+
+    def generate_mono_text
+      @table.each do |codon, aa|
+        if aa == '*'
+          code = 'STOP'
+          aa = '' unless @cuhash
+        else
+          code = @aacode[aa]
+        end
+        if @cuhash
+          percent = @cuhash[codon].to_s.rjust(6)
+          eval("@#{codon} = '#{aa}#{percent}'")
+        else
+          eval("@#{codon} = ' #{code} #{aa} '")
+        end
+      end
+
+      @hydrophilic = [
+        @@properties[:basic].join(" "), "(basic),",
+        @@properties[:polar].join(" "), "(polar),",
+        @@properties[:acidic].join(" "), "(acidic)",
+      ].join(" ")
+      @hydrophobic = @@properties[:nonpolar].join(" ") + " (nonpolar)"
+    end
+
+    def generate_colored_text
+      @table.each do |codon, aa|
+        property, = @@properties.detect {|key, list| list.include?(aa)}
+
+        if aa == '*'
+          if @cuhash
+            color_code = "#{@colors[:stop]}STOP"
+            color_aa = "#{@colors[:stop]}#{aa}"
+          else
+            color_code = "#{@colors[:stop]}STP"
+            case codon
+            when 'tga'
+              color_aa = "#{@colors[:text]}U"
+            when 'tag'
+              color_aa = "#{@colors[:text]}O"
+            else
+              color_aa = "#{@colors[:text]}*"
+            end
+          end
+        else
+          color_code = "#{@colors[property]}#{@aacode[aa]}"
+          if @table.start_codon?(codon)
+            if @cuhash
+              color_aa = "#{@colors[:aa]}#{aa}"
+            else
+              color_aa = "#{@colors[:start]}#{aa}"
+            end
+          else
+            if @cuhash
+              color_aa = "#{@colors[property]}#{aa}"
+            else
+              color_aa = "#{@colors[:aa]}#{aa}"
+            end
+          end
+        end
+
+        if @cuhash
+          percent = @cuhash[codon].to_s.rjust(6)
+          eval("@#{codon} = '#{color_aa}#{@colors[:text]}#{percent}'")
+        else
+          eval("@#{codon} = ' #{color_code} #{color_aa}#{@colors[:text]} '")
+        end
+      end
+
+      @hydrophilic = [
+        "#{@colors[:basic]}basic#{@colors[:text]},",
+        "#{@colors[:polar]}polar#{@colors[:text]},",
+        "#{@colors[:acidic]}acidic#{@colors[:text]}"
+      ].join(" ")
+      @hydrophobic = "#{@colors[:nonpolar]}nonpolar"
+    end
+
+    def output
+      header = <<-END
+        #
+        # = Codon table #{@number} : #{@table.definition}
+        #
+        #   hydrophilic: #{@hydrophilic}
+        #   hydrophobic: #{@hydrophobic}
+      END
+      table = <<-END
+        #
+        # *---------------------------------------------*
+        # |       |              2nd              |     |
+        # |  1st  |-------------------------------| 3rd |
+        # |       |  U    |  C    |  A    |  G    |     |
+        # |-------+-------+-------+-------+-------+-----|
+        # | U   U |#{@ttt}|#{@tct}|#{@tat}|#{@tgt}|  u  |
+        # | U   U |#{@ttc}|#{@tcc}|#{@tac}|#{@tgc}|  c  |
+        # | U   U |#{@tta}|#{@tca}|#{@taa}|#{@tga}|  a  |
+        # |  UUU  |#{@ttg}|#{@tcg}|#{@tag}|#{@tgg}|  g  |
+        # |-------+-------+-------+-------+-------+-----|
+        # |  CCCC |#{@ctt}|#{@cct}|#{@cat}|#{@cgt}|  u  |
+        # | C     |#{@ctc}|#{@ccc}|#{@cac}|#{@cgc}|  c  |
+        # | C     |#{@cta}|#{@cca}|#{@caa}|#{@cga}|  a  |
+        # |  CCCC |#{@ctg}|#{@ccg}|#{@cag}|#{@cgg}|  g  |
+        # |-------+-------+-------+-------+-------+-----|
+        # |   A   |#{@att}|#{@act}|#{@aat}|#{@agt}|  u  |
+        # |  A A  |#{@atc}|#{@acc}|#{@aac}|#{@agc}|  c  |
+        # | AAAAA |#{@ata}|#{@aca}|#{@aaa}|#{@aga}|  a  |
+        # | A   A |#{@atg}|#{@acg}|#{@aag}|#{@agg}|  g  |
+        # |-------+-------+-------+-------+-------+-----|
+        # |  GGGG |#{@gtt}|#{@gct}|#{@gat}|#{@ggt}|  u  |
+        # | G     |#{@gtc}|#{@gcc}|#{@gac}|#{@ggc}|  c  |
+        # | G GGG |#{@gta}|#{@gca}|#{@gaa}|#{@gga}|  a  |
+        # |  GG G |#{@gtg}|#{@gcg}|#{@gag}|#{@ggg}|  g  |
+        # *---------------------------------------------*
+        #
+      END
+      if @cuhash
+        text = table
+      else
+        text = header + table
+      end
+      if Bio::Shell.config[:color]
+        text.gsub(/^\s+#/, @colors[:text])
+      else
+        text.gsub(/^\s+#/, '')
+      end
+    end
+
+  end
+
+  private
+
+  def codontable(num = 1, codon_usage = nil)
+    cct = ColoredCodonTable.new(num, codon_usage)
+    if codon_usage
+      return cct
+    else
+      puts cct.output
+      return cct.table
+    end
+  end
+
+  def codontables
+    tables = Bio::CodonTable::DEFINITIONS
+    tables.sort.each do |i, definition|
+      puts "#{i}\t#{definition}"
+    end
+    return tables
+  end
+
+  def aminoacids
+    names = Bio::AminoAcid.names
+    names.sort.each do |aa, code|
+      if aa.length == 1
+        puts "#{aa}\t#{code}\t#{names[code]}"
+      end
+    end
+    return names
+  end
+
+  def nucleicacids
+    names = Bio::NucleicAcid.names
+    %w(a t g c u r y w s k m b v h d n).each do |base|
+      puts "#{base}\t#{names[base]}\t#{names[base.upcase]}"
+    end
+    return names
+  end
+
+end
+

data/lib/bio/shell/plugin/das.rb

@@ -0,0 +1,58 @@
+#
+# = bio/shell/plugin/keggdas.rb - plugin for KEGG DAS
+#
+# Copyright::   Copyright (C) 2006
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: das.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio
+
+  class DAS
+    def list_sequences
+      result = ""
+      self.get_dsn.each do |dsn|
+        src = dsn.source_id
+        self.get_entry_points(src).each do |ep|
+          data = [src, ep.entry_id, ep.start.to_i, ep.stop.to_i, "# #{ep.description}"].join("\t") + "\n"
+          puts data
+          result += data
+        end
+      end
+      return result
+    end
+  end
+
+end
+
+module Bio::Shell
+
+  private
+
+  # http://www.biodas.org/
+  # http://www.dasregistry.org/
+
+  def das(url = nil)
+    if url
+      @das = Bio::DAS.new(url)
+    else
+      @das ||= keggdas
+    end
+  end
+
+  def keggdas(url = "http://das.hgc.jp/cgi-bin/")
+    das(url)
+  end
+
+  def ensembl(url = "http://das.ensembl.org/")
+    das(url)
+  end
+
+  def wormbase(url = "http://www.wormbase.org/db/")
+    das(url)
+  end
+
+end
+

data/lib/bio/shell/plugin/emboss.rb

@@ -0,0 +1,23 @@
+#
+# = bio/shell/plugin/emboss.rb - methods to use EMBOSS
+#
+# Copyright::   Copyright (C) 2005
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: emboss.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio::Shell
+
+  private
+
+  def seqret(usa)
+    Bio::EMBOSS.seqret(usa)
+  end
+
+  def entret(usa)
+    Bio::EMBOSS.entret(usa)
+  end
+
+end

data/lib/bio/shell/plugin/entry.rb

@@ -0,0 +1,105 @@
+#
+# = bio/shell/plugin/entry.rb - extract entry and sequence
+#
+# Copyright::   Copyright (C) 2005
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: entry.rb,v 1.10 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio::Shell
+
+  private
+
+  # Read a text file and collect the first word of each line in array
+  def readlist(filename)
+    list = []
+    File.open(filename).each do |line|
+      list << line[/^\S+/]
+    end
+    return list
+  end
+
+  # Obtain a Bio::Sequence::NA (DNA) or a Bio::Sequence::AA (Amino Acid)
+  # sequence from
+  #   * String     -- "atgcatgc" or "MQKKP"
+  #   * IO         -- io = IO.popen("gzip -dc db.gz") (first entry only)
+  #   * "filename" -- "gbvrl.gbk" (first entry only)
+  #   * "db:entry" -- "embl:BUM"  (entry is retrieved by the ent method)
+  def getseq(arg)
+    seq = ""
+    if arg.kind_of?(Bio::Sequence)
+      seq = arg
+    elsif arg.respond_to?(:gets) or File.exists?(arg)
+      ent = flatauto(arg)
+    elsif arg[/:/]
+      ent = getobj(arg)
+    else
+      tmp = arg
+    end
+
+    if ent.respond_to?(:seq)
+      tmp = ent.seq
+    elsif ent.respond_to?(:naseq)
+      #seq = ent.naseq
+      tmp = ent.naseq
+    elsif ent.respond_to?(:aaseq)
+      #seq = ent.aaseq
+      tmp = ent.aaseq
+    end
+
+    if tmp and tmp.is_a?(String) and not tmp.empty?
+      #seq = Bio::Sequence.auto(tmp).seq
+      seq = Bio::Sequence.auto(tmp)
+    end
+    return seq
+  end
+
+  # Obtain a database entry from
+  #   * IO          -- IO object (first entry only)
+  #   * "filename"  -- local file (first entry only)
+  #   * "db:entry"  -- local BioFlat, OBDA, EMBOSS, KEGG API
+  def getent(arg)
+    entry = ""
+    db, entry_id = arg.to_s.strip.split(/:/)
+
+    # local file
+    if arg.respond_to?(:gets) or File.exists?(arg)
+      puts "Retrieving entry from file (#{arg})"
+      entry = flatfile(arg)
+
+    # BioFlat in ./.bioruby/bioflat/ or ~/.bioinformatics/.bioruby/bioflat/
+    elsif Bio::Shell.find_flat_dir(db)
+      puts "Retrieving entry from local BioFlat database (#{arg})"
+      entry = flatsearch(db, entry_id)
+
+    # OBDA in ~/.bioinformatics/seqdatabase.ini
+    elsif obdadbs.include?(db)
+      puts "Retrieving entry from OBDA (#{arg})"
+      entry = obdaentry(db, entry_id)
+
+    else
+      # EMBOSS USA in ~/.embossrc
+      str = entret(arg)
+      if $?.exitstatus == 0 and str.length != 0
+        puts "Retrieving entry from EMBOSS (#{arg})"
+        entry = str
+
+      # KEGG API at http://www.genome.jp/kegg/soap/
+      else
+        puts "Retrieving entry from KEGG API (#{arg})"
+        entry = bget(arg)
+      end
+    end
+
+    return entry
+  end
+
+  # Obtain a parsed object from sources that ent() supports.
+  def getobj(arg)
+    str = getent(arg)
+    flatparse(str)
+  end
+
+end