bio 1.0.0 → 1.1.0

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

@@ -3,9 +3,9 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License::     Ruby's
+# License::     The Ruby License
 #
-# $Id: emboss.rb,v 1.2 2006/02/09 20:48:53 k Exp $
+# $Id: emboss.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
 #
 
 module Bio::Shell

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

@@ -3,30 +3,38 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License::     Ruby's
+# License::     The Ruby License
 #
-# $Id: entry.rb,v 1.8 2006/02/27 09:37:14 k Exp $
+# $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 seq(arg)
+  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[/:/]
-      str = ent(arg)
-      ent = flatparse(str)
+      ent = getobj(arg)
     else
       tmp = arg
     end
@@ -34,13 +42,16 @@ module Bio::Shell
     if ent.respond_to?(:seq)
       tmp = ent.seq
     elsif ent.respond_to?(:naseq)
-      seq = ent.naseq
+      #seq = ent.naseq
+      tmp = ent.naseq
     elsif ent.respond_to?(:aaseq)
-      seq = ent.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).seq
+      seq = Bio::Sequence.auto(tmp)
     end
     return seq
   end
@@ -49,7 +60,7 @@ module Bio::Shell
   #   * IO          -- IO object (first entry only)
   #   * "filename"  -- local file (first entry only)
   #   * "db:entry"  -- local BioFlat, OBDA, EMBOSS, KEGG API
-  def ent(arg)
+  def getent(arg)
     entry = ""
     db, entry_id = arg.to_s.strip.split(/:/)
 
@@ -86,8 +97,8 @@ module Bio::Shell
   end
 
   # Obtain a parsed object from sources that ent() supports.
-  def obj(arg)
-    str = ent(arg)
+  def getobj(arg)
+    str = getent(arg)
     flatparse(str)
   end
 

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

@@ -3,9 +3,9 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License:      Ruby's
+# License::     The Ruby License
 #
-# $Id: flatfile.rb,v 1.12 2006/02/09 20:48:53 k Exp $
+# $Id: flatfile.rb,v 1.13 2007/04/05 23:45:11 trevor Exp $
 #
 
 module Bio::Shell

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

@@ -3,18 +3,21 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License::     Ruby's
+# License::     The Ruby License
 #
-# $Id: keggapi.rb,v 1.10 2006/02/09 20:48:53 k Exp $
+# $Id: keggapi.rb,v 1.12 2007/04/05 23:35:41 trevor Exp $
 #
 
 module Bio::Shell
 
   module Private
+
+    module_function
+
     def keggapi_definition2tab(list)
       ary = []
       list.each do |entry|
-        ary << "#{entry.entry_id}:\t#{entry.definition}"
+        ary << "#{entry.entry_id}\t#{entry.definition}"
       end
       return ary
     end
@@ -49,17 +52,20 @@ module Bio::Shell
     if block_given?
       yield result
     else
+      puts result
       return result
     end
   end
 
   def btit(str)
     result = keggapi.btit(str)
+    puts result
     return result
   end
 
   def bconv(str)
     result = keggapi.bconv(str)
+    puts result
     return result
   end
 
@@ -67,25 +73,26 @@ module Bio::Shell
 
   def keggdbs
     list = keggapi.list_databases
-    result = Bio::Shell.keggapi_definition2tab(list).join("\n")
+    result = Bio::Shell::Private.keggapi_definition2tab(list).join("\n")
     puts result
     return list.map {|x| x.entry_id}
   end
 
   def keggorgs
     list = keggapi.list_organisms
-    result = Bio::Shell.keggapi_definition2tab(list).sort.join("\n")
+    result = Bio::Shell::Private.keggapi_definition2tab(list).sort.join("\n")
     puts result
     return list.map {|x| x.entry_id}
   end
 
   def keggpathways(org = "map")
     list = keggapi.list_pathways(org)
-    result = Bio::Shell.keggapi_definition2tab(list).join("\n")
+    result = Bio::Shell::Private.keggapi_definition2tab(list).join("\n")
     puts result
     return list.map {|x| x.entry_id}
   end
 
+  # use KEGG DAS insetad
   def kegggenomeseq(org)
     result = ""
     require 'net/ftp'

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

@@ -4,9 +4,9 @@
 # Copyright::   Copyright (C) 2003, 2005
 #               Natsuhiro Ichinose <ichinose@genome.ist.i.kyoto-u.ac.jp>,
 #               Toshiaki Katayama <k@bioruby.org>
-# License::     Ruby's
+# License::     The Ruby License
 #
-# $Id: midi.rb,v 1.7 2006/02/09 20:48:53 k Exp $
+# $Id: midi.rb,v 1.9 2007/04/05 23:35:41 trevor Exp $
 #
 
 #--
@@ -420,11 +420,11 @@ if $0 == __FILE__
 
   seq_file = ARGV.shift
   mid_file = ARGV.shift
-  style    = ARGV.shift
 
   Bio::FlatFile.auto(seq_file) do |ff|
     ff.each do |f|
-      midi(f.naseq[0..1000], mid_file, style)
+      midifile(mid_file, f.naseq[0..1000])
     end
   end
 end
+

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

@@ -3,9 +3,9 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License:      Ruby's
+# License::     The Ruby License 
 #
-# $Id: obda.rb,v 1.9 2006/02/09 20:48:53 k Exp $
+# $Id: obda.rb,v 1.10 2007/04/05 23:45:11 trevor Exp $
 #
 
 module Bio::Shell

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

@@ -0,0 +1,56 @@
+#
+# = bio/shell/plugin/psort.rb - plugin for PSORT
+#
+# Copyright::   Copyright (C) 2006
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: psort.rb,v 1.3 2007/04/05 23:35:41 trevor Exp $
+#
+
+module Bio::Shell
+
+  private
+
+  def psort1(str)
+    seq = getseq(str)
+    if seq.is_a?(Bio::Sequence::NA)
+      seq = seq.translate
+    end
+
+    psort = Bio::PSORT::PSORT1.imsut
+    fasta = seq.to_fasta
+
+    results = psort.exec(fasta).final_result
+    results.each do |result|
+      puts "#{result["certainty"].to_f*100.0}\t#{result["prediction"]}"
+    end
+    return results.first["prediction"]
+  end
+
+  def psort2(str)
+    seq = getseq(str)
+    if seq.is_a?(Bio::Sequence::NA)
+      seq = seq.translate
+    end
+
+    psort = Bio::PSORT::PSORT2.imsut
+    fasta = seq.to_fasta
+
+    results = psort.exec(fasta).prob.sort_by{|x, y| y}.reverse
+    results.each do |loc, prob|
+      next if prob <= 0.0
+      puts "#{prob}\t#{Bio::PSORT::PSORT2::SclNames[loc]}"
+    end
+    return results.first.first
+  end
+
+  def psort2locations
+    names = Bio::PSORT::PSORT2::SclNames
+    names.sort.each do |loc, desc|
+      puts "#{loc}\t#{desc}"
+    end
+    return names
+  end
+end
+

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

@@ -3,9 +3,9 @@
 #
 # Copyright::   Copyright (C) 2005
 #               Toshiaki Katayama <k@bioruby.org>
-# License::     Ruby's
+# License::     The Ruby License
 #
-# $Id: seq.rb,v 1.17 2006/02/09 20:48:53 k Exp $
+# $Id: seq.rb,v 1.21 2007/04/05 23:35:41 trevor Exp $
 #
 
 module Bio::Shell
@@ -17,7 +17,7 @@ module Bio::Shell
     if str.kind_of?(Bio::Sequence)
       seq = str
     else
-      seq = seq(str)
+      seq = getseq(str)
     end
 
     if seq.is_a?(Bio::Sequence::AA)
@@ -41,12 +41,21 @@ module Bio::Shell
   end
 
 
+  def sixtrans(str)
+    seq = getseq(str)
+    [ 1, 2, 3, -1, -2, -3 ].each do |frame|
+      title = "Translation #{frame.to_s.rjust(2)}"
+      puts seq.translate(frame).to_fasta(title, 60)
+    end
+  end
+
+
   # Displays some basic properties of the sequence.
   def seqstat(str)
     max = 150
-    seq = seq(str)
+    seq = getseq(str)
     rep = "\n* * * Sequence statistics * * *\n\n"
-    if seq.respond_to?(:complement)
+    if seq.moltype == Bio::Sequence::NA
       fwd = seq
       rev = seq.complement
       if seq.length > max
@@ -123,12 +132,12 @@ module Bio::Shell
   # Displays a DNA sequence by ascii art in B-type double helix.
   # Argument need to be at least 16 bases in length.
   def doublehelix(str)
-    seq = seq(str)
-    if str.length < 16
+    seq = getseq(str)
+    if seq.length < 16
       warn "Error: Sequence must be longer than 16 bases."
       return
     end
-    if ! seq.respond_to?(:complement)
+    if seq.moltype != Bio::Sequence::NA
       warn "Error: Sequence must be a DNA sequence."
       return
     end
@@ -151,6 +160,24 @@ end
 
 class String
 
+  def step(window_size)
+    i = 0
+    0.step(self.length - window_size, window_size) do |i|
+      yield self[i, window_size]
+    end
+    yield self[i + window_size .. -1] if i + window_size < self.length
+  end
+
+  def skip(window_size, step_size = 1)
+    i = 0
+    0.step(self.length - window_size, step_size) do |i|
+      yield [self[i, window_size], i + 1, i + window_size]
+    end
+    from = i + step_size
+    to  = [self.length, i + step_size + window_size].min
+    yield [self[from, window_size], from + 1, to] if from + 1 <= to
+  end
+
   def to_naseq
     Bio::Sequence::NA.new(self)
   end

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

@@ -0,0 +1,87 @@
+#
+# = bio/shell/plugin/soap.rb - web services
+#
+# Copyright::   Copyright (C) 2006
+#               Toshiaki Katayama <k@bioruby.org>
+# License::     Ruby's
+#
+# $Id: soap.rb,v 1.1 2007/07/09 11:17:09 k Exp $
+#
+
+module Bio::Shell
+
+  private
+
+  def ncbisoap(wsdl = nil)
+    if wsdl
+      @ncbisoap = Bio::NCBI::SOAP.new(wsdl)
+    else
+      @ncbisoap ||= Bio::NCBI::SOAP.new
+    end
+    return @ncbisoap
+  end
+
+  def ebisoap(wsdl = nil)
+    case wsdl
+    when :ipscan
+      @ebisoap = Bio::EBI::SOAP::InterProScan.new(wsdl)
+    when :emboss
+      @ebisoap = Bio::EBI::SOAP::Emboss.new(wsdl)
+    when :clustalw
+      @ebisoap = Bio::EBI::SOAP::ClustalW.new(wsdl)
+    when :tcoffee
+      @ebisoap = Bio::EBI::SOAP::TCoffee.new(wsdl)
+    when :muscle
+      @ebisoap = Bio::EBI::SOAP::Muscle.new(wsdl)
+    when :fasta
+      @ebisoap = Bio::EBI::SOAP::Fasta.new(wsdl)
+    when :wublast
+      @ebisoap = Bio::EBI::SOAP::WUBlast.new(wsdl)
+    when :mpsrch
+      @ebisoap = Bio::EBI::SOAP::MPsrch.new(wsdl)
+    when :scanps
+      @ebisoap = Bio::EBI::SOAP::ScanPS.new(wsdl)
+    when :msd
+      @ebisoap = Bio::EBI::SOAP::MSD.new(wsdl)
+    when :ontology
+      @ebisoap = Bio::EBI::SOAP::Ontology.new(wsdl)
+    when :citation
+      @ebisoap = Bio::EBI::SOAP::Citation.new(wsdl)
+    when /^http/
+      @ebisoap = Bio::EBI::SOAP.new(wsdl)
+    else
+      @ebisoap ||= Bio::EBI::SOAP.new
+    end
+    return @ebisoap
+  end
+
+  def ddbjsoap(wsdl = nil)
+    case wsdl
+    when :blast
+      @ddbjsoap = Bio::DDBJ::XML::Blast.new
+    when :fasta
+      @ddbjsoap = Bio::DDBJ::XML::Fasta.new
+    when :clustalw
+      @ddbjsoap = Bio::DDBJ::XML::ClustalW.new
+    when :ddbj
+      @ddbjsoap = Bio::DDBJ::XML::DDBJ.new
+    when :gib
+      @ddbjsoap = Bio::DDBJ::XML::Gib.new
+    when :gtop
+      @ddbjsoap = Bio::DDBJ::XML::Gtop.new
+    when :pml
+      @ddbjsoap = Bio::DDBJ::XML::PML.new
+    when :srs
+      @ddbjsoap = Bio::DDBJ::XML::SRS.new
+    when :txsearch
+      @ddbjsoap = Bio::DDBJ::XML::TxSearch.new
+    when /^http/
+      @ddbjsoap = Bio::DDBJ::XML.new(wsdl)
+    else
+      @ddbjsoap ||= Bio::DDBJ::XML.new
+    end
+    return @ddbjsoap
+  end
+
+end
+