bio 1.2.1 → 1.3.0

data/lib/bio/db/embl/common.rb

@@ -5,7 +5,7 @@
 #               Mitsuteru C. Nakao <n@bioruby.org>
 # License::     The Ruby License
 #
-# $Id: common.rb,v 1.12 2007/04/05 23:35:40 trevor Exp $
+# $Id: common.rb,v 1.12.2.5 2008/05/07 12:22:10 ngoto Exp $
 #
 # == Description
 #
@@ -73,6 +73,7 @@
 
 require 'bio/db'
 require 'bio/reference'
+require 'bio/compat/references'
 
 module Bio
 class EMBLDB
@@ -270,33 +271,48 @@ module Common
   def references
     unless @data['references']
       ary = self.ref.map {|ent|
-        hash = Hash.new('')
+        hash = Hash.new
         ent.each {|key, value|
           case key
+          when 'RN'
+            if /\[(\d+)\]/ =~ value.to_s
+              hash['embl_gb_record_number'] = $1.to_i
+            end
+          when 'RC'
+            unless value.to_s.strip.empty?
+              hash['comments'] ||= []
+              hash['comments'].push value
+            end
+          when 'RP'
+            hash['sequence_position'] = value
           when 'RA'
-            hash['authors'] = value.split(/, /)
+            a = value.split(/\, /)
+            a.each do |x|
+              x.sub!(/( [^ ]+)\z/, ",\\1")
+            end
+            hash['authors'] = a
           when 'RT'
             hash['title'] = value
           when 'RL'
-            if value =~ /(.*) (\d+) \((\d+)\), (\d+-\d+) \((\d+)\)$/
-              hash['journal'] = $1
+            if /(.*) (\d+) *(\(([^\)]+)\))?(\, |\:)([a-zA-Z\d]+\-[a-zA-Z\d]+) *\((\d+)\)\.?\z/ =~ value.to_s
+              hash['journal'] = $1.rstrip
               hash['volume']  = $2
-              hash['issue']   = $3
-              hash['pages']   = $4
-              hash['year']    = $5
+              hash['issue']   = $4
+              hash['pages']   = $6
+              hash['year']    = $7
             else
               hash['journal'] = value
             end
-          when 'RX'  # PUBMED, MEDLINE
-            value.split('.').each {|item|
-              tag, xref = item.split(/; /).map {|i| i.strip }
+          when 'RX'  # PUBMED, DOI, (AGRICOLA)
+            value.split(/\. /).each {|item|
+              tag, xref = item.split(/\; /).map {|i| i.strip.sub(/\.\z/, '') }
               hash[ tag.downcase ]  = xref
             }
           end
         }
         Reference.new(hash)
       }
-      @data['references'] = References.new(ary)
+      @data['references'] = ary.extend(Bio::References::BackwardCompatibility)
     end
     @data['references']
   end

data/lib/bio/db/embl/embl.rb

@@ -2,10 +2,12 @@
 # = bio/db/embl/embl.rb - EMBL database class
 #
 # 
-# Copyright::   Copyright (C) 2001-2007 Mitsuteru C. Nakao <n@bioruby.org>
+# Copyright::   Copyright (C) 2001-2007
+#               Mitsuteru C. Nakao <n@bioruby.org>
+#               Jan Aerts <jan.aerts@bbsrc.ac.uk>
 # License::     The Ruby License
 #
-# $Id: embl.rb,v 1.29 2007/04/05 23:35:40 trevor Exp $
+# $Id: embl.rb,v 1.29.2.7 2008/06/17 16:04:36 ngoto Exp $
 #
 # == Description
 #
@@ -29,8 +31,13 @@
 #   http://www.ebi.ac.uk/embl/Documentation/User_manual/usrman.html
 #
 
+require 'date'
 require 'bio/db'
 require 'bio/db/embl/common'
+require 'bio/compat/features'
+require 'bio/compat/references'
+require 'bio/sequence'
+require 'bio/sequence/dblink'
 
 module Bio
 class EMBL < EMBLDB
@@ -120,6 +127,14 @@ class EMBL < EMBLDB
   end
   alias molecule_type molecule
 
+  def data_class
+    id_line('DATA_CLASS')
+  end
+  
+  def topology
+    id_line('TOPOLOGY')
+  end
+  
   # returns DIVISION in the ID line.
   # * Bio::EMBL#division -> String
   def division
@@ -221,8 +236,8 @@ class EMBL < EMBLDB
   # RN RC RP RX RA RT RL
   #
   # Bio::EMBLDB#ref
-
-
+  
+  
   ##
   # DR Line; defabases cross-regerence (>=0)
   # "DR  database_identifier; primary_identifier; secondary_identifier."
@@ -246,7 +261,6 @@ class EMBL < EMBLDB
   # FT Line; feature table data (>=0)
   def ft
     unless @data['FT']
-      @data['FT'] = Array.new
       ary = Array.new
       in_quote = false
       @orig['FT'].each_line do |line|
@@ -276,7 +290,7 @@ class EMBL < EMBLDB
         parse_qualifiers(subary)
       end
 
-      @data['FT'] = Features.new(ary)
+      @data['FT'] = ary.extend(Bio::Features::BackwardCompatibility)
     end
     if block_given?
       @data['FT'].each do |feature|
@@ -311,9 +325,9 @@ class EMBL < EMBLDB
   #
   # CC Line; comments of notes (>=0)
   def cc
-    get('CC')
+    get('CC').to_s.gsub(/^CC   /, '')
   end
-
+  alias comment cc
 
   ##
   # XX Line; spacer line (many)
@@ -355,13 +369,96 @@ class EMBL < EMBLDB
   # @orig[''] as sequence
   # bb Line; (blanks) sequence data (>=1)
   def seq
-    Sequence::NA.new( fetch('').gsub(/ /,'').gsub(/\d+/,'') )
+    Bio::Sequence::NA.new( fetch('').gsub(/ /,'').gsub(/\d+/,'') )
   end
   alias naseq seq
   alias ntseq seq
 
+  #--
   # // Line; termination line (end; 1/entry)
+  #++
+
+  # modified date. Returns Date object, String or nil.
+  def date_modified
+    parse_date(self.dt['updated'])
+  end
+
+  # created date. Returns Date object, String or nil.
+  def date_created
+    parse_date(self.dt['created'])
+  end
+
+  # release number when last updated
+  def release_modified
+    parse_release_version(self.dt['updated'])[0]
+  end
+
+  # release number when created
+  def release_created
+    parse_release_version(self.dt['created'])[0]
+  end
 
+  # entry version number numbered by EMBL
+  def entry_version
+    parse_release_version(self.dt['updated'])[1]
+  end
+
+  # parse date string. Returns Date object.
+  def parse_date(str)
+    begin
+      Date.parse(str)
+    rescue ArgumentError, TypeError, NoMethodError, NameError
+      str
+    end
+  end
+  private :parse_date
+
+  # extracts release and version numbers from DT line
+  def parse_release_version(str)
+    return [ nil, nil ] unless str
+    a = str.split(/[\(\,\)]/)
+    dstr = a.shift
+    rel = nil
+    ver = nil
+    a.each do |x|
+      case x
+      when /Rel\.\s*(.+)/
+        rel = $1.strip
+      when /Version\s*(.+)/
+        ver = $1.strip
+      end
+    end
+    [ rel, ver ]
+  end
+  private :parse_release_version
+
+  # database references (DR).
+  # Returns an array of Bio::Sequence::DBLink objects.
+  def dblinks
+    get('DR').split(/\n/).collect { |x|
+      Bio::Sequence::DBLink.parse_embl_DR_line(x)
+    }
+  end
+
+  # species
+  def species
+    self.fetch('OS')
+  end
+
+  # taxonomy classfication
+  alias classification oc
+
+  # features
+  alias features ft
+
+
+  # converts the entry to Bio::Sequence object
+  # ---
+  # *Arguments*::
+  # *Returns*:: Bio::Sequence object
+  def to_biosequence
+    Bio::Sequence.adapter(self, Bio::Sequence::Adapter::EMBL)
+  end
 
   ### private methods
 
@@ -400,3 +497,4 @@ class EMBL < EMBLDB
 end # class EMBL
 
 end # module Bio
+

data/lib/bio/db/embl/embl_to_biosequence.rb

@@ -0,0 +1,85 @@
+#
+# = bio/db/embl/embl_to_biosequence.rb - Bio::EMBL to Bio::Sequence adapter module
+#
+# Copyright::   Copyright (C) 2008
+#               Naohisa Goto <ng@bioruby.org>,
+# License::     The Ruby License
+#
+# $Id:$
+#
+
+require 'bio/sequence'
+require 'bio/sequence/adapter'
+
+# Internal use only. Normal users should not use this module.
+#
+# Bio::EMBL to Bio::Sequence adapter module.
+# It is internally used in Bio::EMBL#to_biosequence.
+#
+module Bio::Sequence::Adapter::EMBL
+
+  extend Bio::Sequence::Adapter
+
+  private
+
+  def_biosequence_adapter :seq
+
+  def_biosequence_adapter :id_namespace do |orig|
+    'EMBL'
+  end
+
+  def_biosequence_adapter :entry_id
+  
+  def_biosequence_adapter :primary_accession do |orig|
+    orig.accessions[0]
+  end
+
+  def_biosequence_adapter :secondary_accessions do |orig|
+    orig.accessions[1..-1] || []
+  end
+
+  def_biosequence_adapter :molecule_type
+
+  def_biosequence_adapter :data_class
+
+  def_biosequence_adapter :definition, :description
+
+  def_biosequence_adapter :topology
+
+  def_biosequence_adapter :date_created
+
+  def_biosequence_adapter :date_modified
+
+  def_biosequence_adapter :release_created
+
+  def_biosequence_adapter :release_modified
+
+  def_biosequence_adapter :entry_version
+
+  def_biosequence_adapter :division
+
+  def_biosequence_adapter :sequence_version, :version
+
+  def_biosequence_adapter :keywords
+
+  def_biosequence_adapter :species
+
+  def_biosequence_adapter :classification
+
+  #--
+  # unsupported yet
+  # def_biosequence_adapter :organelle do |orig|
+  #   orig.fetch('OG')
+  # end
+  #++
+
+  def_biosequence_adapter :references
+
+  def_biosequence_adapter :features
+
+  def_biosequence_adapter :comments, :cc
+
+  def_biosequence_adapter :dblinks
+
+end #module Bio::Sequence::Adapter::EMBL
+

data/lib/bio/db/embl/format_embl.rb

@@ -0,0 +1,190 @@
+#
+# = bio/db/embl/format_embl.rb - EMBL format generater
+#
+# Copyright::  Copyright (C) 2008
+#              Jan Aerts <jandot@bioruby.org>,
+#              Naohisa Goto <ng@bioruby.org>
+# License::    The Ruby License
+#
+# $Id: format_embl.rb,v 1.1.2.7 2008/06/19 12:45:15 ngoto Exp $
+#
+
+require 'bio/sequence/format'
+
+module Bio::Sequence::Format::NucFormatter
+
+  # INTERNAL USE ONLY, YOU SHOULD NOT USE THIS CLASS.
+  # Embl format output class for Bio::Sequence.
+  class Embl < Bio::Sequence::Format::FormatterBase
+
+    # helper methods
+    include Bio::Sequence::Format::INSDFeatureHelper
+    
+    private
+
+    # wrapping with EMBL style
+    def embl_wrap(prefix, str)
+      wrap(str.to_s, 80, prefix)
+    end
+
+    # Given words (an Array of String) are wrapping with EMBL style.
+    # Each word is never splitted inside the word.
+    def embl_wrap_words(prefix, array)
+      width = 80
+      result = []
+      str = nil
+      array.each do |x|
+        if str then
+          if str.length + 1 + x.length > width then
+            str = nil
+          else
+            str.concat ' '
+            str.concat x
+          end
+        end
+        unless str then
+          str = prefix + x
+          result.push str
+        end
+      end
+      result.join("\n")
+    end
+
+    # format reference
+    # ref:: Bio::Reference object
+    # hash:: (optional) a hash for RN (reference number) administration
+    def reference_format_embl(ref, hash = nil)
+      lines = Array.new
+      if ref.embl_gb_record_number or hash then
+        refno = ref.embl_gb_record_number.to_i
+        hash ||= {}
+        if refno <= 0 or hash[refno] then
+          refno = hash.keys.sort[-1].to_i + 1
+          hash[refno] = true
+        end
+        lines << embl_wrap("RN   ", "[#{refno}]")
+      end
+      if ref.comments then
+        ref.comments.each do |cmnt|
+          lines << embl_wrap("RC   ", cmnt)
+        end
+      end
+      unless ref.sequence_position.to_s.empty? then
+        lines << embl_wrap("RP   ",   "#{ref.sequence_position}")
+      end
+      unless ref.doi.to_s.empty? then
+        lines << embl_wrap("RX   ",   "DOI; #{ref.doi}.")
+      end
+      unless ref.pubmed.to_s.empty? then
+        lines << embl_wrap("RX   ",   "PUBMED; #{ref.pubmed}.")
+      end
+      unless ref.authors.empty? then
+        auth = ref.authors.collect do |x|
+          y = x.to_s.strip.split(/\, *([^\,]+)\z/)
+          y[1].gsub!(/\. +/, '.') if y[1]
+          y.join(' ')
+        end
+        lastauth = auth.pop
+        auth.each { |x| x.concat ',' }
+        auth.push(lastauth.to_s + ';')
+        lines << embl_wrap_words('RA   ', auth)
+      end
+      lines << embl_wrap('RT   ',
+                         (ref.title.to_s.empty? ? '' :
+                          "\"#{ref.title}\"") + ';')
+      unless ref.journal.to_s.empty? then
+        volissue = "#{ref.volume.to_s}"
+        volissue = "#{volissue}(#{ref.issue})" unless ref.issue.to_s.empty? 
+        rl = "#{ref.journal}"
+        rl += " #{volissue}" unless volissue.empty? 
+        rl += ":#{ref.pages}" unless ref.pages.to_s.empty?
+        rl += "(#{ref.year})" unless ref.year.to_s.empty?
+        rl += '.'
+        lines << embl_wrap('RL   ', rl)
+      end
+      lines << "XX"
+      return lines.join("\n")
+    end
+
+    def seq_format_embl(seq)
+      counter = 0
+      result = seq.gsub(/.{1,60}/) do |x|
+        counter += x.length
+        x = x.gsub(/.{10}/, '\0 ')
+        sprintf("     %-66s%9d\n", x, counter)
+      end
+      result.chomp!
+      result
+    end
+
+    def seq_composition(seq)
+      { :a => seq.count('aA'),
+        :c => seq.count('cC'),
+        :g => seq.count('gG'),
+        :t => seq.count('tTuU'),
+        :other => seq.count('^aAcCgGtTuU')
+      }
+    end
+
+    # moleculue type
+    def mol_type_embl
+      if mt = molecule_type then
+        mt
+      elsif f = (features or []).find { |f| f.feature == 'source' } and
+          q = f.qualifiers.find { |q| q.qualifier == 'mol_type' } then
+        q.value
+      else
+        'NA'
+      end
+    end
+
+    # CC line. Comments.
+    def comments_format_embl(cmnts)
+      return '' if !cmnts or cmnts.empty?
+      cmnts = [ cmnts ] unless cmnts.kind_of?(Array)
+      a = []
+      cmnts.each do |str|
+        a.push embl_wrap('CC   ', str)
+      end
+      unless a.empty? then
+        a.push "XX   "
+        a.push '' # dummy to put "\n" at the end of the string
+      end
+      a.join("\n")
+    end
+
+
+    # Erb template of EMBL format for Bio::Sequence
+    erb_template <<'__END_OF_TEMPLATE__'
+ID   <%= primary_accession || entry_id %>; SV <%= sequence_version %>; <%= topology %>; <%= mol_type_embl %>; <%= data_class %>; <%= division %>; <%= seq.length %> BP.
+XX   
+<%= embl_wrap('AC   ', accessions.reject{|a| a.nil?}.join('; ') + ';') %>
+XX   
+DT   <%= format_date(date_created || null_date) %> (Rel. <%= release_created || 0 %>, Created)
+DT   <%= format_date(date_modified || null_date) %> (Rel. <%= release_modified || 0 %>, Last updated, Version <%= entry_version || 0 %>)
+XX   
+<%= embl_wrap('DE   ', definition) %>
+XX   
+<%= embl_wrap('KW   ', (keywords || []).join('; ') + '.') %>
+XX   
+OS   <%= species %>
+<%= embl_wrap('OC   ', (classification || []).join('; ') + '.') %>
+XX   
+<% hash = {}; (references || []).each do |ref| %><%= reference_format_embl(ref, hash) %>
+<% end %><% (dblinks || []).each do |r|
+%>DR   <%= r.database %>; <%= r.id %><% unless r.secondary_ids.empty? %>; <%= r.secondary_ids[0] %><% end %>.
+<% end %><% if dblinks and !dblinks.empty? then
+ %>XX   
+<% end %><%= comments_format_embl(comments)
+%>FH   Key             Location/Qualifiers
+FH   
+<%= format_features_embl(features || []) %>XX   
+SQ   Sequence <%= seq.length %> BP; <% c = seq_composition(seq) %><%= c[:a] %> A; <%= c[:c] %> C; <%= c[:g] %> G; <%= c[:t] %> T; <%= c[:other] %> other;
+<%= seq_format_embl(seq) %>
+//
+__END_OF_TEMPLATE__
+
+  end #class Embl
+
+end #module Bio::Sequence::Format::NucFormatter
+