wwood-bioruby 1.2.11

data/lib/bio/compat/features.rb

@@ -0,0 +1,157 @@
+#
+# = bio/compat/features.rb - Obsoleted Features class
+#
+# Copyright::	Copyright (c) 2002, 2005  Toshiaki Katayama <k@bioruby.org>
+#                             2006        Jan Aerts <jan.aerts@bbsrc.ac.uk>
+#                             2008        Naohisa Goto <ng@bioruby.org>
+# License::	The Ruby License
+#
+# $Id: features.rb,v 1.1.2.2 2008/03/10 13:42:26 ngoto Exp $
+#
+# == Description
+#
+# The Bio::Features class was obsoleted after BioRuby 1.2.1.
+# To keep compatibility, some wrapper methods are provided in this file.
+# As the compatibility methods (and Bio::Features) will soon be removed,
+# Please change your code not to use Bio::Features.
+#
+# Note that Bio::Feature is different from the Bio::Features.
+# Bio::Feature still exists to store DDBJ/GenBank/EMBL feature information.
+
+require 'bio/location'
+
+module Bio
+
+# = DESCRIPTION
+#
+# This class is OBSOLETED, and will soon be removed.
+# Instead of this class, an array is to be used.
+#
+#
+# Container for a list of Feature objects.
+#
+# = USAGE
+#  # First, create some Bio::Feature objects
+#  feature1 = Bio::Feature.new('intron','3627..4059')
+#  feature2 = Bio::Feature.new('exon','4060..4236')
+#  feature3 = Bio::Feature.new('intron','4237..4426')
+#  feature4 = Bio::Feature.new('CDS','join(2538..3626,4060..4236)',
+#                   [ Bio::Feature::Qualifier.new('gene', 'CYP2D6'),
+#                     Bio::Feature::Qualifier.new('translation','MGXXTVMHLL...')
+#                   ])
+#
+#  # And create a container for them
+#  feature_container = Bio::Features.new([ feature1, feature2, feature3, feature4 ])
+#
+#  # Iterate over all features and print
+#  feature_container.each do |feature|
+#    puts feature.feature + "\t" + feature.position
+#    feature.each do |qualifier|
+#      puts "- " + qualifier.qualifier + ": " + qualifier.value
+#    end
+#  end
+#
+#  # Iterate only over CDS features and extract translated amino acid sequences
+#  features.each("CDS") do |feature|
+#    hash = feature.to_hash
+#    name = hash["gene"] || hash["product"] || hash["note"] 
+#    aaseq  = hash["translation"]
+#    pos  = feature.position
+#    if name and seq
+#      puts ">#{gene} #{feature.position}"
+#      puts aaseq
+#    end
+#  end
+class Features
+
+  # module to keep backward compatibility with obsoleted Bio::Features
+  module BackwardCompatibility #:nodoc:
+
+    # Backward compatibility with Bio::Features#features.
+    # Now, features are stored in an array, and
+    # you should change your code not to use this method.
+    def features
+      warn 'Bio::Features is obsoleted. Now, features are stored in an array.'
+      self
+    end
+
+    # Backward compatibility with Bio::Features#append.
+    # Now, references are stored in an array, and
+    # you should change your code not to use this method.
+    def append(feature)
+      warn 'Bio::Features is obsoleted. Now, features are stored in an array.'
+      self.push(feature) if feature.is_a? Feature
+      self
+    end
+  end #module BackwardCompatibility
+
+  # This method should not be used.
+  # Only for backward compatibility of existing code.
+  #
+  # Since Bio::Features is obsoleted,
+  # Bio::Features.new not returns Bio::Features object,
+  # but modifies given _ary_ and returns the _ary_.
+  #
+  # *Arguments*:
+  # * (optional) __: Array of Bio::Feature objects
+  # *Returns*:: the given array
+  def self.new(ary = [])
+    warn 'Bio::Features is obsoleted. Some methods are added to given array to keep backward compatibility.'
+    ary.extend(BackwardCompatibility)
+    ary
+  end
+
+  # Normally, users can not call this method.
+  #
+  # Create a new Bio::Features object.
+  #
+  # *Arguments*:
+  # * (optional) _list of features_: list of Bio::Feature objects
+  # *Returns*:: Bio::Features object
+  def initialize(ary = [])
+    @features = ary
+  end
+
+  # Returns an Array of Feature objects.
+  attr_accessor :features
+
+  # Appends a Feature object to Features.
+  # 
+  # *Arguments*:
+  # * (required) _feature_: Bio::Feature object
+  # *Returns*:: Bio::Features object
+  def append(a)
+    @features.push(a) if a.is_a? Feature
+    return self
+  end
+
+  # Iterates on each feature object.
+  #
+  # *Arguments*:
+  # * (optional) _key_: if specified, only iterates over features with this key
+  def each(arg = nil)
+    @features.each do |x|
+      next if arg and x.feature != arg
+      yield x
+    end
+  end
+
+  # Short cut for the Features#features[n]
+  def [](*arg)
+    @features[*arg]
+  end
+
+  # Short cut for the Features#features.first
+  def first
+    @features.first
+  end
+
+  # Short cut for the Features#features.last
+  def last
+    @features.last
+  end
+
+end # Features
+
+end # Bio
+

data/lib/bio/compat/references.rb

@@ -0,0 +1,128 @@
+#
+# = bio/compat/references.rb - Obsoleted References class
+#
+# Copyright::   Copyright (C) 2008
+#               Toshiaki Katayama <k@bioruby.org>,
+#               Ryan Raaum <ryan@raaum.org>,
+#               Jan Aerts <jandot@bioruby.org>,
+#               Naohisa Goto <ng@bioruby.org>
+# License::     The Ruby License
+#
+# $Id: references.rb,v 1.1.2.1 2008/03/04 10:07:49 ngoto Exp $
+#
+# == Description
+#
+# The Bio::References class was obsoleted after BioRuby 1.2.1.
+# To keep compatibility, some wrapper methods are provided in this file.
+# As the compatibility methods (and Bio::References) will soon be removed,
+# Please change your code not to use Bio::References.
+#
+# Note that Bio::Reference is different from Bio::References.
+# Bio::Reference still exists for storing a reference information
+# in sequence entries.
+
+module Bio
+
+  # = DESCRIPTION
+  #
+  # This class is OBSOLETED, and will soon be removed.
+  # Instead of this class, an array is to be used.
+  #
+  # 
+  # A container class for Bio::Reference objects.
+  #
+  # = USAGE
+  #
+  # This class should NOT be used.
+  #
+  #   refs = Bio::References.new
+  #   refs.append(Bio::Reference.new(hash))
+  #   refs.each do |reference|
+  #     ...
+  #   end
+  #
+  class References
+
+    # module to keep backward compatibility with obsoleted Bio::References
+    module BackwardCompatibility #:nodoc:
+
+      # Backward compatibility with Bio::References#references.
+      # Now, references are stored in an array, and
+      # you should change your code not to use this method.
+      def references
+        warn 'Bio::References is obsoleted. Now, references are stored in an array.'
+        self
+      end
+
+      # Backward compatibility with Bio::References#append.
+      # Now, references are stored in an array, and
+      # you should change your code not to use this method.
+      def append(reference)
+        warn 'Bio::References is obsoleted. Now, references are stored in an array.'
+        self.push(reference) if reference.is_a? Reference
+        self
+      end
+    end #module BackwardCompatibility
+
+    # This method should not be used.
+    # Only for backward compatibility of existing code.
+    #
+    # Since Bio::References is obsoleted,
+    # Bio::References.new not returns Bio::References object,
+    # but modifies given _ary_ and returns the _ary_.
+    #
+    # *Arguments*:
+    # * (optional) __: Array of Bio::Reference objects
+    # *Returns*:: the given array
+    def self.new(ary = [])
+      warn 'Bio::References is obsoleted. Some methods are added to given array to keep backward compatibility.'
+      ary.extend(BackwardCompatibility)
+      ary
+    end
+
+    # Array of Bio::Reference objects
+    attr_accessor :references
+
+    # Normally, users can not call this method.
+    #
+    # Create a new Bio::References object
+    # 
+    #   refs = Bio::References.new
+    # ---
+    # *Arguments*:
+    # * (optional) __: Array of Bio::Reference objects
+    # *Returns*:: Bio::References object
+    def initialize(ary = [])
+      @references = ary
+    end
+
+
+    # Add a Bio::Reference object to the container.
+    #
+    #   refs.append(reference)
+    # ---
+    # *Arguments*:
+    # * (required) _reference_: Bio::Reference object
+    # *Returns*:: current Bio::References object
+    def append(reference)
+      @references.push(reference) if reference.is_a? Reference
+      return self
+    end
+
+    # Iterate through Bio::Reference objects.
+    #
+    #   refs.each do |reference|
+    #     ...
+    #   end
+    # ---
+    # *Block*:: yields each Bio::Reference object
+    def each
+      @references.each do |reference|
+        yield reference
+      end
+    end
+
+  end #class References
+end #module Bio
+
+

data/lib/bio/data/aa.rb

@@ -0,0 +1,353 @@
+#
+# = bio/data/aa.rb - Amino Acids
+#
+# Copyright::	Copyright (C) 2001, 2005
+#		Toshiaki Katayama <k@bioruby.org>
+# License::	The Ruby License
+#
+# $Id: aa.rb,v 0.22 2007/04/06 04:44:51 k Exp $
+#
+
+module Bio
+
+class AminoAcid
+
+  module Data
+
+    # IUPAC code
+    # * http://www.iupac.org/
+    # * http://www.chem.qmw.ac.uk/iubmb/newsletter/1999/item3.html
+    # * http://www.ebi.ac.uk/RESID/faq.html
+
+    NAMES = {
+
+      'A' => 'Ala',
+      'C' => 'Cys',
+      'D' => 'Asp',
+      'E' => 'Glu',
+      'F' => 'Phe',
+      'G' => 'Gly',
+      'H' => 'His',
+      'I' => 'Ile',
+      'K' => 'Lys',
+      'L' => 'Leu',
+      'M' => 'Met',
+      'N' => 'Asn',
+      'P' => 'Pro',
+      'Q' => 'Gln',
+      'R' => 'Arg',
+      'S' => 'Ser',
+      'T' => 'Thr',
+      'V' => 'Val',
+      'W' => 'Trp',
+      'Y' => 'Tyr',
+      'B' => 'Asx',	# D/N
+      'Z' => 'Glx',	# E/Q
+      'J' => 'Xle',	# I/L
+      'U' => 'Sec',	# 'uga' (stop)
+      'O' => 'Pyl',	# 'uag' (stop)
+      'X' => 'Xaa',	# (unknown)
+     
+      'Ala' => 'alanine',
+      'Cys' => 'cysteine',
+      'Asp' => 'aspartic acid',
+      'Glu' => 'glutamic acid',
+      'Phe' => 'phenylalanine',
+      'Gly' => 'glycine',
+      'His' => 'histidine',
+      'Ile' => 'isoleucine',
+      'Lys' => 'lysine',
+      'Leu' => 'leucine',
+      'Met' => 'methionine',
+      'Asn' => 'asparagine',
+      'Pro' => 'proline',
+      'Gln' => 'glutamine',
+      'Arg' => 'arginine',
+      'Ser' => 'serine',
+      'Thr' => 'threonine',
+      'Val' => 'valine',
+      'Trp' => 'tryptophan',
+      'Tyr' => 'tyrosine',
+      'Asx' => 'asparagine/aspartic acid [DN]',
+      'Glx' => 'glutamine/glutamic acid [EQ]',
+      'Xle' => 'isoleucine/leucine [IL]',
+      'Sec' => 'selenocysteine',
+      'Pyl' => 'pyrrolysine',
+      'Xaa' => 'unknown [A-Z]',
+
+    }
+
+    # AAindex FASG760101 - Molecular weight (Fasman, 1976)
+    #   Fasman, G.D., ed.
+    #   Handbook of Biochemistry and Molecular Biology", 3rd ed.,
+    #   Proteins - Volume 1, CRC Press, Cleveland (1976)
+
+    WEIGHT = {
+
+      'A' => 89.09,
+      'C' => 121.15,	# 121.16 according to the Wikipedia
+      'D' => 133.10,
+      'E' => 147.13,
+      'F' => 165.19,
+      'G' => 75.07,
+      'H' => 155.16,
+      'I' => 131.17,
+      'K' => 146.19,
+      'L' => 131.17,
+      'M' => 149.21,
+      'N' => 132.12,
+      'P' => 115.13,
+      'Q' => 146.15,
+      'R' => 174.20,
+      'S' => 105.09,
+      'T' => 119.12,
+      'U' => 168.06,
+      'V' => 117.15,
+      'W' => 204.23,
+      'Y' => 181.19,
+    }
+    
+    
+    ACIDIC_AMINO_ACIDS = %w(D E)
+    BASIC_AMINO_ACIDS = %w(H K R)
+
+    def weight(x = nil)
+      if x
+        if x.length > 1
+          total = 0.0
+          x.each_byte do |byte|
+            aa = byte.chr.upcase
+            if WEIGHT[aa]
+              total += WEIGHT[aa]
+            else
+              raise "Error: invalid amino acid '#{aa}'"
+            end
+          end
+          total -= NucleicAcid.weight[:water] * (x.length - 1)
+        else
+          WEIGHT[x]
+        end
+      else
+        WEIGHT
+      end
+    end
+
+    def [](x)
+      NAMES[x]
+    end
+
+    # backward compatibility
+    def names
+      NAMES
+    end
+    alias aa names
+
+    def name(x)
+      str = NAMES[x]
+      if str and str.length == 3
+        NAMES[str]
+      else
+        str
+      end
+    end
+
+    def to_1(x)
+      case x.to_s.length
+      when 1
+        x
+      when 3
+        three2one(x)
+      else
+        name2one(x)
+      end
+    end
+    alias one to_1
+
+    def to_3(x)
+      case x.to_s.length
+      when 1
+        one2three(x)
+      when 3
+        x
+      else
+        name2three(x)
+      end
+    end
+    alias three to_3
+
+    def one2three(x)
+      if x and x.length != 1
+        raise ArgumentError
+      else
+        NAMES[x]
+      end
+    end
+
+    def three2one(x)
+      if x and x.length != 3
+        raise ArgumentError
+      else
+        reverse[x]
+      end
+    end
+
+    def one2name(x)
+      if x and x.length != 1
+        raise ArgumentError
+      else
+        three2name(NAMES[x])
+      end
+    end
+
+    def name2one(x)
+      str = reverse[x.to_s.downcase]
+      if str and str.length == 3
+        three2one(str)
+      else
+        str
+      end
+    end
+
+    def three2name(x)
+      if x and x.length != 3
+        raise ArgumentError
+      else
+        NAMES[x]
+      end
+    end
+
+    def name2three(x)
+      reverse[x.downcase]
+    end
+
+    def to_re(seq)
+      replace = {
+        'B' => '[DNB]',
+        'Z' => '[EQZ]',
+        'J' => '[ILJ]',
+        'X' => '[ACDEFGHIKLMNPQRSTVWYUOX]',
+      }
+      replace.default = '.'
+
+      str = seq.to_s.upcase
+      str.gsub!(/[^ACDEFGHIKLMNPQRSTVWYUO]/) { |aa|
+        replace[aa]
+      }
+      Regexp.new(str)
+    end
+
+
+    private
+
+
+    def reverse
+      hash = Hash.new
+      NAMES.each do |k, v|
+        hash[v] = k
+      end
+      hash
+    end
+
+  end
+
+
+  # as instance methods
+  include Data
+
+  # as class methods
+  extend Data
+
+
+  private
+
+
+  # override when used as an instance method to improve performance
+  alias orig_reverse reverse
+  def reverse
+    unless @reverse
+      @reverse = orig_reverse
+    end
+    @reverse
+  end
+
+end
+
+end # module Bio
+
+
+if __FILE__ == $0
+
+  puts "### aa = Bio::AminoAcid.new"
+  aa = Bio::AminoAcid.new
+
+  puts "# Bio::AminoAcid['A']"
+  p Bio::AminoAcid['A']
+  puts "# aa['A']"
+  p aa['A']
+
+  puts "# Bio::AminoAcid.name('A'), Bio::AminoAcid.name('Ala')"
+  p Bio::AminoAcid.name('A'), Bio::AminoAcid.name('Ala')
+  puts "# aa.name('A'), aa.name('Ala')"
+  p aa.name('A'), aa.name('Ala')
+
+  puts "# Bio::AminoAcid.to_1('alanine'), Bio::AminoAcid.one('alanine')"
+  p Bio::AminoAcid.to_1('alanine'), Bio::AminoAcid.one('alanine')
+  puts "# aa.to_1('alanine'), aa.one('alanine')"
+  p aa.to_1('alanine'), aa.one('alanine')
+  puts "# Bio::AminoAcid.to_1('Ala'), Bio::AminoAcid.one('Ala')"
+  p Bio::AminoAcid.to_1('Ala'), Bio::AminoAcid.one('Ala')
+  puts "# aa.to_1('Ala'), aa.one('Ala')"
+  p aa.to_1('Ala'), aa.one('Ala')
+  puts "# Bio::AminoAcid.to_1('A'), Bio::AminoAcid.one('A')"
+  p Bio::AminoAcid.to_1('A'), Bio::AminoAcid.one('A')
+  puts "# aa.to_1('A'), aa.one('A')"
+  p aa.to_1('A'), aa.one('A')
+
+  puts "# Bio::AminoAcid.to_3('alanine'), Bio::AminoAcid.three('alanine')"
+  p Bio::AminoAcid.to_3('alanine'), Bio::AminoAcid.three('alanine')
+  puts "# aa.to_3('alanine'), aa.three('alanine')"
+  p aa.to_3('alanine'), aa.three('alanine')
+  puts "# Bio::AminoAcid.to_3('Ala'), Bio::AminoAcid.three('Ala')"
+  p Bio::AminoAcid.to_3('Ala'), Bio::AminoAcid.three('Ala')
+  puts "# aa.to_3('Ala'), aa.three('Ala')"
+  p aa.to_3('Ala'), aa.three('Ala')
+  puts "# Bio::AminoAcid.to_3('A'), Bio::AminoAcid.three('A')"
+  p Bio::AminoAcid.to_3('A'), Bio::AminoAcid.three('A')
+  puts "# aa.to_3('A'), aa.three('A')"
+  p aa.to_3('A'), aa.three('A')
+
+  puts "# Bio::AminoAcid.one2three('A')"
+  p Bio::AminoAcid.one2three('A')
+  puts "# aa.one2three('A')"
+  p aa.one2three('A')
+
+  puts "# Bio::AminoAcid.three2one('Ala')"
+  p Bio::AminoAcid.three2one('Ala')
+  puts "# aa.three2one('Ala')"
+  p aa.three2one('Ala')
+
+  puts "# Bio::AminoAcid.one2name('A')"
+  p Bio::AminoAcid.one2name('A')
+  puts "# aa.one2name('A')"
+  p aa.one2name('A')
+
+  puts "# Bio::AminoAcid.name2one('alanine')"
+  p Bio::AminoAcid.name2one('alanine')
+  puts "# aa.name2one('alanine')"
+  p aa.name2one('alanine')
+
+  puts "# Bio::AminoAcid.three2name('Ala')"
+  p Bio::AminoAcid.three2name('Ala')
+  puts "# aa.three2name('Ala')"
+  p aa.three2name('Ala')
+
+  puts "# Bio::AminoAcid.name2three('alanine')"
+  p Bio::AminoAcid.name2three('alanine')
+  puts "# aa.name2three('alanine')"
+  p aa.name2three('alanine')
+
+  puts "# Bio::AminoAcid.to_re('BZACDEFGHIKLMNPQRSTVWYU')"
+  p Bio::AminoAcid.to_re('BZACDEFGHIKLMNPQRSTVWYU')
+
+end
+