bio 1.4.2 → 1.4.3

data/lib/bio/feature.rb

@@ -5,12 +5,11 @@
 #                             2006        Jan Aerts <jan.aerts@bbsrc.ac.uk>
 # License::	The Ruby License
 #
-# $Id: feature.rb,v 1.13.2.1 2008/03/04 10:12:22 ngoto Exp $
-
-require 'bio/location'
 
 module Bio
 
+  autoload :Locations, 'bio/location' unless const_defined?(:Locations)
+
 # = DESCRIPTION
 # Container for the sequence annotation.
 #

data/lib/bio/io/flatfile/autodetection.rb

@@ -482,7 +482,7 @@ module Bio
             /^seq1 \= .*\, \d+ bp(\r|\r?\n)seq2 \= .*\, \d+ bp(\r|\r?\n)/ ],
 
           fastq  = RuleRegexp[ 'Bio::Fastq',
-            /^\@.+(?:\r|\r?\n)(?:[^\@\+].*(?:\r|\r?\n))+\+.*(?:\r|\r?\n).+(?:\r|\r?\n)/ ],
+            /^\@.+(?:\r|\r?\n)(?:[^\@\+].*(?:\r|\r?\n))+/ ],
 
           fastaformat = RuleProc.new('Bio::FastaFormat',
                                      'Bio::NBRF',

data/lib/bio/io/flatfile/buffer.rb

@@ -43,17 +43,101 @@ module Bio
       # *arg is passed to File.open.
       #
       # Like File.open, a block can be accepted.
+      #
+      # Unlike File.open, the default is binary mode, unless text mode
+      # is explicity specified in mode.
       def self.open_file(filename, *arg)
+        params = _parse_file_open_arg(*arg)
+        if params[:textmode] or /t/ =~ params[:fmode_string].to_s then
+          textmode = true
+        else
+          textmode = false
+        end
         if block_given? then
           File.open(filename, *arg) do |fobj|
+            fobj.binmode unless textmode
             yield self.new(fobj, filename)
           end
         else
           fobj = File.open(filename, *arg)
+          fobj.binmode unless textmode
           self.new(fobj, filename)
         end
       end
 
+      # Parses file open mode parameter.
+      # mode must be an Integer or a String.
+      def self._parse_file_open_mode(mode)
+        modeint = nil
+        modestr = nil
+        begin
+          modeint = mode.to_int
+        rescue NoMethodError
+        end
+        unless modeint then
+          begin
+            modestr = mode.to_str
+          rescue NoMethodError
+          end
+        end
+        if modeint then
+          return { :fmode_integer => modeint }
+        end
+        if modestr then
+          fmode, ext_enc, int_enc = modestr.split(/\:/)
+          ret = { :fmode_string => fmode }
+          ret[:external_encoding] = ext_enc if ext_enc
+          ret[:internal_encoding] = int_enc if int_enc
+          return ret
+        end
+        nil
+      end
+      private_class_method :_parse_file_open_mode
+          
+      # Parses file open arguments
+      def self._parse_file_open_arg(*arg)
+        fmode_hash = nil
+        perm = nil
+
+        elem = arg.shift
+        if elem then
+          fmode_hash = _parse_file_open_mode(elem)
+          if fmode_hash then
+            elem = arg.shift
+            if elem then
+              begin
+                perm = elem.to_int
+              rescue NoMethodError
+              end
+            end
+            elem = arg.shift if perm
+          end
+        end
+        if elem.kind_of?(Hash) then
+          opt = elem.dup
+        else
+          opt = {}
+        end
+        if elem = opt[:mode] then
+          fmode_hash = _parse_file_open_mode(elem)
+        end
+        fmode_hash ||= {}
+        fmode_hash[:perm] = perm if perm
+        unless enc = opt[:encoding].to_s.empty? then
+          ext_enc, int_enc = enc.split(/\:/)
+          fmode_hash[:external_encoding] = ext_enc if ext_enc
+          fmode_hash[:internal_encoding] = int_enc if int_enc
+        end
+
+        [ :external_encoding, :internal_encoding,
+          :textmode, :binmode, :autoclose, :perm ].each do |key|
+          val = opt[key]
+          fmode_hash[key] = val if val
+        end
+        fmode_hash
+      end
+      private_class_method :_parse_file_open_arg
+
       # Creates a new input stream wrapper from URI specified as _uri_.
       # by using OpenURI.open_uri or URI#open.
       # _uri_ must be a String or URI object.

data/lib/bio/sequence.rb

@@ -9,10 +9,6 @@
 #               Jan Aerts <jan.aerts@bbsrc.ac.uk>
 # License::     The Ruby License
 #
-# $Id:$
-#
-
-require 'bio/sequence/compat'
 
 module Bio
 
@@ -74,6 +70,12 @@ class Sequence
   autoload :QualityScore, 'bio/sequence/quality_score'
   autoload :SequenceMasker, 'bio/sequence/sequence_masker'
 
+  #--
+  # require "bio/sequence/compat.rb" here to avoid circular require and
+  # possible superclass mismatch of AA class
+  #++
+  require 'bio/sequence/compat'
+
   include Format
   include SequenceMasker
 

data/lib/bio/sequence/aa.rb

@@ -6,14 +6,12 @@
 #               Ryan Raaum <ryan@raaum.org>
 # License::     The Ruby License
 #
-# $Id: aa.rb,v 1.4 2007/04/05 23:35:41 trevor Exp $
-#
-
-require 'bio/sequence/common'
 
 module Bio
 
-  autoload :AminoAcid, 'bio/data/aa'
+  autoload :AminoAcid, 'bio/data/aa' unless const_defined?(:AminoAcid)
+
+  require 'bio/sequence' unless const_defined?(:Sequence)
 
 class Sequence
 

data/lib/bio/sequence/adapter.rb

@@ -5,10 +5,10 @@
 #               Naohisa Goto <ng@bioruby.org>,
 # License::     The Ruby License
 #
-# $Id:$
-#
 
-require 'bio/sequence'
+module Bio
+
+require 'bio/sequence' unless const_defined?(:Sequence)
 
 # Internal use only. Normal users should not use this module.
 #
@@ -17,7 +17,7 @@ require 'bio/sequence'
 #
 # This module is used by using "extend", not "include".
 # 
-module Bio::Sequence::Adapter
+module Sequence::Adapter
 
   autoload :GenBank,     'bio/db/genbank/genbank_to_biosequence'
   autoload :EMBL,        'bio/db/embl/embl_to_biosequence'
@@ -107,6 +107,6 @@ __END_OF_DEF__
     true
   end
 
-end #module Bio::Sequence::Adapter
-
+end #module Sequence::Adapter
 
+end #module Bio

data/lib/bio/sequence/common.rb

@@ -6,12 +6,12 @@
 #               Ryan Raaum <ryan@raaum.org>
 # License::     The Ruby License
 #
-# $Id:$
-#
 
 module Bio
 
-  autoload :Locations, 'bio/location'
+  autoload :Locations, 'bio/location' unless const_defined?(:Locations)
+
+  require 'bio/sequence' unless const_defined?(:Sequence)
 
 class Sequence
 

data/lib/bio/sequence/compat.rb

@@ -6,17 +6,12 @@
 #               Ryan Raaum <ryan@raaum.org>
 # License::     The Ruby License
 #
-# $Id:$
-#
-
 
 module Bio
 
-class Sequence
+require 'bio/sequence' unless const_defined?(:Sequence)
 
-  autoload :Common, 'bio/sequence/common'
-  autoload :NA,     'bio/sequence/na'
-  autoload :AA,     'bio/sequence/aa'
+class Sequence
 
   # Return sequence as 
   # String[http://corelib.rubyonrails.org/classes/String.html].

data/lib/bio/sequence/dblink.rb

@@ -5,15 +5,15 @@
 #              Naohisa Goto <ng@bioruby.org>
 # License::    The Ruby License
 #
-# $Id: dblink.rb,v 1.1.2.1 2008/06/17 15:44:22 ngoto Exp $
-#
 
-require 'bio/sequence'
+module Bio
+
+require 'bio/sequence' unless const_defined?(:Sequence)
 
 # Bio::Sequence::DBLink stores IDs with the database name.
 # Its main purpose is to store database cross-reference information
 # for a sequence entry.
-class Bio::Sequence::DBLink
+class Sequence::DBLink
 
   # creates a new DBLink object
   def initialize(database, primary_id, *secondary_ids)
@@ -49,6 +49,7 @@ class Bio::Sequence::DBLink
     self.new(*(str.split(/\s*\;\s*/)))
   end
 
-end #class Bio::Sequence::DBLink
+end #class Sequence::DBLink
 
+end #module Bio
 

data/lib/bio/sequence/format.rb

@@ -8,12 +8,6 @@
 #               Jan Aerts <jan.aerts@bbsrc.ac.uk>
 # License::     The Ruby License
 #
-# = TODO
-#
-# porting from N. Goto's feature-output.rb on BioRuby list.
-#
-# $Id: format.rb,v 1.4.2.8 2008/06/17 15:50:05 ngoto Exp $
-#
 
 require 'erb'
 

data/lib/bio/sequence/format_raw.rb

@@ -4,10 +4,6 @@
 # Copyright::  Copyright (C) 2008 Naohisa Goto <ng@bioruby.org>
 # License::    The Ruby License
 #
-# $Id: format_raw.rb,v 1.1.2.1 2008/03/04 11:28:46 ngoto Exp $
-#
-
-require 'bio/sequence/format'
 
 module Bio::Sequence::Format::Formatter
 

data/lib/bio/sequence/generic.rb

@@ -5,12 +5,11 @@
 #               Toshiaki Katayama <k@bioruby.org>
 # License::     The Ruby License
 #
-# $Id: generic.rb,v 1.5 2007/04/05 23:35:41 trevor Exp $
-#
-
-require 'bio/sequence/common'
 
 module Bio
+
+require 'bio/sequence' unless const_defined?(:Sequence)
+
 class Sequence
 
 class Generic < String #:nodoc:

data/lib/bio/sequence/na.rb

@@ -6,15 +6,13 @@
 #               Ryan Raaum <ryan@raaum.org>
 # License::     The Ruby License
 #
-# $Id: na.rb,v 1.7 2007/04/23 16:43:51 trevor Exp $
-#
-
-require 'bio/sequence/common'
 
 module Bio
 
-  autoload :NucleicAcid, 'bio/data/na'
-  autoload :CodonTable,  'bio/data/codontable'
+  autoload :NucleicAcid, 'bio/data/na' unless const_defined?(:NucleicAcid)
+  autoload :CodonTable,  'bio/data/codontable' unless const_defined?(:CodonTable)
+
+  require 'bio/sequence' unless const_defined?(:Sequence)
 
 class Sequence
 

data/lib/bio/sequence/quality_score.rb

@@ -18,6 +18,8 @@
 
 module Bio
 
+require 'bio/sequence' unless const_defined?(:Sequence)
+
 class Sequence
 
   # Bio::Sequence::QualityScore is a name space for quality score modules. 

data/lib/bio/sequence/sequence_masker.rb

@@ -14,6 +14,9 @@
 #
 
 module Bio
+
+require 'bio/sequence' unless const_defined?(:Sequence)
+
 class Sequence
 
   # Bio::Sequence::SequenceMasker is a mix-in module to provide helpful

data/lib/bio/shell/core.rb

@@ -343,6 +343,7 @@ module Bio::Shell::Ghost
       File.open(file, "w") do |f|
         bind = Bio::Shell.cache[:binding]
         list = eval("local_variables", bind)
+        list.collect! { |x| x.to_s }
         list -= ["_"]
         hash = {}
         list.each do |elem|

data/lib/bio/tree.rb

@@ -5,7 +5,6 @@
 #               Naohisa Goto <ng@bioruby.org>
 # License::     The Ruby License
 #
-# $Id:$
 #
 
 require 'matrix'
@@ -913,5 +912,5 @@ end #module Bio
 #---
 # temporary added
 #+++
-require 'bio/db/newick'
+require 'bio/tree/output'
 

data/lib/bio/tree/output.rb

@@ -0,0 +1,264 @@
+#
+# = bio/tree/output.rb - Phylogenetic tree formatter
+#
+# Copyright::   Copyright (C) 2004-2006
+#               Naohisa Goto <ng@bioruby.org>
+# License::     The Ruby License
+#
+#
+# == Description
+#
+# This file contains formatter of Newick, NHX and Phylip distance matrix.
+#
+# == References
+#
+# * http://evolution.genetics.washington.edu/phylip/newick_doc.html
+# * http://www.phylosoft.org/forester/NHX.html
+#
+
+module Bio
+  class Tree
+
+    #---
+    # newick output
+    #+++
+
+    # default options
+    DEFAULT_OPTIONS =
+      { :indent => '  ' }
+
+    def __get_option(key, options)
+      if (r = options[key]) != nil then
+        r
+      elsif @options && (r = @options[key]) != nil then
+        r
+      else
+        DEFAULT_OPTIONS[key]
+      end
+    end
+    private :__get_option
+
+
+    # formats Newick label (unquoted_label or quoted_label)
+    def __to_newick_format_label(str, options)
+      if __get_option(:parser, options) == :naive then
+        return str.to_s
+      end
+      str = str.to_s
+      if /([\(\)\,\:\[\]\_\'\x00-\x1f\x7f])/ =~ str then
+        # quoted_label
+        return "\'" + str.gsub(/\'/, "\'\'") + "\'"
+      end
+      # unquoted_label
+      return str.gsub(/ /, '_')
+    end
+    private :__to_newick_format_label
+
+    # formats leaf
+    def __to_newick_format_leaf(node, edge, options)
+
+      label = __to_newick_format_label(get_node_name(node), options)
+
+      dist = get_edge_distance_string(edge)
+
+      bs = get_node_bootstrap_string(node)
+
+      if  __get_option(:branch_length_style, options) == :disabled
+        dist = nil
+      end
+     
+      case __get_option(:bootstrap_style, options)
+      when :disabled
+        label + (dist ? ":#{dist}" : '')
+      when :molphy
+        label + (dist ? ":#{dist}" : '') + (bs ? "[#{bs}]" : '')
+      when :traditional
+        label + (bs ? bs : '') + (dist ? ":#{dist}" : '')
+      else
+        # default: same as molphy style
+        label + (dist ? ":#{dist}" : '') + (bs ? "[#{bs}]" : '')
+      end
+    end
+    private :__to_newick_format_leaf
+
+    # formats leaf for NHX
+    def __to_newick_format_leaf_NHX(node, edge, options)
+
+      label = __to_newick_format_label(get_node_name(node), options)
+
+      dist = get_edge_distance_string(edge)
+
+      bs = get_node_bootstrap_string(node)
+
+      if  __get_option(:branch_length_style, options) == :disabled
+        dist = nil
+      end
+
+      nhx = {}
+
+      # bootstrap
+      nhx[:B] = bs if bs and !(bs.empty?)
+      # EC number
+      nhx[:E] = node.ec_number if node.instance_eval {
+        defined?(@ec_number) && self.ec_number
+      }
+      # scientific name
+      nhx[:S] = node.scientific_name if node.instance_eval {
+        defined?(@scientific_name) && self.scientific_name
+      }
+      # taxonomy id
+      nhx[:T] = node.taxonomy_id if node.instance_eval {
+        defined?(@taxonomy_id) && self.taxonomy_id
+      }
+
+      # :D (gene duplication or speciation)
+      if node.instance_eval { defined?(@events) && !(self.events.empty?) } then
+        if node.events.include?(:gene_duplication)
+          nhx[:D] = 'Y'
+        elsif node.events.include?(:speciation)
+          nhx[:D] = 'N'
+        end
+      end
+
+      # log likelihood
+      nhx[:L] = edge.log_likelihood if edge.instance_eval {
+        defined?(@log_likelihood) && self.log_likelihood }
+      # width
+      nhx[:W] = edge.width if edge.instance_eval {
+        defined?(@width) && self.width }
+
+      # merges other parameters
+      flag = node.instance_eval { defined? @nhx_parameters }
+      nhx.merge!(node.nhx_parameters) if flag
+      flag = edge.instance_eval { defined? @nhx_parameters }
+      nhx.merge!(edge.nhx_parameters) if flag
+
+      nhx_string = nhx.keys.sort{ |a,b| a.to_s <=> b.to_s }.collect do |key|
+        "#{key.to_s}=#{nhx[key].to_s}"
+      end.join(':')
+      nhx_string = "[&&NHX:" + nhx_string + "]" unless nhx_string.empty?
+     
+      label + (dist ? ":#{dist}" : '') + nhx_string
+    end
+    private :__to_newick_format_leaf_NHX
+
+    #
+    def __to_newick(parents, source, depth, format_leaf,
+                    options, &block)
+      result = []
+      if indent_string = __get_option(:indent, options) then
+        indent0 = indent_string * depth
+        indent  = indent_string * (depth + 1)
+        newline = "\n"
+      else
+        indent0 = indent = newline = ''
+      end
+      out_edges = self.out_edges(source)
+      if block_given? then
+        out_edges.sort! { |edge1, edge2| yield(edge1[1], edge2[1]) }
+      else
+        out_edges.sort! do |edge1, edge2|
+          o1 = edge1[1].order_number
+          o2 = edge2[1].order_number
+          if o1 and o2 then
+            o1 <=> o2
+          else
+            edge1[1].name.to_s <=> edge2[1].name.to_s
+          end
+        end
+      end
+      out_edges.each do |src, tgt, edge|
+        if parents.include?(tgt) then
+          ;;
+        elsif self.out_degree(tgt) == 1 then
+          result << indent + __send__(format_leaf, tgt, edge, options)
+        else
+          result << 
+            __to_newick([ src ].concat(parents), tgt, depth + 1,
+                        format_leaf, options) +
+            __send__(format_leaf, tgt, edge, options)
+        end
+      end
+      indent0 + "(" + newline + result.join(',' + newline) +
+        (result.size > 0 ? newline : '') + indent0 + ')'
+    end
+    private :__to_newick
+
+    # Returns a newick formatted string.
+    # If block is given, the order of the node is sorted
+    # (as the same manner as Enumerable#sort).
+    #
+    # Available options:
+    # <tt>:indent</tt>::
+    #     indent string; set false to disable (default: '  ')
+    # <tt>:bootstrap_style</tt>::
+    #     <tt>:disabled</tt> disables bootstrap representations.
+    #     <tt>:traditional</tt> for traditional style.
+    #     <tt>:molphy</tt> for Molphy style (default).
+    def output_newick(options = {}, &block) #:yields: node1, node2
+      root = @root
+      root ||= self.nodes.first
+      return '();' unless root
+      __to_newick([], root, 0, :__to_newick_format_leaf, options, &block) +
+        __to_newick_format_leaf(root, Edge.new, options) +
+        ";\n"
+    end
+
+    alias newick output_newick
+      
+
+    # Returns a NHX (New Hampshire eXtended) formatted string.
+    # If block is given, the order of the node is sorted
+    # (as the same manner as Enumerable#sort).
+    #
+    # Available options:
+    # <tt>:indent</tt>::
+    #     indent string; set false to disable (default: '  ')
+    #
+    def output_nhx(options = {}, &block) #:yields: node1, node2
+      root = @root
+      root ||= self.nodes.first
+      return '();' unless root
+      __to_newick([], root, 0,
+                  :__to_newick_format_leaf_NHX, options, &block) +
+        __to_newick_format_leaf_NHX(root, Edge.new, options) +
+        ";\n"
+    end
+
+    # Returns formatted text (or something) of the tree
+    # Currently supported format is: :newick, :nhx
+    def output(format, *arg, &block)
+      case format
+      when :newick
+        output_newick(*arg, &block)
+      when :nhx
+        output_nhx(*arg, &block)
+      when :phylip_distance_matrix
+        output_phylip_distance_matrix(*arg, &block)
+      else
+        raise 'Unknown format'
+      end
+    end
+
+    #---
+    # This method isn't suitable to written in this file?
+    #+++
+
+    # Generates phylip-style distance matrix as a string.
+    # if nodes is not given, all leaves in the tree are used.
+    # If the names of some of the given (or default) nodes
+    # are not defined or are empty, the names are automatically generated.
+    def output_phylip_distance_matrix(nodes = nil, options = {})
+      nodes = self.leaves unless nodes
+      names = nodes.collect do |x|
+        y = get_node_name(x)
+        y = sprintf("%x", x.__id__.abs) if y.empty?
+        y
+      end
+      m = self.distance_matrix(nodes)
+      Bio::Phylip::DistanceMatrix.generate(m, names, options)
+    end
+
+  end #class Tree
+
+end #module Bio