wwood-bioruby 1.2.11

data/lib/bio/appl/paml/common_report.rb

@@ -0,0 +1,38 @@
+#
+# = bio/appl/paml/common_report.rb - basic report class for PAML results
+#
+# Copyright::  Copyright (C) 2008
+#              Naohisa Goto <ng@bioruby.org>
+#
+# License::    The Ruby License
+#
+# == Description
+#
+# This file contains Bio::PAML::Common::Report, a basic report class
+# for PAML program's results.
+#
+# == References
+#
+# * http://abacus.gene.ucl.ac.uk/software/paml.html
+#
+
+require 'bio/appl/paml/common'
+
+module Bio::PAML
+  class Common
+
+    # UNDER CONSTRUCTION.
+    #
+    # Bio::PAML::Common::Report is a basic report class for PAML program's
+    # results. It will have common function for baseml and codeml.
+    #
+    # Normally, users should not use this class directly.
+    class Report
+
+      # Creates a new Report object.
+      def initialize(str)
+      end
+    end #class Report
+
+  end #class Common
+end #module Bio::PAML

data/lib/bio/appl/paml/yn00.rb

@@ -0,0 +1,103 @@
+#
+# = bio/appl/paml/baseml.rb - Wrapper for running PAML program yn00
+#
+# Copyright::  Copyright (C) 2008
+#              Naohisa Goto <ng@bioruby.org>
+#
+# License::    The Ruby License
+#
+# == Description
+#
+# This file contains Bio::PAML::Yn00, a wrapper class running yn00.
+#
+# == References
+#
+# * http://abacus.gene.ucl.ac.uk/software/paml.html
+#
+
+require 'bio/appl/paml/common'
+
+module Bio::PAML
+
+  # == Description
+  #
+  # Bio::PAML::Yn00 is a wrapper for running PAML yn00 program. 
+  #
+  # Because most of the methods in this class are inherited from
+  # Bio::PAML::Common, see documents of Bio::PAML::Common for details.
+  #
+  # == Examples
+  #
+  # Example 1:
+  #
+  #   require 'bio'
+  #   # Reads multi-fasta formatted file and gets a Bio::Alignment object.
+  #   alignment = Bio::FlatFile.open(Bio::Alignment::MultiFastaFormat,
+  #                                  'example.fst').alignment
+  #   # Creates a Yn00 object
+  #   baseml = Bio::PAML::Yn00.new
+  #   # Sets parameters
+  #   baseml.parameters[:verbose] = 1
+  #   baseml.parameters[:icode] = 0
+  #   # You can also set many parameters at a time.
+  #   baseml.parameters.update({ :weighting => 0, :commonf3x4 => 0 })
+  #   # Executes yn00 with the alignment
+  #   report = yn00.query(alignment)
+  #
+  class Yn00 < Common
+
+    autoload  :Report,  'bio/appl/paml/yn00/report'
+
+    # Default program name
+    DEFAULT_PROGRAM = 'yn00'.freeze
+
+    # Default parameters when running baseml.
+    #
+    # The parameters whose values are different from the baseml defalut
+    # value (described in pamlDOC.pdf) in PAML 4.1 are:
+    #  seqfile, outfile, treefile, ndata, noisy, verbose
+    #
+    DEFAULT_PARAMETERS = {
+      # Essential argumemts
+      :seqfile     => nil,
+      :outfile     => nil,
+      # Optional arguments
+      :verbose     => 1,
+      :icode       => 0,
+      :weighting   => 0,
+      :commonf3x4  => 0
+    }
+
+    # Runs the program on the internal parameters with the specified
+    # sequence alignment.
+    # Note that parameters[:seqfile] and parameters[:outfile]
+    # are always modified.
+    #
+    # For other important information, see the document of
+    # Bio::PAML::Common#query.
+    #
+    # ---
+    # *Arguments*:
+    # * (required) _alignment_: Bio::Alignment object or similar object
+    # *Returns*:: Report object
+    def query(alignment)
+      super(alignment)
+    end
+
+    # Runs the program on the internal parameters with the specified
+    # sequence alignment as a String object.
+    #
+    # For other important information, see the document of
+    # query and Bio::PAML::Common#query_by_string methods.
+    #
+    # ---
+    # *Arguments*:
+    # * (required) _alignment_: Bio::Alignment object or similar object
+    # *Returns*:: Report object
+    def query_by_string(alignment = nil)
+      super(alignment)
+    end
+
+  end #class Yn00
+end #module Bio::PAML
+

data/lib/bio/appl/paml/yn00/report.rb

@@ -0,0 +1,32 @@
+#
+# = bio/appl/paml/baseml/report.rb - parser class for PAML program yn00
+#
+# Copyright::  Copyright (C) 2008
+#              Naohisa Goto <ng@bioruby.org>
+#
+# License::    The Ruby License
+#
+# == Description
+#
+# This file contains Bio::PAML::Yn00::Report, a parser class for a result
+# of yn00.
+#
+# == References
+#
+# * http://abacus.gene.ucl.ac.uk/software/paml.html
+#
+
+require 'bio/appl/paml/yn00'
+
+module Bio::PAML
+  class Yn00
+
+    # UNDER CONSTRUCTION.
+    #
+    # Bio::PAML::Yn00::Report is a parser class for a yn00 result.
+    #
+    class Report < Bio::PAML::Common::Report
+    end #class Report
+
+  end #class Yn00
+end #module Bio::PAML

data/lib/bio/appl/phylip/alignment.rb

@@ -0,0 +1,133 @@
+#
+# = bio/appl/phylip/alignment.rb - phylip multiple alignment format parser
+#
+# Copyright:: Copyright (C) 2006
+#             GOTO Naohisa <ng@bioruby.org>
+#
+# License:: The Ruby License
+#
+#  $Id: alignment.rb,v 1.2 2007/04/05 23:35:40 trevor Exp $
+#
+# = About Bio::Phylip::PhylipFormat
+#
+# Please refer document of Bio::Phylip::PhylipFormat class.
+#
+
+module Bio
+  module Phylip
+
+    # This is phylip multiple alignment format parser.
+    # The two formats, interleaved and non-interleaved, are
+    # automatically determined.
+    #
+    class PhylipFormat
+
+      # create a new object from a string
+      def initialize(str)
+        @data = str.strip.split(/(?:\r\n|\r|\n)/)
+        @first_line = @data.shift
+        @number_of_sequences, @alignment_length =
+          @first_line.to_s.strip.split(/\s+/).collect { |x| x.to_i }
+      end
+
+      # number of sequences
+      attr_reader :number_of_sequences
+
+      # alignment length
+      attr_reader :alignment_length
+
+      # If the alignment format is "interleaved", returns true.
+      # If not, returns false.
+      # It would mistake to determine if the alignment is very short.
+      def interleaved?
+        unless defined? @interleaved_flag then
+          if /\A +/ =~ @data[1].to_s then
+            @interleaved_flag = false
+          else
+            @interleaved_flag = true
+          end
+        end
+        @interleaved_flag
+      end
+
+      # Gets the alignment. Returns a Bio::Alignment object.
+      def alignment
+        unless defined? @alignment then
+          do_parse
+          a = Bio::Alignment.new
+          (0...@number_of_sequences).each do |i|
+            a.add_seq(@sequences[i], @sequence_names[i])
+          end
+          @alignment = a
+        end
+        @alignment
+      end
+
+      private
+
+      def do_parse
+        if interleaved? then
+          do_parse_interleaved
+        else
+          do_parse_noninterleaved
+        end
+      end
+
+      def do_parse_interleaved
+        first_block = @data[0, @number_of_sequences]
+        @data[0, @number_of_sequences] = ''
+        @sequence_names = Array.new(@number_of_sequences) { '' }
+        @sequences = Array.new(@number_of_sequences) do
+          ' ' * @alignment_length
+        end
+        first_block.each_with_index do |x, i|
+          n, s = x.split(/ +/, 2)
+          if s.nil? # if there is no space separator, assume 10 character name
+            n = x[0..9]
+            s = x[10..(x.length-1)]
+          end
+          @sequence_names[i] = n
+          @sequences[i].replace(s.gsub(/\s+/, ''))
+        end
+        i = 0
+        @data.each do |x|
+          if x.strip.length <= 0 then
+            i = 0
+          else
+            @sequences[i] << x.gsub(/\s+/, '')
+            i = (i + 1) % @number_of_sequences
+          end
+        end
+        @data.clear
+        true
+      end
+
+      def do_parse_noninterleaved
+        @sequence_names = Array.new(@number_of_sequences) { '' }
+        @sequences = Array.new(@number_of_sequences) do
+          ' ' * @alignment_length
+        end
+        curseq = nil
+        i = 0
+        @data.each do |x|
+          next if x.strip.length <= 0
+          if !curseq or
+              curseq.length > @alignment_length or /^\s/ !~ x then
+            p i
+            n, s = x.strip.split(/ +/, 2)
+            @sequence_names[i] = n
+            curseq = @sequences[i]
+            curseq.replace(s.gsub(/\s+/, ''))
+            i += 1
+          else
+            curseq << x.gsub(/\s+/, '')
+          end
+        end
+        @data.clear
+        true
+      end
+
+    end #class PhylipFormat
+  end #module Phylip
+end #module Bio
+

data/lib/bio/appl/phylip/distance_matrix.rb

@@ -0,0 +1,96 @@
+#
+# = bio/appl/phylip/distance_matrix.rb - phylip distance matrix parser
+#
+# Copyright:: Copyright (C) 2006
+#             GOTO Naohisa <ng@bioruby.org>
+#
+# License:: The Ruby License
+#
+#  $Id: distance_matrix.rb,v 1.3 2007/04/05 23:35:40 trevor Exp $
+#
+# = About Bio::Phylip::DistanceMatrix
+#
+# Please refer document of Bio::Phylip::DistanceMatrix class.
+#
+
+require 'matrix'
+
+module Bio
+  module Phylip
+
+    # This is a parser class for phylip distance matrix data
+    # created by dnadist, protdist, or restdist commands.
+    #
+    class DistanceMatrix
+
+      # creates a new distance matrix object
+      def initialize(str)
+        data = str.strip.split(/(?:\r\n|\r|\n)/)
+        @otus = data.shift.to_s.strip.to_i
+        prev = nil
+        data.collect! do |x|
+          if /\A +/ =~ x and prev then
+            prev.concat x.strip.split(/\s+/)
+            nil
+          else
+            prev = x.strip.split(/\s+/)
+            prev
+          end
+        end
+        data.compact!
+        if data.size != @otus then
+          raise "inconsistent data (OTUs=#{@otus} but #{data.size} rows)"
+        end
+        @otu_names = data.collect { |x| x.shift }
+        mat = data.collect do |x|
+          if x.size != @otus then
+            raise "inconsistent data (OTUs=#{@otus} but #{x.size} columns)"
+          end
+          x.collect { |y| y.to_f }
+        end
+        @matrix = Matrix.rows(mat, false)
+        @original_matrix = Matrix.rows(data, false)
+      end
+
+      # distance matrix (returns Ruby's Matrix object)
+      attr_reader :matrix
+
+      # matrix contains values as original strings.
+      # Use it when you doubt precision of floating-point numbers.
+      attr_reader :original_matrix
+
+      # number of OTUs
+      attr_reader :otus
+
+      # names of OTUs
+      attr_reader :otu_names
+
+      # Generates a new phylip distance matrix formatted text as a string.
+      def self.generate(matrix, otu_names = nil, options = {})
+        if matrix.row_size != matrix.column_size then
+          raise "must be a square matrix"
+        end
+        otus = matrix.row_size
+        names = (0...otus).collect do |i|
+          name = ((otu_names and otu_names[i]) or "OTU#{i.to_s}")
+          name
+        end
+        data = (0...otus).collect do |i|
+          x = (0...otus).collect { |j|  sprintf("%9.6f", matrix[i, j]) }
+          x.unshift(sprintf("%-10s", names[i])[0, 10])
+
+          str = x[0, 7].join(' ') + "\n"
+          7.step(otus + 1, 7) do |k|
+            str << ' ' + x[k, 7].join(' ') + "\n"
+          end
+          str
+        end
+        sprintf("%5d\n", otus) + data.join('')
+      end
+        
+    end #class DistanceMatrix
+
+  end #module Phylip
+
+end #module Bio
+

data/lib/bio/appl/probcons.rb

@@ -0,0 +1,41 @@
+#
+# = bio/appl/probcons.rb - ProbCons application wrapper class
+#
+# Copyright:: Copyright (C) 2006-2007
+#             Jeffrey Blakeslee and John Conery University of Oregon <jeffb@uoregon.edu>
+#             Naohisa Goto <ng@bioruby.org>
+# License::   The Ruby License
+#
+#  $Id: probcons.rb,v 1.1 2007/07/16 12:25:50 ngoto Exp $
+#
+# Bio::Probcons is a wrapper class to execute ProbCons
+# (Probabilistic Consistency-based Multiple Alignment
+# of Amino Acid Sequences).
+#
+# == References
+# 
+# * http://probcons.stanford.edu/
+# * Do, C.B., Mahabhashyam, M.S.P., Brudno, M., and Batzoglou, S.
+#   ProbCons: Probabilistic Consistency-based Multiple Sequence Alignment.
+#   Genome Research 15: 330-340, 2005.
+#
+
+
+module Bio
+
+  # Bio::Probcons is a wrapper class to execute PROBCONS
+  # (Probabilistic Consistency-based Multiple Alignment
+  # of Amino Acid Sequences).
+  #
+  # Please refer documents in bio/apple/probcons.rb for references.
+  class Probcons < Bio::Alignment::FactoryTemplate::FileInStdoutOut
+
+    # default program name
+    DEFAULT_PROGRAM = 'probcons'.freeze
+
+    # default report parser
+    DEFAULT_PARSER = Bio::Alignment::MultiFastaFormat
+
+  end #class Probcons
+
+end #module Bio