forester/ruby/evoruby/lib/evo/sequence/sequence.rb

   1 #
   2 # = lib/evo/sequence/sequence.rb - Sequence class
   3 #
   4 # Copyright::  Copyright (C) 2006-2007 Christian M. Zmasek
   5 # License::    GNU Lesser General Public License (LGPL)
   6 #
   7 # $Id: sequence.rb,v 1.10 2009/01/07 02:48:20 cmzmasek Exp $
   8
   9 require 'set'
  10
  11 module Evoruby
  12
  13     class Sequence
  14
  15         def initialize( name,
  16                 molecular_sequence_str,
  17                 accession = nil,
  18                 accession_source = nil,
  19                 taxonomy = nil,
  20                 symbol = nil,
  21                 secondary_accession = nil,
  22                 secondary_accession_source = nil )
  23             @name               = String.new( name.strip() )
  24             @molecular_sequence = String.new( molecular_sequence_str )
  25             if ( accession == nil )
  26                 @accession = String.new()
  27             else
  28                 @accession = String.new( accession.strip() )
  29             end
  30             if ( accession_source == nil )
  31                 @accession_source = String.new()
  32             else
  33                 @accession_source = String.new( accession_source.strip() )
  34             end
  35             @taxonomy = taxonomy
  36             if ( symbol == nil )
  37                 @symbol = String.new()
  38             else
  39                 @symbol = String.new( symbol.strip() )
  40             end
  41             if ( secondary_accession == nil )
  42                 @secondary_accession = String.new()
  43             else
  44                 @secondary_accession = String.new( secondary_accession.strip() )
  45             end
  46             if ( secondary_accession_source == nil )
  47                 @secondary_accession_source = String.new()
  48             else
  49                 @secondary_accession_source = String.new( secondary_accession_source.strip() )
  50             end
  51         end
  52
  53         def copy
  54             if get_taxonomy == nil
  55                 Sequence.new( get_name, get_sequence_as_string, get_accession, get_accession_source, nil, get_symbol, get_secondary_accession, get_secondary_accession_source )
  56             else
  57                 Sequence.new( get_name, get_sequence_as_string, get_accession, get_accession_source, get_taxonomy.copy, get_symbol, get_secondary_accession, get_secondary_accession_source )
  58             end
  59         end
  60
  61         def get_name()
  62             @name
  63         end
  64
  65         def set_name( name )
  66             @name = name
  67         end
  68
  69         def get_sequence_as_string()
  70             @molecular_sequence
  71         end
  72
  73         def get_accession()
  74             @accession
  75         end
  76
  77         def get_accession_source()
  78             @accession_source
  79         end
  80
  81         def get_secondary_accession()
  82             @secondary_accession
  83         end
  84
  85         def get_secondary_accession_source()
  86             @secondary_accession_source
  87         end
  88
  89         def get_symbol()
  90             @symbol
  91         end
  92
  93         def get_taxonomy()
  94             @taxonomy
  95         end
  96
  97         def get_length()
  98             @molecular_sequence.length
  99         end
 100
 101         def get_residue( position )
 102             get_slice( position, 1 )
 103         end
 104
 105         def get_character_code( position )
 106             @molecular_sequence.getbyte( position )
 107         end
 108
 109         def get_gap_ratio()
 110             return get_gap_length().to_f / get_length()
 111         end
 112
 113         def get_gap_length()
 114             counter = 0
 115             for i in 0 ... get_length()
 116                 if ( Util.is_aa_gap_character?( get_character_code( i ) ) )
 117                     counter += 1
 118                 end
 119             end
 120             return counter;
 121         end
 122
 123         def delete_residue!( position )
 124             if ( position < 0 || position >= get_length() )
 125                 error_msg = "attempt to delete residue at postion out of range"
 126                 raise ArgumentError, error_msg
 127             end
 128             @molecular_sequence.slice!( position )
 129         end
 130
 131         def get_slice( start, length )
 132             if ( start < 0 || start + length > get_length() )
 133                 error_msg = "attempt to get sequence residue(s) at postion out of range"
 134                 raise ArgumentError, error_msg
 135             end
 136             @molecular_sequence.slice( start, length )
 137         end
 138
 139         def get_slice!( start, length )
 140             if ( start < 0 || start + length > get_length() )
 141                 error_msg = "attempt to get sequence residue(s) at postion out of range"
 142                 raise ArgumentError, error_msg
 143             end
 144             @molecular_sequence.slice!( start, length )
 145         end
 146
 147         def get_subsequence( first, last )
 148             if ( last < first )
 149                 error_msg = "attempt to get subsequence from " + first + " to " + last
 150                 raise ArgumentError, error_msg
 151             end
 152             return Sequence.new( get_name, @molecular_sequence.slice( first, last - first + 1 ) )
 153         end
 154
 155         def append!( molecular_sequence_str )
 156             @molecular_sequence.concat( molecular_sequence_str )
 157         end
 158
 159         def to_str
 160             return "[" + @name + "] " + @molecular_sequence
 161         end
 162
 163         def to_fasta
 164             return ">" + @name + Constants::LINE_DELIMITER  + @molecular_sequence
 165         end
 166
 167
 168     end # class Sequence
 169
 170 end # module Evoruby