added Seq.shuffle method and tests

[biopieces.git] / code_ruby / lib / maasha / seq.rb

diff --git a/code_ruby/lib/maasha/seq.rb b/code_ruby/lib/maasha/seq.rb
index 913f4d979928f96472558192a48da3e434a15f7c..e72250d596cc5d1dd55adac34d3a15b3503048fa 100644 (file)
--- a/code_ruby/lib/maasha/seq.rb
+++ b/code_ruby/lib/maasha/seq.rb
@@ -27,6 +27,9 @@ require 'maasha/seq/digest'
 require 'maasha/seq/trim'
 require 'narray'
 
+autoload :BackTrack, 'maasha/seq/backtrack.rb'
+autoload :Dynamic,   'maasha/seq/dynamic.rb'
+
 # Residue alphabets
 DNA     = %w[a t c g]
 RNA     = %w[a u c g]
@@ -64,15 +67,16 @@ TRANS_TAB11 = {
   "GTG" => "V", "GCG" => "A", "GAG" => "E", "GGG" => "G"
 }
 
-# Quality scores bases
-SCORE_BASE = 64
-SCORE_MIN  = 0
-SCORE_MAX  = 40
 
 # Error class for all exceptions to do with Seq.
 class SeqError < StandardError; end
 
 class Seq
+  # Quality scores bases
+  SCORE_BASE = 64
+  SCORE_MIN  = 0
+  SCORE_MAX  = 40
+
   include Digest
   include Trim
 
@@ -386,7 +390,12 @@ class Seq
     seq_new
   end
 
-  # Method to shuffle a sequence readomly inline.
+  # Method to return a new Seq object with shuffled sequence.
+  def shuffle
+    Seq.new(self.seq_name, self.seq.split('').shuffle!.join, self.type, self.qual)
+  end
+
+  # Method to shuffle a sequence randomly inline.
   def shuffle!
     self.seq = self.seq.split('').shuffle!.join
     self
@@ -535,13 +544,30 @@ class Seq
   # Method that determines if a quality score string can be
   # absolutely identified as base 33.
   def qual_base33?
-    self.qual.match(/[!-:]/)
+    self.qual.match(/[!-:]/) ? true : false
   end
 
   # Method that determines if a quality score string can be
   # absolutely identified as base 64.
   def qual_base64?
-    self.qual.match(/[K-h]/)
+    self.qual.match(/[K-h]/) ? true : false
+  end
+
+  # Method to determine if a quality score is valid.
+  def qual_valid?(encoding)
+    raise SeqError, "Missing qual" if self.qual.nil?
+
+    case encoding.downcase
+    when "sanger"     then return true if self.qual.match(/^[!-~]*$/)
+    when "454"        then return true if self.qual.match(/^[@-~]*$/)
+    when "solexa"     then return true if self.qual.match(/^[;-~]*$/)
+    when "illumina13" then return true if self.qual.match(/^[@-~]*$/)
+    when "illumina15" then return true if self.qual.match(/^[@-~]*$/)
+    when "illumina18" then return true if self.qual.match(/^[!-~]*$/)
+    else raise SeqError, "unknown quality score encoding: #{encoding}"
+    end
+
+    false
   end
 
   # Method to convert quality scores inbetween formats.
@@ -549,7 +575,7 @@ class Seq
   # 454        base 64, range  0-40 
   # Solexa     base 64, range -5-40 
   # Illumina13 base 64, range  0-40 
-  # Illumina15 base 64, range  3-40 
+  # Illumina15 base 64, range  0-40 
   # Illumina18 base 33, range  0-41 
   def convert_scores!(from, to)
     unless from == to
@@ -572,7 +598,7 @@ class Seq
       when "illumina13" then na_qual += 64
       when "illumina15" then na_qual += 64
       when "illumina18" then na_qual += 33
-      else raise SeqError, "unknown quality score encoding: #{from}"
+      else raise SeqError, "unknown quality score encoding: #{to}"
       end
 
       self.qual = na_qual.to_s