implemented hamming_dist in inline C

[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 f76466f583af2a18699f3aa4ee40f67fb88e250e..1d8f3c00594e9208f434251eb23c342413df9506 100644 (file)
--- a/code_ruby/lib/maasha/seq.rb
+++ b/code_ruby/lib/maasha/seq.rb
@@ -27,8 +27,12 @@ require 'maasha/seq/digest'
 require 'maasha/seq/trim'
 require 'narray'
 
-autoload :BackTrack, 'maasha/seq/backtrack.rb'
-autoload :Dynamic,   'maasha/seq/dynamic.rb'
+autoload :BackTrack,   'maasha/seq/backtrack'
+autoload :Dynamic,     'maasha/seq/dynamic'
+autoload :Homopolymer, 'maasha/seq/homopolymer'
+autoload :Hamming,     'maasha/seq/hamming'
+autoload :Levenshtein, 'maasha/seq/levenshtein'
+autoload :Ambiguity,   'maasha/seq/ambiguity'
 
 # Residue alphabets
 DNA     = %w[a t c g]
@@ -67,7 +71,6 @@ TRANS_TAB11 = {
   "GTG" => "V", "GCG" => "A", "GAG" => "E", "GGG" => "G"
 }
 
-
 # Error class for all exceptions to do with Seq.
 class SeqError < StandardError; end
 
@@ -383,8 +386,18 @@ class Seq
 
   # Method to determine the Hamming Distance between
   # two Sequence objects (case insensitive).
-  def hamming_distance(seq)
-    self.seq.upcase.hamming_distance(seq.seq.upcase)
+  def hamming_distance(entry, options = nil)
+    if options and options[:ambiguity]
+      Hamming.distance(self.seq, entry.seq)
+    else
+      self.seq.upcase.hamming_distance(entry.seq.upcase)
+    end
+  end
+
+  # Method to determine the Edit Distance between
+  # two Sequence objects (case insensitive).
+  def edit_distance(entry)
+    Levenshtein.distance(self.seq, entry.seq)
   end
 
   # Method that generates a random sequence of a given length and type.
@@ -416,6 +429,15 @@ class Seq
     self
   end
 
+  # Method to add two Seq objects.
+  def +(entry)
+    new_entry = Seq.new()
+    new_entry.seq  = self.seq  + entry.seq
+    new_entry.type = self.type              if self.type == entry.type
+    new_entry.qual = self.qual + entry.qual if self.qual and entry.qual
+    new_entry
+  end
+
   # Method to concatenate sequence entries.
   def <<(entry)
     raise SeqError, "sequences of different types" unless self.type == entry.type
@@ -427,6 +449,25 @@ class Seq
     self
   end
 
+  # Index method for Seq objects.
+  def [](*args)
+    entry = Seq.new
+    entry.seq_name = self.seq_name
+    entry.seq      = self.seq[*args]
+    entry.type     = self.type
+    entry.qual     = self.qual[*args] unless self.qual.nil?
+
+    entry
+  end
+
+  # Index assignment method for Seq objects.
+  def []=(*args, entry)
+    self.seq[*args]  = entry.seq[*args]
+    self.qual[*args] = entry.qual[*args] unless self.qual.nil?
+
+    self
+  end
+
   # Method that returns a subsequence of from a given start position
   # and of a given length.
   def subseq(start, length = self.length - start)
@@ -487,23 +528,6 @@ class Seq
     comp
   end
 
-  # Method that returns the length of the longest homopolymeric stretch
-  # found in a sequence.
-  def homopol_max(min = 1)
-    return 0 if self.seq.nil? or self.seq.empty?
-
-    found = false
-
-    self.seq.upcase.scan(/A{#{min},}|T{#{min},}|G{#{min},}|C{#{min},}|N{#{min},}/) do |match|
-      found = true
-      min   = match.size > min ? match.size : min
-    end
-
-    return 0 unless found
- 
-    min
-  end
-
   # Method that returns the percentage of hard masked residues
   # or N's in a sequence.
   def hard_mask