fixed bug in pair assembler

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

diff --git a/code_ruby/lib/maasha/seq/assemble.rb b/code_ruby/lib/maasha/seq/assemble.rb
index 7745b5bed7141bd1ad4d500c71b9ac585b2ed0b2..c6f528b8ce9800061cd5aabdfada95b1911a5c11 100644 (file)
--- a/code_ruby/lib/maasha/seq/assemble.rb
+++ b/code_ruby/lib/maasha/seq/assemble.rb
@@ -22,32 +22,43 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+require 'inline'
+require 'maasha/seq/ambiguity'
+
+# Class containing methods to assemble two overlapping sequences into a single.
 class Assemble
+  extend Ambiguity
+
+  # Class method to assemble two Seq objects.
   def self.pair(entry1, entry2, options = {})
     assemble = self.new(entry1, entry2, options)
     assemble.match
   end
 
+  # Method to initialize an Assembly object.
   def initialize(entry1, entry2, options)
     @entry1  = entry1
     @entry2  = entry2
     @options = options
     @options[:mismatches_max] ||= 0
     @options[:overlap_min]    ||= 1
-    @options[:overlap_max]    ||= entry1.length
-    @options[:overlap_max]      = [@options[:overlap_max], entry1.length, entry2.length].min
   end
 
+  # Method to locate overlapping matches between two sequences.
   def match
-    overlap = @options[:overlap_max]
-
-    na_seq1 = NArray.to_na(@entry1.seq.downcase, "byte")
-    na_seq2 = NArray.to_na(@entry2.seq.downcase, "byte")
+    if @options[:overlap_max]
+      overlap = [@options[:overlap_max], @entry1.length, @entry2.length].min
+    else
+      overlap = [@entry1.length, @entry2.length].min
+    end
 
     while overlap >= @options[:overlap_min]
-      hamming_dist = (na_seq1[-1 * overlap .. -1] ^ na_seq2[0 ... overlap]).count_true
+      mismatches_max = (overlap * @options[:mismatches_max] * 0.01).round
+      
+      offset1 = @entry2.length - overlap
+      offset2 = 0
 
-      if hamming_dist <= percent2real(overlap, @options[:mismatches_max])
+      if match_C(@entry1.seq, @entry2.seq, offset1, offset2, overlap, mismatches_max)
         entry_left  = @entry1[0 ... @entry1.length - overlap]
         entry_right = @entry2[overlap .. -1]
 
@@ -60,8 +71,11 @@ class Assemble
           entry_overlap = @entry1[-1 * overlap .. -1]
         end
 
+        entry_left.seq.downcase!
+        entry_overlap.seq.upcase!
+        entry_right.seq.downcase!
         entry_merged          = entry_left + entry_overlap + entry_right
-        entry_merged.seq_name = @entry1.seq_name + ":overlap=#{overlap}:hamming=#{hamming_dist}"
+        entry_merged.seq_name = @entry1.seq_name + ":overlap=#{overlap}"
 
         return entry_merged
       end
@@ -70,10 +84,9 @@ class Assemble
     end
   end
 
-  def percent2real(length, percent)
-    (length * percent * 0.01).round
-  end
-
+  # Method to merge sequence and quality scores in an overlap.
+  # The residue with the highest score at mismatch positions is selected.
+  # The quality scores of the overlap are the mean of the two sequences.
   def merge_overlap(entry_overlap1, entry_overlap2)
     na_seq = NArray.byte(entry_overlap1.length, 2)
     na_seq[true, 0] = NArray.to_na(entry_overlap1.seq.downcase, "byte")
@@ -92,6 +105,59 @@ class Assemble
 
     merged
   end
+
+  inline do |builder|
+    add_ambiguity_macro(builder)
+
+    # C method for determining if two strings of equal length match
+    # given a maximum allowed mismatches and allowing for IUPAC
+    # ambiguity codes. Returns true if match, else false.
+    builder.c %{
+      VALUE match_C(
+        VALUE _string1,       // String 1
+        VALUE _string2,       // String 2
+        VALUE _offset1,       // Offset 1
+        VALUE _offset2,       // Offset 2
+        VALUE _length,        // String length
+        VALUE _max_mismatch   // Maximum mismatches
+      )
+      {
+        char         *string1      = StringValuePtr(_string1);
+        char         *string2      = StringValuePtr(_string2);
+        unsigned int  offset1      = FIX2UINT(_offset1);
+        unsigned int  offset2      = FIX2UINT(_offset2);
+        unsigned int  length       = FIX2UINT(_length);
+        unsigned int  max_mismatch = FIX2UINT(_max_mismatch);
+
+        unsigned int max_match = length - max_mismatch;
+        unsigned int match     = 0;
+        unsigned int mismatch  = 0;
+        unsigned int i         = 0;
+
+        for (i = 0; i < length; i++)
+        {
+          if (MATCH(string1[i + offset1], string2[i + offset2]))
+          {
+            match++;
+
+            if (match >= max_match) {
+              return Qtrue;
+            }
+          }
+          else
+          {
+            mismatch++;
+
+            if (mismatch > max_mismatch) {
+              return Qfalse;
+            }
+          }
+        }
+
+        return Qfalse;
+      }
+    }
+  end
 end