X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=code_ruby%2Flib%2Fmaasha%2Fseq%2Ftrim.rb;h=2ca94542b65c3ed3e6d536b460095295b1f13181;hb=3656f15bb59e2eb7eff628bae117db6479b2f03f;hp=a322dc943c313c09840973e53c1f524027bb337d;hpb=e643e1f687105eea667af1e3d9281842319099a5;p=biopieces.git

diff --git a/code_ruby/lib/maasha/seq/trim.rb b/code_ruby/lib/maasha/seq/trim.rb
index a322dc9..2ca9454 100644
--- a/code_ruby/lib/maasha/seq/trim.rb
+++ b/code_ruby/lib/maasha/seq/trim.rb
@@ -1,3 +1,4 @@
+# Copyright (C) 2007-2013 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -22,7 +23,11 @@
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 require 'inline'
+require 'maasha/seq/backtrack'
 
+include BackTrack
+
+# Error class for all exceptions to do with Trim.
 class TrimError < StandardError; end
 
 # Module containing methods for end trimming sequences with suboptimal quality
@@ -33,9 +38,9 @@ module Trim
   def quality_trim_right(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos = trim_right_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos = trim_right_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
 
-    self.subseq(0, pos)
+    self[0, pos]
   end
 
   # Method to progressively trim a Seq object sequence from the right end until
@@ -43,9 +48,12 @@ module Trim
   def quality_trim_right!(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos = trim_right_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos = trim_right_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
+
+    self.seq  = self.seq[0, pos]
+    self.qual = self.qual[0, pos] if self.qual
 
-    self.subseq!(0, pos)
+    self
   end
 
   # Method to progressively trim a Seq object sequence from the left end until
@@ -53,9 +61,9 @@ module Trim
   def quality_trim_left(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos = trim_left_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos = trim_left_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
 
-    self.subseq(pos)
+    self[pos .. -1]
   end
 
   # Method to progressively trim a Seq object sequence from the left end until
@@ -63,9 +71,12 @@ module Trim
   def quality_trim_left!(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos = trim_left_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos = trim_left_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
+
+    self.seq  = self.seq[pos .. -1]
+    self.qual = self.qual[pos .. -1] if self.qual
 
-    self.subseq!(pos)
+    self
   end
 
   # Method to progressively trim a Seq object sequence from both ends until a
@@ -73,12 +84,12 @@ module Trim
   def quality_trim(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos_right = trim_right_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
-    pos_left  = trim_left_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos_right = trim_right_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
+    pos_left  = trim_left_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
 
     pos_left = pos_right if pos_left > pos_right
 
-    self.subseq(pos_left, pos_right - pos_left)
+    self[pos_left ... pos_right]
   end
 
   # Method to progressively trim a Seq object sequence from both ends until a
@@ -86,12 +97,47 @@ module Trim
   def quality_trim!(min_qual, min_len = 1)
     check_trim_args(min_qual)
 
-    pos_right = trim_right_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
-    pos_left  = trim_left_pos_c(self.qual, self.length, min_qual, min_len, SCORE_BASE)
+    pos_right = trim_right_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
+    pos_left  = trim_left_pos_c(self.qual, self.length, min_qual, min_len, Seq::SCORE_BASE)
 
     pos_left = pos_right if pos_left > pos_right
 
-    self.subseq!(pos_left, pos_right - pos_left)
+    self.seq  = self.seq[pos_left ... pos_right]
+    self.qual = self.qual[pos_left ... pos_right] if self.qual
+
+    self
+  end
+
+  # Method to locate a pattern in a sequence and trim all sequence to the left
+  # including the pattern.
+  def patmatch_trim_left!(pattern, options = {})
+    match = self.patmatch(pattern, options)
+
+    if match
+      stop = self.length
+
+      self.seq  = self.seq[match.pos + match.length .. stop]
+      self.qual = self.qual[match.pos + match.length .. stop] if self.qual
+
+      return self
+    end
+
+    nil
+  end
+
+  # Method to locate a pattern in a sequence and trim all sequence to the right
+  # including the pattern.
+  def patmatch_trim_right!(pattern, options = {})
+    match = self.patmatch(pattern, options)
+
+    if match
+      self.seq  = self.seq[0 ... match.pos]
+      self.qual = self.qual[0 ... match.pos] if self.qual
+
+      return self
+    end
+
+    nil
   end
 
   private
@@ -101,8 +147,8 @@ module Trim
   def check_trim_args(min_qual)
     raise TrimError, "no sequence"      if self.seq.nil?
     raise TrimError, "no quality score" if self.qual.nil?
-    unless (SCORE_MIN .. SCORE_MAX).include? min_qual
-      raise TrimError, "minimum quality value: #{min_qual} out of range #{SCORE_MIN} .. #{SCORE_MAX}"
+    unless (Seq::SCORE_MIN .. Seq::SCORE_MAX).include? min_qual
+      raise TrimError, "minimum quality value: #{min_qual} out of range #{Seq::SCORE_MIN} .. #{Seq::SCORE_MAX}"
     end
   end
 
@@ -131,7 +177,7 @@ module Trim
         {
           c = 0;
 
-          while ((c < min_len) && ((c + i) < len) && (qual[len - (c + i) - 1] - score_base > min_qual))
+          while ((c < min_len) && ((c + i) < len) && (qual[len - (c + i) - 1] - score_base >= min_qual))
             c++;
 
           if (c == min_len)
@@ -169,7 +215,7 @@ module Trim
         {
           c = 0;
 
-          while ((c < min_len) && ((c + i) < len) && (qual[c + i] - score_base > min_qual))
+          while ((c < min_len) && ((c + i) < len) && (qual[c + i] - score_base >= min_qual))
             c++;
 
           if (c == min_len)