From: martinahansen <martinahansen@74ccb610-7750-0410-82ae-013aeee3265d>
Date: Fri, 8 Feb 2013 07:48:39 +0000 (+0000)
Subject: fixed quality score validity check
X-Git-Url: https://git.donarmstrong.com/?p=biopieces.git;a=commitdiff_plain;h=c54a2cc53ee2a7db8028ef593be593e527dc50c2

fixed quality score validity check

git-svn-id: http://biopieces.googlecode.com/svn/trunk@2091 74ccb610-7750-0410-82ae-013aeee3265d
---

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index 44a69ed..96be2c3 100755
--- a/bp_bin/read_fastq
+++ b/bp_bin/read_fastq
@@ -65,12 +65,15 @@ Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
             else
               raise SeqError, "Could not auto-detect quality score encoding"
             end
-          elsif first
-            #raise SeqError, "Quality score outside valid range" unless entry.qual_valid?(encoding)
-            first = false
           end
 
           entry.convert_scores!(encoding, 'illumina13')
+
+          if first
+            raise SeqError, "Quality score outside valid range" unless entry.qual_valid?('illumina13')
+            first = false
+          end
+
           output.puts entry.to_bp
           num += 1
 
diff --git a/code_ruby/lib/maasha/seq.rb b/code_ruby/lib/maasha/seq.rb
index f2d93d1..8833acb 100644
--- a/code_ruby/lib/maasha/seq.rb
+++ b/code_ruby/lib/maasha/seq.rb
@@ -552,12 +552,12 @@ class Seq
     raise SeqError, "Missing qual" if self.qual.nil?
 
     case encoding.downcase
-    when "sanger"     then return true if self.qual.match(/^[!-I]*$/)
-    when "454"        then return true if self.qual.match(/^[@-h]*$/)
-    when "solexa"     then return true if self.qual.match(/^[;-h]*$/)
-    when "illumina13" then return true if self.qual.match(/^[@-h]*$/)
-    when "illumina15" then return true if self.qual.match(/^[@-h]*$/)
-    when "illumina18" then return true if self.qual.match(/^[!-J]*$/)
+    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
 
diff --git a/code_ruby/test/maasha/test_seq.rb b/code_ruby/test/maasha/test_seq.rb
index 171e6b1..a7145ef 100755
--- a/code_ruby/test/maasha/test_seq.rb
+++ b/code_ruby/test/maasha/test_seq.rb
@@ -520,12 +520,12 @@ class TestSeq < Test::Unit::TestCase
   end
 
   def test_Seq_qual_valid_returns_correctly
-    tests = [["sanger",      0, 40, 33],
-             ["454",         0, 40, 64],
-             ["solexa",     -5, 40, 64],
-             ["illumina13",  0, 40, 64],
-             ["illumina15",  0, 40, 64],
-             ["illumina18",  0, 41, 33]]
+    tests = [["sanger",      0, 93, 33],
+             ["454",         0, 62, 64],
+             ["solexa",     -5, 62, 64],
+             ["illumina13",  0, 62, 64],
+             ["illumina15",  0, 62, 64],
+             ["illumina18",  0, 93, 33]]
 
     tests.each do |test|
       @entry.qual = (test[1] + test[-1]).chr + (test[2] + test[-1]).chr