From 3f75fc4e83c14346f4e8687f6e78692ee623d97f Mon Sep 17 00:00:00 2001
From: ebanks <ebanks@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Mon, 6 Apr 2009 03:51:35 +0000
Subject: [PATCH] Unfortunately, because BWA occasionally outputs crazy reads,
 we need to make sure not to have an ArrayIndexOutOfBoundsException thrown.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@297 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../sting/playground/gatk/walkers/MismatchHistoWalker.java    | 4 +++-
 java/src/org/broadinstitute/sting/utils/Utils.java            | 1 +
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/java/src/org/broadinstitute/sting/playground/gatk/walkers/MismatchHistoWalker.java b/java/src/org/broadinstitute/sting/playground/gatk/walkers/MismatchHistoWalker.java
index 9165ac8cd..bf6c3ce1f 100755
--- a/java/src/org/broadinstitute/sting/playground/gatk/walkers/MismatchHistoWalker.java
+++ b/java/src/org/broadinstitute/sting/playground/gatk/walkers/MismatchHistoWalker.java
@@ -37,10 +37,12 @@ public class MismatchHistoWalker extends ReadWalker<Integer, Integer> {
 
             int start = read.getAlignmentStart()-1;
             int stop  = read.getAlignmentEnd();
+            // sometimes BWA outputs screwy reads
+            if ( stop >= context.getReferenceContig().getBases().length )
+                return 0;
 
             List<Byte> refSeq = Utils.subseq(context.getReferenceContig().getBases(), start, stop);
             List<Byte> readBases = Utils.subseq(read.getReadBases());
-
             assert(refSeq.size() == readBases.size());
 
             // it's actually faster to reallocate a resized array than to use ArrayLists...
diff --git a/java/src/org/broadinstitute/sting/utils/Utils.java b/java/src/org/broadinstitute/sting/utils/Utils.java
index 81c3d1726..64d2b6738 100755
--- a/java/src/org/broadinstitute/sting/utils/Utils.java
+++ b/java/src/org/broadinstitute/sting/utils/Utils.java
@@ -163,6 +163,7 @@ public class Utils {
     }
 
     public static ArrayList<Byte> subseq(byte[] fullArray, int start, int end) {
+        assert end < fullArray.length;
         ArrayList<Byte> dest = new ArrayList<Byte>(end - start + 1);
         for (int i = start; i < end; i++) {
             dest.add(fullArray[i]);