From c5de06a6417c58ac917bfa499d2a905317fc88c3 Mon Sep 17 00:00:00 2001
From: chartl <chartl@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Tue, 21 Jun 2011 14:07:58 +0000
Subject: [PATCH] Fixing up the RefSeqCodec so a bad entry in RefSeq (some
 transcripts are odd and have a negative length which may signify something
 special (?) ) doesn't cause failure, but issues a warning instead.
 Integration tests pass.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@6021 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../sting/gatk/refdata/features/refseq/RefSeqCodec.java   | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java b/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
index 044258c6c..461aab9a5 100644
--- a/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
+++ b/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
@@ -6,6 +6,7 @@ import org.broad.tribble.readers.LineReader;
 import org.broadinstitute.sting.gatk.refdata.ReferenceDependentFeatureCodec;
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.Utils;
 import org.broadinstitute.sting.utils.exceptions.UserException;
 
 import java.util.ArrayList;
@@ -35,7 +36,12 @@ public class RefSeqCodec implements ReferenceDependentFeatureCodec<RefSeqFeature
         String fields[] = line.split("\t");
         if (fields.length < 3) throw new TribbleException("RefSeq (decodeLoc) : Unable to parse line -> " + line + ", we expected at least 3 columns, we saw " + fields.length);
         String contig_name = fields[2];
-        return new RefSeqFeature(genomeLocParser.createGenomeLoc(contig_name, Integer.parseInt(fields[4])+1, Integer.parseInt(fields[5])));
+        try {
+            return new RefSeqFeature(genomeLocParser.createGenomeLoc(contig_name, Integer.parseInt(fields[4])+1, Integer.parseInt(fields[5])));
+        } catch ( UserException.MalformedGenomeLoc e ) {
+            Utils.warnUser("RefSeq file is potentially incorrect, as some transcripts or exons have a negative length ("+fields[2]+")");
+            return null;
+        }
     }
 
     /** Fills this object from a text line in RefSeq (UCSC) text dump file */