From 069554efe5765dfbdb2023141188966abcdfacda Mon Sep 17 00:00:00 2001
From: Andrey Sivachenko <asivache@broadinstitute.org>
Date: Wed, 17 Aug 2011 14:05:19 -0400
Subject: [PATCH 1/3] somatic indel detector does not die on reads that are too
 long (likely contain a huge deletion) anymore; instead print a warning and
 ignore the read

---
 .../walkers/indels/SomaticIndelDetectorWalker.java  | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
index 64cff2c0d..7e8985e03 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
@@ -469,8 +469,17 @@ public class SomaticIndelDetectorWalker extends ReadWalker<Integer,Integer> {
                 // let's double check now that the read fits after the shift
                 if ( read.getAlignmentEnd() > normal_context.getStop()) {
                     // ooops, looks like the read does not fit into the window even after the latter was shifted!!
-                    throw new UserException.BadArgumentValue("window_size", "Read "+read.getReadName()+": out of coverage window bounds. Probably window is too small, so increase the value of the window_size argument.\n"+
-                                             "Read length="+read.getReadLength()+"; cigar="+read.getCigarString()+"; start="+
+                    // we used to die over such reads and require user to run with larger window size. Now we
+                    // just print a warning and discard the read (this means that our counts can be slightly off in
+                    // th epresence of such reads)
+                    //throw new UserException.BadArgumentValue("window_size", "Read "+read.getReadName()+": out of coverage window bounds. Probably window is too small, so increase the value of the window_size argument.\n"+
+                    //                         "Read length="+read.getReadLength()+"; cigar="+read.getCigarString()+"; start="+
+                    //                         read.getAlignmentStart()+"; end="+read.getAlignmentEnd()+
+                    //                         "; window start (after trying to accomodate the read)="+normal_context.getStart()+"; window end="+normal_context.getStop());
+                    System.out.println("WARNING: Read "+read.getReadName()+
+                             " is out of coverage window bounds. Probably window is too small and the window_size value must be increased.\n"+
+                             "  The read is ignored in this run (so all the counts/statistics reported will not include it).\n"+
+                                             "  Read length="+read.getReadLength()+"; cigar="+read.getCigarString()+"; start="+
                                              read.getAlignmentStart()+"; end="+read.getAlignmentEnd()+
                                              "; window start (after trying to accomodate the read)="+normal_context.getStart()+"; window end="+normal_context.getStop());
                 }

From 710d34633e951ea6261890652401924175af7ba7 Mon Sep 17 00:00:00 2001
From: Andrey Sivachenko <asivache@broadinstitute.org>
Date: Wed, 17 Aug 2011 15:16:23 -0400
Subject: [PATCH 2/3] now the reads that are too long are truly ignored (fix of
 the fix)

---
 .../sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java    | 1 +
 1 file changed, 1 insertion(+)

diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
index 7e8985e03..b2d30235a 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/indels/SomaticIndelDetectorWalker.java
@@ -482,6 +482,7 @@ public class SomaticIndelDetectorWalker extends ReadWalker<Integer,Integer> {
                                              "  Read length="+read.getReadLength()+"; cigar="+read.getCigarString()+"; start="+
                                              read.getAlignmentStart()+"; end="+read.getAlignmentEnd()+
                                              "; window start (after trying to accomodate the read)="+normal_context.getStart()+"; window end="+normal_context.getStop());
+                    return 1;
                 }
             }
 

From a423546cdd2c759d0d5d0994a9e5ce8d781cff1d Mon Sep 17 00:00:00 2001
From: Andrey Sivachenko <asivache@broadinstitute.org>
Date: Wed, 17 Aug 2011 15:17:31 -0400
Subject: [PATCH 3/3] fix: RefSeq contains records with zero coding length and
 the refsec codec/feature used to crash on those; now such records are
 ignored, with warning printed (once)

---
 .../refdata/features/refseq/RefSeqCodec.java     | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/public/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java b/public/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
index 461aab9a5..89ee65532 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java
@@ -1,5 +1,6 @@
 package org.broadinstitute.sting.gatk.refdata.features.refseq;
 
+import org.apache.commons.io.filefilter.FalseFileFilter;
 import org.broad.tribble.Feature;
 import org.broad.tribble.TribbleException;
 import org.broad.tribble.readers.LineReader;
@@ -20,7 +21,7 @@ public class RefSeqCodec implements ReferenceDependentFeatureCodec<RefSeqFeature
      * The parser to use when resolving genome-wide locations.
      */
     private GenomeLocParser genomeLocParser;
-
+    private boolean zero_coding_length_user_warned = false;
     /**
      * Set the parser to use when resolving genetic data.
      * @param genomeLocParser The supplied parser.
@@ -60,9 +61,20 @@ public class RefSeqCodec implements ReferenceDependentFeatureCodec<RefSeqFeature
         else if ( fields[3].length()==1 && fields[3].charAt(0)=='-') feature.setStrand(-1);
         else throw new UserException.MalformedFile("Expected strand symbol (+/-), found: "+fields[3] + " for line=" + line);
 
+        int coding_start = Integer.parseInt(fields[6])+1;
+        int coding_stop = Integer.parseInt(fields[7]);
+
+        if ( coding_start > coding_stop ) {
+            if ( ! zero_coding_length_user_warned ) {
+                Utils.warnUser("RefSeq file contains transcripts with zero coding length. "+
+                        "Such transcripts will be ignored (this warning is printed only once)");
+                zero_coding_length_user_warned = true;
+            }
+            return null;
+        }
 
         feature.setTranscript_interval(genomeLocParser.createGenomeLoc(contig_name, Integer.parseInt(fields[4])+1, Integer.parseInt(fields[5])));
-        feature.setTranscript_coding_interval(genomeLocParser.createGenomeLoc(contig_name, Integer.parseInt(fields[6])+1, Integer.parseInt(fields[7])));
+        feature.setTranscript_coding_interval(genomeLocParser.createGenomeLoc(contig_name, coding_start, coding_stop));
         feature.setGene_name(fields[12]);
         String[] exon_starts = fields[9].split(",");
         String[] exon_stops = fields[10].split(",");