From f1109e90709da70b63032768760f5b89a7ffee8d Mon Sep 17 00:00:00 2001
From: aaron <aaron@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Sat, 25 Jul 2009 22:36:29 +0000
Subject: [PATCH] Added the interator to SAMDataSource to prevent seeing
 dupplicate reads, only in a byReads traversal.  The iterator  discards any
 reads in the current interval that would have been seen in the previous
 interval.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1317 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../simpleDataSources/SAMDataSource.java         | 16 +++++++++++++++-
 .../gatk/iterators/IntervalOverlapIterator.java  |  6 +++++-
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java b/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
index 3341d573b..d8921047a 100755
--- a/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
+++ b/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
@@ -81,6 +81,8 @@ public class SAMDataSource implements SimpleDataSource {
     // A pool of SAM iterators.
     private SAMResourcePool resourcePool = null;
 
+    private GenomeLoc mLastInterval = null;
+
     /**
      * Returns a histogram of reads that were screened out, grouped by the nature of the error.
      * @return Histogram of reads.  Will not be null.
@@ -153,14 +155,26 @@ public class SAMDataSource implements SimpleDataSource {
                     reads.getDownsamplingFraction(),
                     reads.getSafetyChecking(),
                     reads.getSupplementalFilters());
-        } else if (shard.getShardType() == Shard.ShardType.LOCUS || shard.getShardType() == Shard.ShardType.INTERVAL) {
+        } else if (shard.getShardType() == Shard.ShardType.LOCUS) {
             iterator = seekLocus(shard.getGenomeLoc());
             iterator = applyDecoratingIterators(false,
                     iterator,
                     reads.getDownsamplingFraction(),
                     reads.getSafetyChecking(),
                     reads.getSupplementalFilters());
+        } else if (shard.getShardType() == Shard.ShardType.INTERVAL) {
+            iterator = seekLocus(shard.getGenomeLoc());
+            iterator = applyDecoratingIterators(false,
+                    iterator,
+                    reads.getDownsamplingFraction(),
+                    reads.getSafetyChecking(),
+                    reads.getSupplementalFilters());
+
+            // add the new overlapping detection iterator, if we have a last interval
+            if (mLastInterval != null && queryOverlapping) iterator = new IntervalOverlapIterator(iterator,mLastInterval,false);
+            mLastInterval = shard.getGenomeLoc();
         } else {
+
             throw new StingException("seek: Unknown shard type");
         }
 
diff --git a/java/src/org/broadinstitute/sting/gatk/iterators/IntervalOverlapIterator.java b/java/src/org/broadinstitute/sting/gatk/iterators/IntervalOverlapIterator.java
index 8b4f197be..bb2caab9c 100644
--- a/java/src/org/broadinstitute/sting/gatk/iterators/IntervalOverlapIterator.java
+++ b/java/src/org/broadinstitute/sting/gatk/iterators/IntervalOverlapIterator.java
@@ -4,6 +4,7 @@ import net.sf.samtools.SAMRecord;
 import org.broadinstitute.sting.gatk.Reads;
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.StingException;
 
 import java.util.Iterator;
 
@@ -103,7 +104,10 @@ public class IntervalOverlapIterator implements StingSAMIterator {
      * @return true if it overlaps, false otherwise
      */
     private boolean isOverlaping(SAMRecord rec) {
-        return mLoc.overlapsP(GenomeLocParser.createGenomeLoc(rec));
+        boolean overlap = mLoc.overlapsP(GenomeLocParser.createGenomeLoc(rec));
+        if (overlap && this.throwException)
+            throw new StingException("IntervalOverlapIterator found a overlapping read " + rec.getReadName() + " with overlap " + this.mLoc.toString());
+        return overlap;
     }
 
 }