From 9f519af06d7a64fed078732cd11947210153a1af Mon Sep 17 00:00:00 2001
From: kcibul <kcibul@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Mon, 15 Mar 2010 15:40:09 +0000
Subject: [PATCH] new method to filter out overlapping PE reads

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3002 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../sting/utils/pileup/ReadBackedPileup.java  | 35 +++++++++++++++++++
 1 file changed, 35 insertions(+)

diff --git a/java/src/org/broadinstitute/sting/utils/pileup/ReadBackedPileup.java b/java/src/org/broadinstitute/sting/utils/pileup/ReadBackedPileup.java
index 06918b8dc..6b99ceb3f 100755
--- a/java/src/org/broadinstitute/sting/utils/pileup/ReadBackedPileup.java
+++ b/java/src/org/broadinstitute/sting/utils/pileup/ReadBackedPileup.java
@@ -173,6 +173,41 @@ public class ReadBackedPileup implements Iterable<PileupElement> {
         }
     }
 
+    /**
+     * Returns a new ReadBackedPileup where only one read from an overlapping read
+     * pair is retained.  If the two reads in question disagree to their basecall,
+     * neither read is retained.  If they agree on the base, the read with the higher
+     * quality observation is retained
+     *
+     * @return the newly filtered pileup
+     */
+    public ReadBackedPileup getOverlappingFragementFilteredPileup() {
+        Map<String, PileupElement> filteredPileup = new HashMap<String, PileupElement>();
+
+        for ( PileupElement p : pileup ) {
+            String readName = p.getRead().getReadName();
+
+            // if we've never seen this read before, life is good
+            if (!filteredPileup.containsKey(readName)) {
+                filteredPileup.put(readName, p);
+            } else {
+                PileupElement existing = filteredPileup.get(readName);
+
+                // if the reads disagree at this position, throw them both out.  Otherwise
+                // keep the element with the higher quality score
+                if (existing.getBase() != p.getBase()) {
+                    filteredPileup.remove(readName);
+                } else {
+                    if (existing.getQual() < p.getQual()) {
+                        filteredPileup.put(readName, p);
+                    }
+                }
+            }
+        }
+
+        return new ReadBackedPileup(loc, new ArrayList<PileupElement>(filteredPileup.values()));
+    }
+
     /**
      * Returns a new ReadBackedPileup that is free of mapping quality zero reads in this pileup.  Note that this
      * does not copy the data, so both ReadBackedPileups should not be changed.  Doesn't make an unnecessary copy