a couple of light changes to GenomeLocSortedSet.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1708 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/GenomeLocSortedSet.java

@@ -2,6 +2,7 @@ package org.broadinstitute.sting.utils;
 
 import net.sf.samtools.SAMSequenceDictionary;
 import net.sf.samtools.SAMSequenceRecord;
+import org.apache.log4j.Logger;
 
 import java.util.*;
 
@@ -34,8 +35,10 @@ import java.util.*;
  *         that element.
  */
 public class GenomeLocSortedSet extends AbstractSet<GenomeLoc> {
+    private static Logger logger = Logger.getLogger(GenomeLocSortedSet.class);
+
     // our private storage for the GenomeLoc's
-    private final ArrayList<GenomeLoc> mArray = new ArrayList<GenomeLoc>();
+    private List<GenomeLoc> mArray = new ArrayList<GenomeLoc>();
 
     /** default constructor */
     public GenomeLocSortedSet() {
@@ -119,6 +122,9 @@ public class GenomeLocSortedSet extends AbstractSet<GenomeLoc> {
                     (e.getStart() < g.getStart()) && !haveAdded) {
                 mArray.add(mArray.indexOf(g), e);
                 return true;
+            } else if (haveAdded && ((e.getContigIndex() > e.getContigIndex()) ||
+                    (g.getContigIndex() == e.getContigIndex() && e.getStart() > g.getStart()))) {
+                return true;
             }
         }
         /** we're at the end and we haven't found locations that should fall after it,
@@ -262,4 +268,49 @@ public class GenomeLocSortedSet extends AbstractSet<GenomeLoc> {
         return ret;
     }
 
+
+    public boolean addAllRegions(List<GenomeLoc> locations) {
+        this.mArray.addAll(locations);
+        Collections.sort(this.mArray);
+        this.mArray = GenomeLocSortedSet.mergeOverlappingLocations(this.mArray);
+        return true;
+    }
+
+/**
+     * merge a list of genome locs that may be overlapping, returning the list of unique genomic locations
+     *
+     * @param raw the unchecked genome loc list
+     *
+     * @return the list of merged locations
+     */
+    public static List<GenomeLoc> mergeOverlappingLocations(final List<GenomeLoc> raw) {
+        logger.debug("  Raw locations are: " + Utils.join(", ", raw));
+        if (raw.size() <= 1)
+            return raw;
+        else {
+            ArrayList<GenomeLoc> merged = new ArrayList<GenomeLoc>();
+            Iterator<GenomeLoc> it = raw.iterator();
+            GenomeLoc prev = it.next();
+            while (it.hasNext()) {
+                GenomeLoc curr = it.next();
+                if (prev.contiguousP(curr)) {
+                    prev = prev.merge(curr);
+                } else {
+                    merged.add(prev);
+                    prev = curr;
+                }
+            }
+            merged.add(prev);
+            return merged;
+        }
+    }
+
+    /**
+     * convert this object to a list
+     * @return the lists
+     */
+    public List<GenomeLoc> toList() {
+        return this.mArray;
+    }
+
 }

java/test/org/broadinstitute/sting/utils/GenomeLocSortedSetTest.java

@@ -4,6 +4,7 @@ import net.sf.samtools.SAMFileHeader;
 import org.broadinstitute.sting.BaseTest;
 import org.broadinstitute.sting.utils.sam.ArtificialSAMUtils;
 import static org.junit.Assert.assertTrue;
+import static org.junit.Assert.assertEquals;
 import org.junit.Before;
 import org.junit.Test;
 
@@ -203,4 +204,31 @@ public class GenomeLocSortedSetTest extends BaseTest {
         }
         assertTrue(seqNumber == GenomeLocSortedSetTest.NUMBER_OF_CHROMOSOMES);
     }
+
+    @Test
+    public void testAddAll() {
+        mSortedSet = GenomeLocSortedSet.createSetFromSequenceDictionary(this.header.getSequenceDictionary());
+        GenomeLocSortedSet set = GenomeLocSortedSet.createSetFromSequenceDictionary(this.header.getSequenceDictionary());
+        // we should have sequence
+        assertTrue(mSortedSet.size() == GenomeLocSortedSetTest.NUMBER_OF_CHROMOSOMES);
+        mSortedSet.addAllRegions(set.toList());
+        assertTrue(mSortedSet.size() == GenomeLocSortedSetTest.NUMBER_OF_CHROMOSOMES);                        
+    }
+
+    @Test
+    public void testAddAll2() {
+        mSortedSet = new GenomeLocSortedSet();
+        GenomeLocSortedSet mSortedSet2 = new GenomeLocSortedSet();
+        for (int x=0; x < 200; x = x + 2) {
+            mSortedSet.add(GenomeLocParser.createGenomeLoc(1,x));
+        }
+        assertEquals(100, mSortedSet.size());
+        for (int x=1; x < 201; x = x + 2) {
+            mSortedSet2.add(GenomeLocParser.createGenomeLoc(1,x));
+        }
+        assertEquals(100, mSortedSet2.size());
+        mSortedSet.addAllRegions(mSortedSet2.toList());
+        assertEquals(1, mSortedSet.size());
+    }
+
 }