More comprehensive tracking of position when bin trees are sparse.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5344 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/datasources/reads/BinTree.java

@@ -59,15 +59,6 @@ public class BinTree {
         this.bins = bins;
     }
 
-    public GATKBin getLowestLevelBin() {
-        for(int i = bins.length-1; i >= 0; i--) {
-            if(bins[i] != null)
-                return bins[i];
-        }
-        return null;
-    }
-
-
     /**
      * Retrieve the bins from the bin tree.
      * @return list of bins.
@@ -84,6 +75,31 @@ public class BinTree {
         return bins.length;
     }
 
+    /**
+     * Returns the start of the region covered by this bin tree.
+     * @return Start of the region covered by this bin tree.
+     */
+    public int getStart() {
+        return binTreeStart;
+    }
+
+    /**
+     * Returns the end of the region covered by this bin tree.
+     * @return End of the region covered by this bin tree.
+     */
+    public int getStop() {
+        return binTreeStop;
+    }
+
+    /**
+     * Returns true if the location of this bin tree is before the given position.
+     * @param locus Locus to test.
+     * @return True if this bin sits completely before the given locus; false otherwise.
+     */
+    public boolean isBefore(final GenomeLoc locus) {
+        return binTreeStop < locus.getStart();
+    }
+
     /**
      * Checks overlap between this bin tree and other bin trees.
      * @param position the position over which to detect overlap.

java/src/org/broadinstitute/sting/gatk/datasources/reads/LowMemoryIntervalSharder.java

@@ -95,8 +95,8 @@ public class LowMemoryIntervalSharder implements Iterator<FilePointer> {
                 GATKBAMIndex index = (GATKBAMIndex)dataSource.getIndex(reader);
                 BinTree binTree = getNextOverlappingBinTree((GATKBAMIndex)dataSource.getIndex(reader),currentLocus);
                 if(binTree != null) {
-                    coveredRegionStart = Math.max(coveredRegionStart,index.getFirstLocusInBin(binTree.getLowestLevelBin().toBin()));
-                    coveredRegionStop = Math.min(coveredRegionStop,index.getLastLocusInBin(binTree.getLowestLevelBin().toBin()));
+                    coveredRegionStart = Math.max(coveredRegionStart,binTree.getStart());
+                    coveredRegionStop = Math.min(coveredRegionStop,binTree.getStop());
                     coveredRegion = loci.getGenomeLocParser().createGenomeLoc(currentLocus.getContig(),coveredRegionStart,coveredRegionStop);
 
                     GATKBAMFileSpan fileSpan = generateFileSpan(index,binTree,currentLocus);
@@ -164,7 +164,7 @@ public class LowMemoryIntervalSharder implements Iterator<FilePointer> {
             return null;
 
         BinTree binTree = binTreeIterator.peek();
-        while(index.getLastLocusInBin(binTree.getLowestLevelBin().toBin()) < locus.getStart()) {
+        while(binTree.isBefore(locus)) {
             binTreeIterator.next(); // Before the point of interest.  Consume this one.
             binTree = binTreeIterator.peek();
         }

java/src/org/broadinstitute/sting/gatk/datasources/reads/ReadShardStrategy.java

@@ -96,7 +96,7 @@ public class ReadShardStrategy implements ShardStrategy {
         this.locations = locations;
 
         if(locations != null)
-            filePointerIterator = IntervalSharder.shardIntervals(this.dataSource,locations);
+            filePointerIterator = SAMDataSource.TRY_LOW_MEMORY_SHARDING ? new LowMemoryIntervalSharder(this.dataSource,locations) : IntervalSharder.shardIntervals(this.dataSource,locations);
         else
             filePointerIterator = filePointers.iterator();