Added a interface that all our SAMRecord iterators should try to code to. This is in the effort to keep our code generic

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@609 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/dataSources/simpleDataSources/SAMDataSource.java

@@ -269,7 +269,7 @@ public class SAMDataSource implements SimpleDataSource {
     private BoundedReadIterator fastMappedReadSeek(long readCount, MergingSamRecordIterator2 iter) throws SimpleDataSourceLoadException {
         BoundedReadIterator bound;// is this the first time we're doing this?
         if (lastReadPos == null) {
-            lastReadPos = new GenomeLoc(iter.getMergedHeader().getSequenceDictionary().getSequence(0).getSequenceIndex(), 0, 0);
+            lastReadPos = new GenomeLoc(iter.getHeader().getSequenceDictionary().getSequence(0).getSequenceIndex(), 0, 0);
             iter.queryContained(lastReadPos.getContig(), 1, -1);
             bound = new BoundedReadIterator(iter, readCount);
             this.readsTaken = readCount;
@@ -339,7 +339,7 @@ public class SAMDataSource implements SimpleDataSource {
             }
             // in case we're run out of reads, get out
             else {
-                throw new StingException("Danger");
+                throw new StingException("Danger: weve run out reads in fastMappedReadSeek");
                 //return null;
             }
             bound = new BoundedReadIterator(iter, readCount);

java/src/org/broadinstitute/sting/gatk/executive/LinearMicroScheduler.java

@@ -7,9 +7,11 @@ import org.broadinstitute.sting.gatk.dataSources.shards.ShardStrategy;
 import org.broadinstitute.sting.gatk.dataSources.simpleDataSources.SAMDataSource;
 import org.broadinstitute.sting.gatk.dataSources.simpleDataSources.SimpleDataSourceLoadException;
 import org.broadinstitute.sting.gatk.iterators.MergingSamRecordIterator2;
+import org.broadinstitute.sting.gatk.iterators.StingSAMIterator;
 import org.broadinstitute.sting.gatk.traversals.TraversalEngine;
 import org.broadinstitute.sting.gatk.traversals.TraverseLociByReference;
 import org.broadinstitute.sting.gatk.walkers.LocusWalker;
+import org.broadinstitute.sting.gatk.walkers.ReadWalker;
 import org.broadinstitute.sting.gatk.walkers.Walker;
 import org.broadinstitute.sting.utils.GenomeLoc;
 
@@ -21,7 +23,7 @@ import java.util.List;
  */
 public class LinearMicroScheduler extends MicroScheduler {
 
-    private TraverseLociByReference traversalEngine = null;
+    private TraversalEngine traversalEngine = null;
 
     public TraversalEngine getTraversalEngine() {
         return traversalEngine;
@@ -46,12 +48,15 @@ public class LinearMicroScheduler extends MicroScheduler {
         ShardStrategy shardStrategy = getShardStrategy( reference, locations );
         SAMDataSource dataSource = getReadsDataSource();
 
+        // determine if we're a read walker: they get a slightly different, but not in any way worse execute methodology
+        boolean readwalker = (walker instanceof ReadWalker) ? true : false;
+
         boolean walkerInitialized = false;
         Object accumulator = null;
 
         for(Shard shard: shardStrategy) {
 
-            MergingSamRecordIterator2 readShard = null;
+            StingSAMIterator readShard = null;
             try {
                 readShard = (MergingSamRecordIterator2)dataSource.seek( shard );
             }
@@ -63,7 +68,7 @@ public class LinearMicroScheduler extends MicroScheduler {
             LocusContextProvider locusProvider = new LocusContextProvider( readShard );
 
             // set the sam header of the traversal engine
-            traversalEngine.setSAMHeader(readShard.getMergedHeader());
+            traversalEngine.setSAMHeader(readShard.getHeader());
 
             if (!walkerInitialized) {
                 walker.initialize();
@@ -71,7 +76,7 @@ public class LinearMicroScheduler extends MicroScheduler {
                 walkerInitialized = true;
             }
 
-            accumulator = traversalEngine.traverse( walker, shard, referenceProvider, locusProvider, accumulator );
+            accumulator = ((TraverseLociByReference)traversalEngine).traverse( walker, shard, referenceProvider, locusProvider, accumulator );
             readShard.close();
         }
 

java/src/org/broadinstitute/sting/gatk/iterators/BoundedReadIterator.java

@@ -33,7 +33,7 @@ import java.util.Iterator;
  * This class implements a read iterator that is bounded by the number of reads
  * it will produce over the iteration.
  */
-public class BoundedReadIterator implements CloseableIterator<SAMRecord>, Iterable<SAMRecord> {
+public class BoundedReadIterator implements StingSAMIterator {
 
     // the genome loc we're bounding
     final private long readCount;
@@ -70,11 +70,11 @@ public class BoundedReadIterator implements CloseableIterator<SAMRecord>, Iterab
     }
 
 
-    public SAMFileHeader getMergedHeader() {
+    public SAMFileHeader getHeader() {
         // todo: this is bad, we need an iterface out there for samrecords that supports getting the header,
         // regardless of the merging
         if (iterator instanceof MergingSamRecordIterator2)
-            return ((MergingSamRecordIterator2)iterator).getMergedHeader();
+            return ((MergingSamRecordIterator2)iterator).getHeader();
         else
             return null;
     }

java/src/org/broadinstitute/sting/gatk/iterators/MergingSamRecordIterator2.java

@@ -15,7 +15,6 @@ import edu.mit.broad.picard.sam.ReservedTagConstants;
 import edu.mit.broad.picard.sam.SamFileHeaderMerger;
 import edu.mit.broad.picard.util.PeekableIterator;
 import net.sf.samtools.*;
-import net.sf.samtools.util.CloseableIterator;
 import org.apache.log4j.Logger;
 
 import java.lang.reflect.Constructor;
@@ -29,7 +28,7 @@ import java.util.PriorityQueue;
  * iterable stream. The underlying iterators/files must all have the same sort order unless
  * the requested output format is unsorted, in which case any combination is valid.
  */
-public class MergingSamRecordIterator2 implements CloseableIterator<SAMRecord>, Iterable<SAMRecord> {
+public class MergingSamRecordIterator2 implements StingSAMIterator {
     protected PriorityQueue<ComparableSamRecordIterator> pq = null;
     protected final SamFileHeaderMerger samHeaderMerger;
     protected final SAMFileHeader.SortOrder sortOrder;
@@ -152,6 +151,7 @@ public class MergingSamRecordIterator2 implements CloseableIterator<SAMRecord>,
         }
 
         final ComparableSamRecordIterator iterator = this.pq.poll();
+
         if (iterator == null) {
             return null;
         }
@@ -244,7 +244,7 @@ public class MergingSamRecordIterator2 implements CloseableIterator<SAMRecord>,
     }
 
     /** Returns the merged header that the merging iterator is working from. */
-    public SAMFileHeader getMergedHeader() {
+    public SAMFileHeader getHeader() {
         return this.samHeaderMerger.getMergedHeader();
     }
 

java/src/org/broadinstitute/sting/gatk/iterators/StingSAMIterator.java

@@ -0,0 +1,38 @@
+package org.broadinstitute.sting.gatk.iterators;
+
+import net.sf.samtools.SAMRecord;
+import net.sf.samtools.SAMFileHeader;
+import net.sf.samtools.util.CloseableIterator;
+/**
+ *
+ * User: aaron
+ * Date: May 6, 2009
+ * Time: 5:30:41 PM
+ *
+ * The Broad Institute
+ * SOFTWARE COPYRIGHT NOTICE AGREEMENT 
+ * This software and its documentation are copyright 2009 by the
+ * Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
+ *
+ * This software is supplied without any warranty or guaranteed support whatsoever. Neither
+ * the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
+ *
+ */
+
+/**
+ * @author aaron
+ * @version 1.0
+ * @date May 6, 2009
+ * <p/>
+ * Interface ClosableGetHeaderIterator
+ * <p/>
+ * A descriptions should go here. Blame aaron if it's missing.
+ */
+public interface StingSAMIterator extends CloseableIterator<SAMRecord>, Iterable<SAMRecord> {
+
+    /**
+     * gets the header from the iterator
+     * @return the samfileheader for the iterator, null if one is not available
+     */
+    public SAMFileHeader getHeader();
+}

java/test/org/broadinstitute/sting/gatk/traversals/TraverseReadsTest.java

@@ -148,30 +148,30 @@ public class TraverseReadsTest extends BaseTest {
             Shard shard = shardStrategy.next();
             BoundedReadIterator readIter = null;
             try {
-                readIter = (BoundedReadIterator)dataSource.seek(shard);
+                readIter = (BoundedReadIterator) dataSource.seek(shard);
             }
             catch (SimpleDataSourceLoadException ex) {
                 throw new RuntimeException(ex);
             }
 
-                //LocusContextProvider locusProvider = new LocusContextProvider( readIter );
+            //LocusContextProvider locusProvider = new LocusContextProvider( readIter );
 
-                // set the sam header of the traversal engine
+            // set the sam header of the traversal engine
-                traversalEngine.setSAMHeader(readIter.getMergedHeader());
+            traversalEngine.setSAMHeader(readIter.getHeader());
 
-                if (!walkerInitialized) {
+            if (!walkerInitialized) {
-                    countReadWalker.initialize();
+                countReadWalker.initialize();
-                    accumulator = ((ReadWalker<?, ?>) countReadWalker).reduceInit();
+                accumulator = ((ReadWalker<?, ?>) countReadWalker).reduceInit();
-                    walkerInitialized = true;
+                walkerInitialized = true;
 
-                }
+            }
-                if (shard == null) {
+            if (shard == null) {
-                    fail("Shard == null");
+                fail("Shard == null");
-                }
+            }
 
 
-                accumulator = traversalEngine.traverse(countReadWalker, shard, readIter, accumulator);
+            accumulator = traversalEngine.traverse(countReadWalker, shard, readIter, accumulator);
-                readIter.close();
+            readIter.close();
 
         }
 
@@ -181,7 +181,7 @@ public class TraverseReadsTest extends BaseTest {
         if (!(accumulator instanceof Integer)) {
             fail("Count read walker should return an interger.");
         }
-        if (((Integer)accumulator) != 9721) {
+        if (((Integer) accumulator) != 9721) {
             fail("there should be 9721 mapped reads in the index file");
         }
     }
@@ -229,30 +229,30 @@ public class TraverseReadsTest extends BaseTest {
             Shard shard = shardStrategy.next();
             BoundedReadIterator readIter = null;
             try {
-                readIter = (BoundedReadIterator)dataSource.seek(shard);
+                readIter = (BoundedReadIterator) dataSource.seek(shard);
             }
             catch (SimpleDataSourceLoadException ex) {
                 throw new RuntimeException(ex);
             }
 
-                //LocusContextProvider locusProvider = new LocusContextProvider( readIter );
+            //LocusContextProvider locusProvider = new LocusContextProvider( readIter );
 
-                // set the sam header of the traversal engine
+            // set the sam header of the traversal engine
-                traversalEngine.setSAMHeader(readIter.getMergedHeader());
+            traversalEngine.setSAMHeader(readIter.getHeader());
 
-                if (!walkerInitialized) {
+            if (!walkerInitialized) {
-                    countReadWalker.initialize();
+                countReadWalker.initialize();
-                    accumulator = ((ReadWalker<?, ?>) countReadWalker).reduceInit();
+                accumulator = ((ReadWalker<?, ?>) countReadWalker).reduceInit();
-                    walkerInitialized = true;
+                walkerInitialized = true;
 
-                }
+            }
-                if (shard == null) {
+            if (shard == null) {
-                    fail("Shard == null");
+                fail("Shard == null");
-                }
+            }
 
 
-                accumulator = traversalEngine.traverse(countReadWalker, shard, readIter, accumulator);
+            accumulator = traversalEngine.traverse(countReadWalker, shard, readIter, accumulator);
-                readIter.close();
+            readIter.close();
 
         }
 
@@ -262,7 +262,7 @@ public class TraverseReadsTest extends BaseTest {
         if (!(accumulator instanceof Integer)) {
             fail("Count read walker should return an interger.");
         }
-        if (((Integer)accumulator) != 10000) {
+        if (((Integer) accumulator) != 10000) {
             fail("there should be 9721 mapped reads in the index file");
         }
     }