Cleanup LinearMicroScheduler in preparation for TraverseByLoci inclusion.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@634 348d0f76-0448-11de-a6fe-93d51630548a

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

@@ -38,15 +38,17 @@ public class SAMDataSource implements SimpleDataSource {
     /** our log, which we want to capture anything from this class */
     protected static Logger logger = Logger.getLogger(SAMDataSource.class);
 
-    // are we set to locus mode or read mode for dividing
-    private boolean locusMode = true;
-
     // How strict should we be with SAM/BAM parsing?
     protected SAMFileReader.ValidationStringency strictness = SAMFileReader.ValidationStringency.SILENT;
 
     // our list of readers
     private final List<File> samFileList = new ArrayList<File>();
 
+    /**
+     * SAM header file.
+     */
+    private final SAMFileHeader header;
+
     // used for the reads case, the last count of reads retrieved
     private long readsTaken = 0;
 
@@ -87,6 +89,7 @@ public class SAMDataSource implements SimpleDataSource {
 
         }
 
+        header  = createHeaderMerger().getMergedHeader();
     }
 
     /**
@@ -120,18 +123,13 @@ public class SAMDataSource implements SimpleDataSource {
     public MergingSamRecordIterator2 seekLocus(GenomeLoc location) throws SimpleDataSourceLoadException {
 
         // right now this is pretty damn heavy, it copies the file list into a reader list every time
-        SamFileHeaderMerger headerMerger = CreateHeader();
+        SamFileHeaderMerger headerMerger = createHeaderMerger();
 
         // make a merging iterator for this record
         MergingSamRecordIterator2 iter = new MergingSamRecordIterator2(headerMerger);
 
-
         // we do different things for locus and read modes
-        if (locusMode) {
+        iter.queryOverlapping(location.getContig(), (int) location.getStart(), (int) location.getStop() + 1);
-            iter.queryOverlapping(location.getContig(), (int) location.getStart(), (int) location.getStop() + 1);
-        } else {
-            iter.queryContained(location.getContig(), (int) location.getStart(), (int) location.getStop() + 1);
-        }
 
         // return the iterator
         return iter;
@@ -168,6 +166,13 @@ public class SAMDataSource implements SimpleDataSource {
         includeUnmappedReads = seeUnMappedReads;
     }
 
+    /**
+     * Gets the (potentially merged) SAM file header.
+     * @return SAM file header.
+     */
+    public SAMFileHeader getHeader() {
+        return header; 
+    }
 
     /**
      * <p>
@@ -180,7 +185,7 @@ public class SAMDataSource implements SimpleDataSource {
     private BoundedReadIterator seekRead(ReadShard shard) throws SimpleDataSourceLoadException {
 
         BoundedReadIterator bound = null;
-        SamFileHeaderMerger headerMerger = CreateHeader();
+        SamFileHeaderMerger headerMerger = createHeaderMerger();
         MergingSamRecordIterator2 iter = null;
 
         if (!intoUnmappedReads) {
@@ -193,7 +198,7 @@ public class SAMDataSource implements SimpleDataSource {
             if (iter != null) {
                 iter.close();
             }
-            iter = new MergingSamRecordIterator2(CreateHeader());
+            iter = new MergingSamRecordIterator2(createHeaderMerger());
             bound = toUnmappedReads(shard.getSize(), iter);
         }
 
@@ -204,7 +209,7 @@ public class SAMDataSource implements SimpleDataSource {
         return bound;
     }
 
-    private SamFileHeaderMerger CreateHeader() {
+    private SamFileHeaderMerger createHeaderMerger() {
         // TODO: make extremely less horrible
         List<SAMFileReader> lst = GetReaderList();
 
@@ -314,7 +319,7 @@ public class SAMDataSource implements SimpleDataSource {
                         iter.close();
                         // now merge the headers
                         // right now this is pretty damn heavy, it copies the file list into a reader list every time
-                        SamFileHeaderMerger mg = CreateHeader();
+                        SamFileHeaderMerger mg = createHeaderMerger();
                         iter = new MergingSamRecordIterator2(mg);
                         iter.queryContained(lastReadPos.getContig(), 1, Integer.MAX_VALUE);
                         return new BoundedReadIterator(iter,readCount);

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

@@ -7,7 +7,6 @@ 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.StingSAMIterator;
-import org.broadinstitute.sting.gatk.traversals.TraversalEngine;
 import org.broadinstitute.sting.gatk.traversals.TraverseByReads;
 import org.broadinstitute.sting.gatk.traversals.TraverseLociByReference;
 import org.broadinstitute.sting.gatk.traversals.TraverseReads;
@@ -24,15 +23,8 @@ import java.util.List;
 /** A micro-scheduling manager for single-threaded execution of a traversal. */
 public class LinearMicroScheduler extends MicroScheduler {
 
-    private TraversalEngine traversalEngine = null;
-
     private boolean isAReadWalker = false;
 
-    /** get the traversal engine */
-    public TraversalEngine getTraversalEngine() {
-        return traversalEngine;
-    }
-
     /**
      * Create a new linear microscheduler to process the given reads and reference.
      *
@@ -62,31 +54,16 @@ public class LinearMicroScheduler extends MicroScheduler {
         ShardStrategy shardStrategy = getShardStrategy(reference, locations);
         SAMDataSource dataSource = getReadsDataSource();
 
-        boolean walkerInitialized = false;
+        walker.initialize();
-        Object accumulator = null;
+        Object accumulator = ((LocusWalker<?, ?>) walker).reduceInit();
 
         for (Shard shard : shardStrategy) {
 
-            StingSAMIterator readShard = null;
+            StingSAMIterator readShard = dataSource.seek(shard);
-            try {
-                readShard = dataSource.seek(shard);
-            }
-            catch (SimpleDataSourceLoadException ex) {
-                throw new RuntimeException(ex);
-            }
 
             ReferenceProvider referenceProvider = new ReferenceProvider(reference, shard.getGenomeLoc());
             LocusContextProvider locusProvider = new LocusContextProvider(readShard);
 
-            // set the sam header of the traversal engine
-            traversalEngine.setSAMHeader(readShard.getHeader());
-
-            if (!walkerInitialized) {
-                walker.initialize();
-                accumulator = ((LocusWalker<?, ?>) walker).reduceInit();
-                walkerInitialized = true;
-            }
-
             if (!isAReadWalker) {
                 accumulator = ((TraverseLociByReference) traversalEngine).traverse(walker, shard, referenceProvider, locusProvider, accumulator);
             } else {

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

@@ -37,6 +37,9 @@ public abstract class MicroScheduler {
 
     protected IndexedFastaSequenceFile reference;
 
+    protected TraversalEngine traversalEngine = null;
+
+
     /**
      * MicroScheduler factory function.  Create a microscheduler appropriate for reducing the
      * selected walker.
@@ -70,7 +73,9 @@ public abstract class MicroScheduler {
      * of the microscheduler shouldn't need to know anything about the traversal engine.
      * @return The traversal engine.
      */
-    public abstract TraversalEngine getTraversalEngine();
+    public TraversalEngine getTraversalEngine() {
+        return traversalEngine;
+    }
 
     /**
      * Walks a walker over the given list of intervals.
@@ -115,6 +120,12 @@ public abstract class MicroScheduler {
         catch( FileNotFoundException ex ) {
             throw new RuntimeException( ex );
         }
+
+        // Side effect: initialize the traversal engine with reads data.
+        // TODO: Give users a dedicated way of getting the header so that the MicroScheduler
+        //       doesn't have to bend over backward providing legacy getters and setters.
+        traversalEngine.setSAMHeader(dataSource.getHeader());
+
         return dataSource;
     }
 

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

@@ -29,10 +29,4 @@ import net.sf.samtools.util.CloseableIterator;
  * 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();
 }