Added ability to sort reads on the fly

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@83 348d0f76-0448-11de-a6fe-93d51630548a

playground/java/src/org/broadinstitute/sting/gatk/GenomeAnalysisTK.java

@@ -26,6 +26,7 @@ public class GenomeAnalysisTK extends CommandLineProgram {
     @Option(shortName="DBSNP", doc="DBSNP file", optional=true) public String DBSNP_FILE = null;
     @Option(shortName="THREADED_IO", doc="If true, enables threaded I/O operations", optional=true) public String ENABLED_THREADED_IO = "false";
     @Option(shortName="U", doc="If true, enables unsafe operations, nothing will be checked at runtime.  You better know what you are doing if you set this flag.", optional=false) public String UNSAFE = "false";
+    @Option(shortName="SORT_ON_FLY", doc="If true, enables on fly sorting of reads file.", optional=false) public String ENABLED_SORT_ON_FLY = "false";
 
     public static HashMap<String, Object> MODULES = new HashMap<String,Object>();
     public static void addModule(final String name, final Object walker) {
@@ -103,6 +104,7 @@ public class GenomeAnalysisTK extends CommandLineProgram {
         }
 
         engine.setSafetyChecking(! UNSAFE.toLowerCase().equals("true"));
+        engine.setSortOnFly(! ENABLED_SORT_ON_FLY.toLowerCase().equals("true"));
 
         engine.initialize(ENABLED_THREADED_IO.toLowerCase().equals("true"));
         //engine.testReference();

playground/java/src/org/broadinstitute/sting/gatk/TraversalEngine.java

@@ -77,6 +77,8 @@ public class TraversalEngine {
 
     public boolean DEBUGGING = false;
     public boolean beSafeP = true;
+    public boolean SORT_ON_FLY = false;
+    public int MAX_ON_FLY_SORTS = 100000;
     public long N_RECORDS_TO_PRINT = 100000;
     public int THREADED_IO_BUFFER_SIZE = 10000;
 
@@ -117,6 +119,11 @@ public class TraversalEngine {
             System.out.printf("*** Turning off safety checking, I hope you know what you are doing.  Errors will result in debugging assert failures and other inscrutable messages...%n");
         this.beSafeP = beSafeP;
     }
+    public void setSortOnFly( final boolean SORT_ON_FLY ) {
+        if ( SORT_ON_FLY )
+            System.out.println("Sorting read file on the fly: max reads allowed is " + MAX_ON_FLY_SORTS);
+        this.SORT_ON_FLY = SORT_ON_FLY;
+    }
 
     // --------------------------------------------------------------------------------------------------------------
     //
@@ -306,9 +313,11 @@ public class TraversalEngine {
             throw new RuntimeIOException(ex);
         }
 
-        if ( beSafeP )
+        if ( SORT_ON_FLY )
+            samReadIter = new SortSamIterator(samReadIter, MAX_ON_FLY_SORTS);
+        else if ( beSafeP )
             samReadIter = new VerifyingSamIterator(samReadIter);
-
+        
         if ( THREADED_IO ) {
             System.out.printf("Enabling threaded I/O with buffer of %d reads%n", THREADED_IO_BUFFER_SIZE);
             samReadIter = new ThreadedIterator<SAMRecord>(samReadIter, THREADED_IO_BUFFER_SIZE);
@@ -455,7 +464,7 @@ public class TraversalEngine {
     }
 
     public void verifySortOrder(final boolean requiresSortedOrder) {
-        if ( beSafeP && samReader.getFileHeader().getSortOrder() != SAMFileHeader.SortOrder.coordinate ) {
+        if ( beSafeP && !SORT_ON_FLY && samReader.getFileHeader().getSortOrder() != SAMFileHeader.SortOrder.coordinate ) {
             final String msg = "SAM file is not sorted in coordinate order (according to header) Walker type with given arguments requires a sorted file for correct processing";
             if ( requiresSortedOrder || strictness == SAMFileReader.ValidationStringency.STRICT )
                 throw new RuntimeIOException(msg);

playground/java/src/org/broadinstitute/sting/gatk/iterators/SortSamIterator.java

@@ -0,0 +1,64 @@
+package org.broadinstitute.sting.gatk.iterators;
+
+import net.sf.samtools.SAMRecord;
+import net.sf.samtools.SAMFileReader;
+import net.sf.samtools.util.RuntimeIOException;
+
+import java.util.Arrays;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.Iterator;
+
+import org.broadinstitute.sting.utils.GenomeLoc;
+import org.broadinstitute.sting.utils.Utils;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: mdepristo
+ * Date: Mar 15, 2009
+ * Time: 6:02:31 PM
+ * To change this template use File | Settings | File Templates.
+ */
+public class SortSamIterator implements Iterator<SAMRecord> {
+
+    Iterator<ComparableSAMRecord> it;
+
+    public SortSamIterator(Iterator<SAMRecord> unsortedIter, int maxSorts) {
+
+        ArrayList<ComparableSAMRecord> list = new ArrayList<ComparableSAMRecord>();
+        while (unsortedIter.hasNext()) {
+            list.add(new ComparableSAMRecord(unsortedIter.next()));
+            // choose an arbitrary length to limit sorting for now
+            if (list.size() > maxSorts)
+                throw new UnsupportedOperationException("Can not sort files with more than 100K reads on the fly!");
+        }
+        Collections.sort(list);
+        it = list.iterator();
+    }
+
+    public boolean hasNext() { return it.hasNext(); }
+    public SAMRecord next()  { return it.next().getRecord(); }
+
+    public void remove() {
+        throw new UnsupportedOperationException("Can not remove records from a SAM file via an iterator!");
+    }
+
+    private class ComparableSAMRecord implements Comparable<ComparableSAMRecord> {
+
+        private SAMRecord record;
+
+        public ComparableSAMRecord(SAMRecord record) {
+            this.record = record;
+        }
+
+        public SAMRecord getRecord() {
+            return record;
+        }
+
+        public int compareTo(ComparableSAMRecord o) {
+            GenomeLoc myLoc = Utils.genomicLocationOf(record);
+            GenomeLoc hisLoc = Utils.genomicLocationOf(o.getRecord());
+            return myLoc.compareTo(hisLoc);
+        }
+    }
+}