Add removeProgramRecords argument

* Add unit test for the removeProgramRecords

public/java/src/org/broadinstitute/sting/gatk/GenomeAnalysisEngine.java

@@ -813,7 +813,8 @@ public class GenomeAnalysisEngine {
                 getWalkerBAQQualityMode(),
                 refReader,
                 getBaseRecalibration(),
-                argCollection.defaultBaseQualities);
+                argCollection.defaultBaseQualities,
+                argCollection.removeProgramRecords);
     }
 
     /**

public/java/src/org/broadinstitute/sting/gatk/arguments/GATKArgumentCollection.java

@@ -249,6 +249,9 @@ public class GATKArgumentCollection {
     @Argument(fullName = "validation_strictness", shortName = "S", doc = "How strict should we be with validation", required = false)
     public SAMFileReader.ValidationStringency strictnessLevel = SAMFileReader.ValidationStringency.SILENT;
 
+    @Argument(fullName = "remove_program_records", shortName = "rpr", doc = "Should we remove program records from the SAM header", required = false)
+    public boolean removeProgramRecords = false;
+
     @Argument(fullName = "unsafe", shortName = "U", doc = "If set, enables unsafe operations: nothing will be checked at runtime.  For expert users only who know what they are doing.  We do not support usage of this argument.", required = false)
     public ValidationExclusion.TYPE unsafe;
 

public/java/src/org/broadinstitute/sting/gatk/datasources/reads/SAMDataSource.java

@@ -89,6 +89,11 @@ public class SAMDataSource {
      */
     private final SAMFileReader.ValidationStringency validationStringency;
 
+    /**
+     * Do we want to remove the program records from this data source?
+     */
+    private final boolean removeProgramRecords;
+
     /**
      * Store BAM indices for each reader present.
      */
@@ -200,7 +205,8 @@ public class SAMDataSource {
                 BAQ.QualityMode.DONT_MODIFY,
                 null, // no BAQ
                 null, // no BQSR
-                (byte) -1);
+                (byte) -1,
+                false);
     }
 
     /**
@@ -233,7 +239,8 @@ public class SAMDataSource {
             BAQ.QualityMode qmode,
             IndexedFastaSequenceFile refReader,
             BaseRecalibration bqsrApplier,
-            byte defaultBaseQualities) {
+            byte defaultBaseQualities,
+            boolean removeProgramRecords) {
         this.readMetrics = new ReadMetrics();
         this.genomeLocParser = genomeLocParser;
 
@@ -249,6 +256,7 @@ public class SAMDataSource {
             dispatcher = null;
 
         validationStringency = strictness;
+        this.removeProgramRecords = removeProgramRecords;
         if(readBufferSize != null)
             ReadShard.setReadBufferSize(readBufferSize);
         else {
@@ -748,7 +756,7 @@ public class SAMDataSource {
         private synchronized void createNewResource() {
             if(allResources.size() > maxEntries)
                 throw new ReviewedStingException("Cannot create a new resource pool.  All resources are in use.");
-            SAMReaders readers = new SAMReaders(readerIDs, validationStringency);
+            SAMReaders readers = new SAMReaders(readerIDs, validationStringency, removeProgramRecords);
             allResources.add(readers);
             availableResources.add(readers);
         }
@@ -777,9 +785,11 @@ public class SAMDataSource {
         /**
          * Derive a new set of readers from the Reads metadata.
          * @param readerIDs reads to load.
+         * TODO: validationStringency is not used here
          * @param validationStringency validation stringency.
+         * @param removeProgramRecords indicate whether to clear program records from the readers
          */
-        public SAMReaders(Collection<SAMReaderID> readerIDs, SAMFileReader.ValidationStringency validationStringency) {
+        public SAMReaders(Collection<SAMReaderID> readerIDs, SAMFileReader.ValidationStringency validationStringency, boolean removeProgramRecords) {
             final int totalNumberOfFiles = readerIDs.size();
             int readerNumber = 1;
             final SimpleTimer timer = new SimpleTimer().start();
@@ -790,6 +800,9 @@ public class SAMDataSource {
             long lastTick = timer.currentTime();
             for(final SAMReaderID readerID: readerIDs) {
                 final ReaderInitializer init = new ReaderInitializer(readerID).call();
+                if (removeProgramRecords) {
+                    init.reader.getFileHeader().setProgramRecords(new ArrayList<SAMProgramRecord>());
+                }
                 if (threadAllocation.getNumIOThreads() > 0) {
                     inputStreams.put(init.readerID, init.blockInputStream); // get from initializer
                 }

public/java/test/org/broadinstitute/sting/gatk/datasources/reads/SAMDataSourceUnitTest.java

@@ -24,9 +24,12 @@
 
 package org.broadinstitute.sting.gatk.datasources.reads;
 
+import static org.testng.Assert.assertEquals;
+import static org.testng.Assert.assertTrue;
 import static org.testng.Assert.fail;
 import net.sf.picard.reference.IndexedFastaSequenceFile;
 import net.sf.samtools.SAMFileReader;
+import net.sf.samtools.SAMProgramRecord;
 import net.sf.samtools.SAMRecord;
 import org.broadinstitute.sting.BaseTest;
 import org.broadinstitute.sting.commandline.Tags;
@@ -36,6 +39,7 @@ import org.broadinstitute.sting.gatk.iterators.StingSAMIterator;
 import org.broadinstitute.sting.gatk.resourcemanagement.ThreadAllocation;
 import org.broadinstitute.sting.utils.GenomeLocParser;
 import org.broadinstitute.sting.utils.GenomeLoc;
+import org.broadinstitute.sting.utils.baq.BAQ;
 import org.broadinstitute.sting.utils.fasta.CachingIndexedFastaSequenceFile;
 import org.broadinstitute.sting.utils.exceptions.UserException;
 import org.testng.annotations.AfterMethod;
@@ -143,4 +147,73 @@ public class SAMDataSourceUnitTest extends BaseTest {
             fail("testLinearBreakIterateAll: We Should get a UserException.CouldNotReadInputFile exception");
         }
     }
+
+    /** Test that we clear program records when requested */
+    @Test
+    public void testRemoveProgramRecords() {
+        logger.warn("Executing testRemoveProgramRecords");
+
+        // setup the data
+        readers.add(new SAMReaderID(new File(b37GoodBAM),new Tags()));
+
+        // use defaults
+        SAMDataSource data = new SAMDataSource(readers,
+                new ThreadAllocation(),
+                null,
+                genomeLocParser,
+                false,
+                SAMFileReader.ValidationStringency.SILENT,
+                null,
+                null,
+                new ValidationExclusion(),
+                new ArrayList<ReadFilter>(),
+                false);
+
+        List<SAMProgramRecord> defaultProgramRecords = data.getHeader().getProgramRecords();
+        assertTrue(defaultProgramRecords.size() != 0, "testRemoveProgramRecords: No program records found when using default constructor");
+
+        boolean removeProgramRecords = false;
+        data = new SAMDataSource(readers,
+                new ThreadAllocation(),
+                null,
+                genomeLocParser,
+                false,
+                SAMFileReader.ValidationStringency.SILENT,
+                null,
+                null,
+                new ValidationExclusion(),
+                new ArrayList<ReadFilter>(),
+                false,
+                BAQ.CalculationMode.OFF,
+                BAQ.QualityMode.DONT_MODIFY,
+                null, // no BAQ
+                null, // no BQSR
+                (byte) -1,
+                removeProgramRecords);
+
+        List<SAMProgramRecord> dontRemoveProgramRecords = data.getHeader().getProgramRecords();
+        assertEquals(dontRemoveProgramRecords, defaultProgramRecords, "testRemoveProgramRecords: default program records differ from removeProgramRecords = false");
+
+        removeProgramRecords = true;
+        data = new SAMDataSource(readers,
+                new ThreadAllocation(),
+                null,
+                genomeLocParser,
+                false,
+                SAMFileReader.ValidationStringency.SILENT,
+                null,
+                null,
+                new ValidationExclusion(),
+                new ArrayList<ReadFilter>(),
+                false,
+                BAQ.CalculationMode.OFF,
+                BAQ.QualityMode.DONT_MODIFY,
+                null, // no BAQ
+                null, // no BQSR
+                (byte) -1,
+                removeProgramRecords);
+
+        List<SAMProgramRecord> doRemoveProgramRecords = data.getHeader().getProgramRecords();
+        assertTrue(doRemoveProgramRecords.isEmpty(), "testRemoveProgramRecords: program records not cleared when removeProgramRecords = true");
+    }
 }