Moving the --sites_only argument up into the VCFWriter itself so that any walkers that write VCFs can choose not to emit genotypes

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5088 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/io/storage/VCFWriterStorage.java

@@ -1,6 +1,5 @@
 package org.broadinstitute.sting.gatk.io.storage;
 
-import org.broad.tribble.readers.LineReader;
 import org.broad.tribble.source.BasicFeatureSource;
 import org.broad.tribble.vcf.*;
 import org.broad.tribble.util.variantcontext.VariantContext;
@@ -11,7 +10,6 @@ import java.io.*;
 
 import net.sf.samtools.util.BlockCompressedOutputStream;
 import org.broadinstitute.sting.utils.exceptions.UserException;
-import org.broadinstitute.sting.utils.text.XReadLines;
 
 /**
  * Provides temporary and permanent storage for genotypes in VCF format.
@@ -37,7 +35,7 @@ public class VCFWriterStorage implements Storage<VCFWriterStorage>, VCFWriter {
         else if ( stub.getOutputStream() != null ) {
             this.file = null;
             this.stream = stub.getOutputStream();
-            writer = new StandardVCFWriter(stream);
+            writer = new StandardVCFWriter(stream, stub.doNotWriteGenotypes());
         }
         else
             throw new ReviewedStingException("Unable to create target to which to write; storage was provided with neither a file nor a stream.");
@@ -62,7 +60,7 @@ public class VCFWriterStorage implements Storage<VCFWriterStorage>, VCFWriter {
         }
 
         // The GATK/Tribble can't currently index block-compressed files on the fly.  Disable OTF indexing even if the user explicitly asked for it.
-        return new StandardVCFWriter(file, this.stream, indexOnTheFly && !stub.isCompressed());
+        return new StandardVCFWriter(file, this.stream, indexOnTheFly && !stub.isCompressed(), stub.doNotWriteGenotypes());
     }
 
 

java/src/org/broadinstitute/sting/gatk/io/stubs/VCFWriterArgumentTypeDescriptor.java

@@ -25,17 +25,13 @@
 
 package org.broadinstitute.sting.gatk.io.stubs;
 
-import org.broad.tribble.vcf.VCFHeaderLine;
 import org.broad.tribble.vcf.VCFWriter;
 import org.broadinstitute.sting.commandline.*;
-import org.broadinstitute.sting.gatk.CommandLineExecutable;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
-import org.broadinstitute.sting.utils.classloader.JVMUtils;
 import org.broadinstitute.sting.utils.exceptions.ReviewedStingException;
 
 import java.io.File;
 import java.io.OutputStream;
-import java.lang.annotation.Annotation;
 import java.util.*;
 
 /**
@@ -46,6 +42,7 @@ import java.util.*;
  */
 public class VCFWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor {
     private static final String NO_HEADER_ARG_NAME = "NO_HEADER";
+    private static final String SITES_ONLY_ARG_NAME = "sites_only";
     private static final HashSet<String> SUPPORTED_ZIPPED_SUFFIXES = new HashSet<String>();
 
     //
@@ -95,7 +92,7 @@ public class VCFWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor {
 
     @Override
     public List<ArgumentDefinition> createArgumentDefinitions( ArgumentSource source ) {
-        return Arrays.asList( createDefaultArgumentDefinition(source),createNoHeaderArgumentDefinition());
+        return Arrays.asList( createDefaultArgumentDefinition(source),createNoHeaderArgumentDefinition(),createSitesOnlyArgumentDefinition());
     }
 
     /**
@@ -111,7 +108,7 @@ public class VCFWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor {
     public Object createTypeDefault(ParsingEngine parsingEngine,ArgumentSource source,Class<?> type) {
         if(!source.isRequired())
             throw new ReviewedStingException("BUG: tried to create type default for argument type descriptor that can't support a type default.");        
-        VCFWriterStub stub = new VCFWriterStub(engine, defaultOutputStream, false, argumentSources, false);
+        VCFWriterStub stub = new VCFWriterStub(engine, defaultOutputStream, false, argumentSources, false, false);
         engine.addOutput(stub);
         return stub;
     }
@@ -139,10 +136,11 @@ public class VCFWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor {
         boolean compress = writerFileName != null && SUPPORTED_ZIPPED_SUFFIXES.contains(getFileSuffix(writerFileName));
 
         boolean skipWritingHeader = argumentIsPresent(createNoHeaderArgumentDefinition(),matches);
+        boolean doNotWriteGenotypes = argumentIsPresent(createSitesOnlyArgumentDefinition(),matches);
 
         // Create a stub for the given object.
-        VCFWriterStub stub = (writerFile != null) ? new VCFWriterStub(engine, writerFile, compress, argumentSources, skipWritingHeader)
+        VCFWriterStub stub = (writerFile != null) ? new VCFWriterStub(engine, writerFile, compress, argumentSources, skipWritingHeader, doNotWriteGenotypes)
-                                                  : new VCFWriterStub(engine, defaultOutputStream, compress, argumentSources, skipWritingHeader);
+                                                  : new VCFWriterStub(engine, defaultOutputStream, compress, argumentSources, skipWritingHeader, doNotWriteGenotypes);
 
         // WARNING: Side effects required by engine!
         parsingEngine.addTags(stub,getArgumentTags(matches));
@@ -171,6 +169,26 @@ public class VCFWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor {
                                        null );
     }
 
+    /**
+     * Creates the optional compression level argument for the BAM file.
+     * @return Argument definition for the BAM file itself.  Will not be null.
+     */
+    private ArgumentDefinition createSitesOnlyArgumentDefinition() {
+        return new ArgumentDefinition( ArgumentIOType.ARGUMENT,
+                                       boolean.class,
+                                       SITES_ONLY_ARG_NAME,
+                                       SITES_ONLY_ARG_NAME,
+                                       "Just output sites without genotypes (i.e. only the first 8 columns of the VCF)",
+                                       false,
+                                       true,
+                                       false,
+                                       true,
+                                       null,
+                                       null,
+                                       null,
+                                       null );
+    }
+
 
     /**
      * Returns a lower-cased version of the suffix of the provided file.

java/src/org/broadinstitute/sting/gatk/io/stubs/VCFWriterStub.java

@@ -28,15 +28,12 @@ package org.broadinstitute.sting.gatk.io.stubs;
 import java.io.File;
 import java.io.PrintStream;
 import java.io.OutputStream;
-import java.util.ArrayList;
 import java.util.Collection;
-import java.util.TreeSet;
 
 import org.broad.tribble.util.variantcontext.VariantContext;
 import org.broad.tribble.vcf.VCFHeader;
 import org.broad.tribble.vcf.VCFHeaderLine;
 import org.broad.tribble.vcf.VCFWriter;
-import org.broadinstitute.sting.commandline.CommandLineUtils;
 import org.broadinstitute.sting.gatk.CommandLineExecutable;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
 import org.broadinstitute.sting.gatk.io.OutputTracker;
@@ -87,6 +84,11 @@ public class VCFWriterStub implements Stub<VCFWriter>, VCFWriter {
      */
     private final boolean skipWritingHeader;
 
+    /**
+     * Should we not write genotypes even when provided?
+     */
+    private final boolean doNotWriteGenotypes;
+
     /**
      * Connects this stub with an external stream capable of serving the
      * requests of the consumer of this stub.
@@ -98,13 +100,14 @@ public class VCFWriterStub implements Stub<VCFWriter>, VCFWriter {
      * @param genotypeFile  file to (ultimately) create.
      * @param isCompressed  should we compress the output stream?
      */
-    public VCFWriterStub(GenomeAnalysisEngine engine, File genotypeFile, boolean isCompressed, Collection<Object> argumentSources, boolean skipWritingHeader) {
+    public VCFWriterStub(GenomeAnalysisEngine engine, File genotypeFile, boolean isCompressed, Collection<Object> argumentSources, boolean skipWritingHeader, boolean doNotWriteGenotypes) {
         this.engine = engine;
         this.genotypeFile = genotypeFile;
         this.genotypeStream = null;
         this.isCompressed = isCompressed;
         this.argumentSources = argumentSources;
         this.skipWritingHeader = skipWritingHeader;
+        this.doNotWriteGenotypes = doNotWriteGenotypes;
     }
 
     /**
@@ -112,13 +115,14 @@ public class VCFWriterStub implements Stub<VCFWriter>, VCFWriter {
      * @param genotypeStream  stream to (ultimately) write.
      * @param isCompressed  should we compress the output stream?
      */
-    public VCFWriterStub(GenomeAnalysisEngine engine, OutputStream genotypeStream, boolean isCompressed, Collection<Object> argumentSources, boolean skipWritingHeader) {
+    public VCFWriterStub(GenomeAnalysisEngine engine, OutputStream genotypeStream, boolean isCompressed, Collection<Object> argumentSources, boolean skipWritingHeader, boolean doNotWriteGenotypes) {
         this.engine = engine;
         this.genotypeFile = null;
         this.genotypeStream = new PrintStream(genotypeStream);
         this.isCompressed = isCompressed;
         this.argumentSources = argumentSources;
         this.skipWritingHeader = skipWritingHeader;
+        this.doNotWriteGenotypes = doNotWriteGenotypes;
     }
 
     /**
@@ -145,6 +149,14 @@ public class VCFWriterStub implements Stub<VCFWriter>, VCFWriter {
         return isCompressed;
     }
 
+    /**
+     * Should we tell the VCF writer not to write genotypes?
+     * @return true if the writer should not write genotypes.
+     */
+    public boolean doNotWriteGenotypes() {
+        return doNotWriteGenotypes;
+    }
+
     /**
      * Retrieves the header to use when creating the new file.
      * @return header to use when creating the new file.

java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedArgumentCollection.java

@@ -44,10 +44,6 @@ public class UnifiedArgumentCollection {
     @Argument(fullName = "pcr_error_rate", shortName = "pcr_error", doc = "The PCR error rate to be used for computing fragment-based likelihoods", required = false)
     public Double PCR_error = DiploidSNPGenotypeLikelihoods.DEFAULT_PCR_ERROR_RATE;
 
-    // control the output
-    @Argument(fullName = "sites_only", shortName = "sites_only", doc = "Should we output just sites without genotypes (i.e. only the first 8 columns of the VCF)?", required = false)
-    public boolean SITES_ONLY = false;
-
     @Argument(fullName = "genotype", shortName = "genotype", doc = "Should we output confident genotypes (i.e. including ref calls) or just the variants?", required = false)
     public boolean GENOTYPE_MODE = false;
 
@@ -109,7 +105,6 @@ public class UnifiedArgumentCollection {
         uac.GLmodel = GLmodel;
         uac.heterozygosity = heterozygosity;
         uac.PCR_error = PCR_error;
-        uac.SITES_ONLY = SITES_ONLY;
         uac.GENOTYPE_MODE = GENOTYPE_MODE;
         uac.ALL_BASES_MODE = ALL_BASES_MODE;
         uac.NO_SLOD = NO_SLOD;

java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java

@@ -129,9 +129,6 @@ public class UnifiedGenotyper extends LocusWalker<VariantCallContext, UnifiedGen
         annotationEngine = new VariantAnnotatorEngine(getToolkit(), Arrays.asList(annotationClassesToUse), annotationsToUse);
         UG_engine = new UnifiedGenotyperEngine(getToolkit(), UAC, logger, verboseWriter, annotationEngine, samples);
 
-        if ( UAC.SITES_ONLY )
-            samples.clear();
-
         // initialize the header
         writer.writeHeader(new VCFHeader(getHeaderInfo(), samples)) ;
     }

java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyperEngine.java

@@ -400,9 +400,6 @@ public class UnifiedGenotyperEngine {
             vcCall = variantContexts.iterator().next(); // we know the collection will always have exactly 1 element.
         }
 
-        if ( UAC.SITES_ONLY )
-            vcCall = VariantContext.modifyGenotypes(vcCall, null);
-
         VariantCallContext call = new VariantCallContext(vcCall, passesCallThreshold(phredScaledConfidence, atTriggerTrack));
         call.setRefBase(refContext.getBase());
         return call;