a fix for a bug Eric found: if your first call contains fewer samples than calls at other loci, your VCFHeader got setup incorrectly.

Also moved a buch of Lists over to Sets for consistancy.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1859 348d0f76-0448-11de-a6fe-93d51630548a

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

@@ -98,10 +98,12 @@ public class UnifiedGenotyper extends LocusWalker<Pair<List<GenotypeCall>, Genot
         if ( VARIANTS_FILE != null )
             writer = GenotypeWriterFactory.create(VAR_FORMAT, GenomeAnalysisEngine.instance.getSAMFileHeader(), VARIANTS_FILE,
                                                   "UnifiedGenotyper",
-                                                  this.getToolkit().getArguments().referenceFile.getName());
+                                                  this.getToolkit().getArguments().referenceFile.getName(),
+                                                  samples);
         else
             writer = GenotypeWriterFactory.create(VAR_FORMAT, GenomeAnalysisEngine.instance.getSAMFileHeader(), out, "UnifiedGenotyper",
-                                                  this.getToolkit().getArguments().referenceFile.getName());
+                                                  this.getToolkit().getArguments().referenceFile.getName(),
+                                                  samples);
         callsMetrics = new CallMetrics();
     }
 

java/src/org/broadinstitute/sting/playground/gatk/walkers/variantstovcf/VariantsToVCF.java

@@ -49,7 +49,7 @@ public class VariantsToVCF extends RefWalker<Integer, Integer> {
 
     public static VCFHeader getHeader(GATKArgumentCollection args, Set<String> sampleNames) {
         Map<String, String> metaData = new HashMap<String, String>();
-        List<String> additionalColumns = new ArrayList<String>();
+        Set<String> additionalColumns = new HashSet<String>();
 
         // Don't output the data for now because it kills our unit test MD5s and is optional
         // TODO - figure out what to do here

java/src/org/broadinstitute/sting/utils/genotype/GenotypeWriterFactory.java

@@ -9,6 +9,8 @@ import org.broadinstitute.sting.utils.genotype.vcf.VCFGenotypeWriterAdapter;
 
 import java.io.File;
 import java.io.PrintStream;
+import java.util.List;
+import java.util.Set;
 
 
 /**
@@ -31,7 +33,12 @@ public class GenotypeWriterFactory {
      * @param destination the destination file
      * @return the genotype writer object
      */
-    public static GenotypeWriter create(GENOTYPE_FORMAT format, SAMFileHeader header, File destination, String source, String referenceName ) {
+    public static GenotypeWriter create(GENOTYPE_FORMAT format,
+                                        SAMFileHeader header,
+                                        File destination,
+                                        String source,
+                                        String referenceName,
+                                        Set<String> sampleNames ) {
         switch (format) {
             case GLF:
                 return new GLFWriter(header.toString(), destination);
@@ -40,20 +47,25 @@ public class GenotypeWriterFactory {
             case GELI_BINARY:
                 return new GeliAdapter(destination, header);
             case VCF:
-                return new VCFGenotypeWriterAdapter(source, referenceName, destination);
+                return new VCFGenotypeWriterAdapter(source, referenceName, destination, sampleNames);
             default:
                 throw new StingException("Genotype writer " + format.toString() + " is not implemented");
         }
     }
 
-    public static GenotypeWriter create(GENOTYPE_FORMAT format, SAMFileHeader header, PrintStream destination, String source, String referenceName ) {
+    public static GenotypeWriter create(GENOTYPE_FORMAT format,
+                                        SAMFileHeader header,
+                                        PrintStream destination,
+                                        String source,
+                                        String referenceName,
+                                        Set<String> sampleNames ) {
         switch (format) {
             case GELI:
                 return new GeliTextWriter(destination);
             case GLF:
                 return new GLFWriter(header.toString(), destination);
             case VCF:
-                return new VCFGenotypeWriterAdapter(source, referenceName, destination);
+                return new VCFGenotypeWriterAdapter(source, referenceName, destination, sampleNames);
             default:
                 throw new StingException("Genotype writer to " + format.toString() + " to standard output is not implemented");
         }

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFGenotypeWriterAdapter.java

@@ -21,23 +21,26 @@ public class VCFGenotypeWriterAdapter implements GenotypeWriter {
     private String mSource;
     private String mReferenceName;
     private boolean mInitialized = false;
+    private final Set<String> mSampleNames = new HashSet<String>();
     private final File mFile;
     private final OutputStream mStream;
 
-    public VCFGenotypeWriterAdapter(String source, String referenceName, File writeTo) {
+    public VCFGenotypeWriterAdapter(String source, String referenceName, File writeTo, Set<String> sampleNames) {
         mReferenceName = referenceName;
         mSource = source;
         mFile = writeTo;
         if (mFile == null) throw new RuntimeException("VCF output file must not be null");
         mStream = null;
+        mSampleNames.addAll(sampleNames);
     }
 
-    public VCFGenotypeWriterAdapter(String source, String referenceName, OutputStream writeTo) {
+    public VCFGenotypeWriterAdapter(String source, String referenceName, OutputStream writeTo, Set<String> sampleNames) {
         mReferenceName = referenceName;
         mSource = source;
         mFile = null;
         mStream = writeTo;
         if (mStream == null) throw new RuntimeException("VCF output stream must not be null");
+        mSampleNames.addAll(sampleNames);
 
     }
 
@@ -49,7 +52,6 @@ public class VCFGenotypeWriterAdapter implements GenotypeWriter {
      */
     private void lazyInitialize(List<Genotype> genotypes, File file, OutputStream stream) {
         Map<String, String> hInfo = new HashMap<String, String>();
-        List<String> sampleNames = getSampleNames(genotypes);
 
         // setup the header fields
         hInfo.put("format", "VCRv3.2");
@@ -57,7 +59,7 @@ public class VCFGenotypeWriterAdapter implements GenotypeWriter {
         hInfo.put("reference", mReferenceName);
 
         // setup the sample names
-        mHeader = new VCFHeader(hInfo, sampleNames);
+        mHeader = new VCFHeader(hInfo, mSampleNames);
         if (mFile == null)
             mWriter = new VCFWriter(mHeader, stream);
         else

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFHeader.java

@@ -23,7 +23,7 @@ public class VCFHeader {
     private final Map<String, String> mMetaData = new HashMap<String, String>();
 
     // the list of auxillary tags
-    private final List<String> mGenotypeSampleNames = new ArrayList<String>();
+    private final Set<String> mGenotypeSampleNames = new HashSet<String>();
 
     // the character string that indicates meta data
     public static final String METADATA_INDICATOR = "##";
@@ -56,7 +56,7 @@ public class VCFHeader {
      * @param metaData            the meta data associated with this header
      * @param genotypeSampleNames the genotype format field, and the sample names
      */
-    public VCFHeader(Map<String, String> metaData, List<String> genotypeSampleNames) {
+    public VCFHeader(Map<String, String> metaData, Set<String> genotypeSampleNames) {
         for (String key : metaData.keySet()) mMetaData.put(key, metaData.get(key));
         for (String col : genotypeSampleNames) {
             if (!col.equals("FORMAT"))
@@ -107,7 +107,7 @@ public class VCFHeader {
      *
      * @return a list of the genotype column names, which may be empty if hasGenotypingData() returns false
      */
-    public List<String> getGenotypeSamples() {
+    public Set<String> getGenotypeSamples() {
         return mGenotypeSampleNames;
     }
 

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFReader.java

@@ -126,7 +126,7 @@ public class VCFReader implements Iterator<VCFRecord>, Iterable<VCFRecord> {
     protected VCFHeader createHeader(List<String> headerStrings) {
 
         Map<String, String> metaData = new HashMap<String, String>();
-        List<String> auxTags = new ArrayList<String>();
+        Set<String> auxTags = new HashSet<String>();
         // iterate over all the passed in strings
         for (String str : headerStrings) {
             Matcher matcher = pMeta.matcher(str);

java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFHeaderTest.java

@@ -19,7 +19,7 @@ public class VCFHeaderTest extends BaseTest {
 
     private Set<VCFHeader.HEADER_FIELDS> headerFields = new LinkedHashSet<VCFHeader.HEADER_FIELDS>();
     private Map<String, String> metaData = new HashMap();
-    private List<String> additionalColumns = new ArrayList<String>();
+    private Set<String> additionalColumns = new HashSet<String>();
 
     /**
      * give it fake data, and make sure we get back the right fake data

java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFRecordTest.java

@@ -4,10 +4,7 @@ import org.broadinstitute.sting.BaseTest;
 import org.junit.Assert;
 import org.junit.Test;
 
-import java.util.ArrayList;
+import java.util.*;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
 
 
 /**
@@ -126,7 +123,7 @@ public class VCFRecordTest extends BaseTest {
      */
     public static VCFHeader createFakeHeader() {
         Map<String, String> metaData = new HashMap();
-        List<String> additionalColumns = new ArrayList<String>();
+        Set<String> additionalColumns = new HashSet<String>();
         metaData.put("format", "VCRv3.2"); // required
         metaData.put("two", "2");
         additionalColumns.add("FORMAT");

java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFWriterTest.java

@@ -18,7 +18,7 @@ import java.util.*;
 public class VCFWriterTest extends BaseTest {
     private Set<VCFHeader.HEADER_FIELDS> headerFields = new LinkedHashSet<VCFHeader.HEADER_FIELDS>();
     private Map<String, String> metaData = new HashMap();
-    private List<String> additionalColumns = new ArrayList<String>();
+    private Set<String> additionalColumns = new HashSet<String>();
     private File fakeVCFFile = new File("FAKEVCFFILEFORTESTING.vcf");
 
     /** test, using the writer and reader, that we can output and input a VCF file without problems */
@@ -45,7 +45,7 @@ public class VCFWriterTest extends BaseTest {
      * create a fake header of known quantity
      * @return a fake VCF header
      */
-    public static VCFHeader createFakeHeader(Map<String, String> metaData, List<String> additionalColumns) {
+    public static VCFHeader createFakeHeader(Map<String, String> metaData, Set<String> additionalColumns) {
         metaData.put("format", "VCRv3.2"); // required
         metaData.put("two", "2");
         additionalColumns.add("FORMAT");