Final performance optimization for GenotypesContext

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/util/VariantEvalUtils.java

@@ -267,7 +267,7 @@ public class VariantEvalUtils {
      * @return a new VariantContext with just the requested sample
      */
     public VariantContext getSubsetOfVariantContext(VariantContext vc, String sampleName) {
-        return getSubsetOfVariantContext(vc, new HashSet<String>(Arrays.asList(sampleName)));
+        return getSubsetOfVariantContext(vc, Collections.singleton(sampleName));
     }
 
     /**

public/java/src/org/broadinstitute/sting/utils/variantcontext/GenotypesContext.java

@@ -646,28 +646,18 @@ public class GenotypesContext implements List<Genotype> {
      */
     @Requires("samples != null")
     @Ensures("result != null")
-    public GenotypesContext subsetToSamples( final Collection<String> samples ) {
-        return subsetToSamples(new HashSet<String>(samples));
-    }
-
-    /**
-     * {@link #subsetToSamples(java.util.Collection)}
-     * @param samples
-     * @return
-     */
-    @Requires("samples != null")
-    @Ensures("result != null")
     public GenotypesContext subsetToSamples( final Set<String> samples ) {
-        if ( samples.size() == size() )
+        final int nSamples = samples.size();
+        final int nGenotypes = size();
+
+        if ( nSamples == nGenotypes )
             return this;
-        else if ( samples.isEmpty() )
+        else if ( nSamples == 0 )
             return NO_GENOTYPES;
-        else {
+        else { // nGenotypes < nSamples
-            GenotypesContext subset = create(samples.size());
+            final GenotypesContext subset = create(samples.size());
-            for ( final Genotype g : getGenotypes() ) {
+            for ( final String sample : samples ) {
-                if ( samples.contains(g.getSampleName()) ) {
+                subset.add(get(sample));
-                    subset.add(g);
-                }
             }
             return subset;
         }

public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContext.java

@@ -726,19 +726,21 @@ public class VariantContext implements Feature { // to enable tribble intergrati
      * @throws IllegalArgumentException if sampleName isn't bound to a genotype
      */
     public GenotypesContext getGenotypes(String sampleName) {
-        return getGenotypes(Arrays.asList(sampleName));
+        return getGenotypes(Collections.singleton(sampleName));
     }
 
     /**
      * Returns a map from sampleName -> Genotype for each sampleName in sampleNames.  Returns a map
      * for consistency with the multi-get function.
      *
+     * For testing convenience only
+     *
      * @param sampleNames a unique list of sample names
      * @return
      * @throws IllegalArgumentException if sampleName isn't bound to a genotype
      */
-    public GenotypesContext getGenotypes(Collection<String> sampleNames) {
+    protected GenotypesContext getGenotypes(Collection<String> sampleNames) {
-        return getGenotypes().subsetToSamples(sampleNames);
+        return getGenotypes().subsetToSamples(new HashSet<String>(sampleNames));
     }
 
     public GenotypesContext getGenotypes(Set<String> sampleNames) {

public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContextUtils.java

@@ -86,6 +86,7 @@ public class VariantContextUtils {
                 for ( Allele allele : vc.getAlternateAlleles() ) {
                     int altChromosomes = vc.getCalledChrCount(allele);
                     alleleCounts.add(altChromosomes);
+                    // todo -- this is a performance problem
                     String freq = String.format(makePrecisionFormatStringFromDenominatorValue(totalChromosomes), ((double)altChromosomes / totalChromosomes));
                     alleleFreqs.add(freq);
                 }