V5 improvements to VariantContext. Now fully supports genotypes. Filtering enabled. Significant tests throughout system. Support for rebuilding variant contexts from subsets of genotypes. Some code cleanup around repository

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2721 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/oneoffprojects/variantcontext/Genotype.java

@@ -13,6 +13,11 @@ public class Genotype extends AttributedObject {
     private List<Allele> alleles = new ArrayList<Allele>();
     private String sampleName = null;
 
+    // todo -- do genotypes need to have locations?  Or is it sufficient to have an
+    // todo -- associated VC with a location?  One nasty implication is that people will have to
+    // todo -- pass around both a Variant Context and genotypes.  Although users can always just package up
+    // the associated genotypes into the VC itself.
+
     public Genotype(VariantContext vc, List<String> alleles, String sampleName, double negLog10PError) {
         this(resolveAlleles(vc, alleles), sampleName, negLog10PError);
     }

java/src/org/broadinstitute/sting/oneoffprojects/variantcontext/TestVariantContextWalker.java

@@ -23,7 +23,7 @@ public class TestVariantContextWalker extends RodWalker<Integer, Integer> {
                 for (ReferenceOrderedDatum d : dbsnpList) {
                     rodDbSNP dbsnpRecord = (rodDbSNP)d;
                     if ( dbsnpRecord.getLocation().getStart() == context.getLocation().getStart() ) {
-                        VariantContext vc = VariantContextAdaptors.dbsnp2VariantContext(dbsnpRecord);
+                        VariantContext vc = VariantContextAdaptors.dbsnpToVariantContext(dbsnpRecord);
                         if ( vc != null ) {
                             n++;
                             System.out.printf("%s%n", vc);
@@ -40,7 +40,7 @@ public class TestVariantContextWalker extends RodWalker<Integer, Integer> {
                 int n = 0;
                 for (ReferenceOrderedDatum d : vcfList) {
                     RodVCF vcfRecord = (RodVCF)d;
-                    VariantContext vc = VariantContextAdaptors.vcf2VariantContext(vcfRecord);
+                    VariantContext vc = VariantContextAdaptors.vcfToVariantContext(vcfRecord);
                     if ( vc != null ) {
                         n++;
                         System.out.printf("%s%n", vc);

java/src/org/broadinstitute/sting/oneoffprojects/variantcontext/VariantContext.java

@@ -2,6 +2,7 @@ package org.broadinstitute.sting.oneoffprojects.variantcontext;
 
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.StingException;
+import org.broadinstitute.sting.utils.BaseUtils;
 
 import java.util.*;
 
@@ -22,7 +23,8 @@ public class VariantContext extends AttributedObject {
 
     Type type = Type.UNDETERMINED;
 
-    // todo -- add FILTER
+    Set<Object> filters = new HashSet<Object>();
+
 
     // ---------------------------------------------------------------------------------------------------------
     //
@@ -31,20 +33,17 @@ public class VariantContext extends AttributedObject {
     // ---------------------------------------------------------------------------------------------------------
 
     public VariantContext(GenomeLoc loc) {
-        if ( loc == null ) { throw new StingException("GenomeLoc cannot be null"); }
+        super();
 
+        if ( loc == null ) { throw new StingException("GenomeLoc cannot be null"); }
         this.loc = loc;
     }
 
-    protected VariantContext(VariantContext parent, Set<Genotype> genotypes, HashMap<Object, Object> attributes) {
+    public VariantContext(GenomeLoc loc, Collection<Allele> alleles) {
-        this(parent.getLocation(), parent.getAlleles(), genotypes, attributes);
+        this(loc, alleles, (Map<Object, Object>)null);
     }
 
-    // todo -- add more convenience methods
+    public VariantContext(GenomeLoc loc, Collection<Allele> alleles, Map<Object, Object> attributes) {
-    public VariantContext(GenomeLoc loc, Set<Allele> alleles)   { this(loc, alleles, null); }
-    public VariantContext(GenomeLoc loc, List<Allele> alleles ) { this(loc, alleles, null); }
-
-    public VariantContext(GenomeLoc loc, List<Allele> alleles, Map<Object, Object> attributes) {
         this(loc);
 
         HashSet<Allele> alleleSet = new HashSet<Allele>();
@@ -59,30 +58,114 @@ public class VariantContext extends AttributedObject {
         validate();
     }
 
-    public VariantContext(GenomeLoc loc, Set<Allele> alleles, Map<Object, Object> attributes) {
+    public VariantContext(GenomeLoc loc, Collection<Allele> alleles, Collection<Genotype> genotypes) {
         this(loc);
         setAlleles(alleles);
-        setAttributes(attributes);
+        setGenotypes(genotypes);
         validate();
     }
 
-    public VariantContext(GenomeLoc loc, Set<Allele> alleles, Set<Genotype> genotypes, Map<Object, Object> attributes) {
+    public VariantContext(GenomeLoc loc, Collection<Allele> alleles,
+                          Collection<Genotype> genotypes, Map<Object, Object> attributes,
+                          double negLog10PError, Collection<Object> filters) {
         this(loc);
         setAlleles(alleles);
         setGenotypes(genotypes);
         setAttributes(attributes);
+        setNegLog10PError(negLog10PError);
+        setFilters(filters);
         validate();
     }
 
-
+    public VariantContext(GenomeLoc loc, Collection<Allele> alleles,
-    public VariantContext(GenomeLoc loc, Set<Allele> alleles, Map<String, Genotype> genotypes, Map<Object, Object> attributes) {
+                          Map<String, Genotype> genotypes, Map<Object, Object> attributes,
+                          double negLog10PError, Collection<Object> filters) {
         this(loc);
         setAlleles(alleles);
         setGenotypes(genotypes);
         setAttributes(attributes);
+        setNegLog10PError(negLog10PError);
+        setFilters(filters);
         validate();
     }
 
+    // todo -- add clone method
+
+    // ---------------------------------------------------------------------------------------------------------
+    //
+    // Selectors
+    //
+    // ---------------------------------------------------------------------------------------------------------
+
+    public VariantContext subContextFromGenotypes(Genotype genotype) {
+        return subContextFromGenotypes(Arrays.asList(genotype));
+    }
+
+    public VariantContext subContextFromGenotypes(Collection<Genotype> genotypes) {
+        // todo -- we should check for uniqueness of genotypes
+        return subContextFromGenotypes(new HashSet<Genotype>(genotypes), getAttributes());
+    }
+
+    public VariantContext subContextFromGenotypes(Collection<Genotype> genotypes, Map<Object, Object> attributes) {
+        return new VariantContext(getLocation(), allelesOfGenotypes(genotypes), genotypes, attributes, getNegLog10PError(), getFilters());
+    }
+
+    /** helper routnine for subcontext */
+    private Set<Allele> allelesOfGenotypes(Collection<Genotype> genotypes) {
+        Set<Allele> alleles = new HashSet<Allele>();
+
+        boolean addedref = false;
+        for ( Genotype g : genotypes ) {
+            for ( Allele a : g.getAlleles() ) {
+                addedref = addedref || a.isReference();
+                if ( a.isCalled() )
+                    alleles.add(a);
+            }
+        }
+        if ( ! addedref ) alleles.add(getReference());
+
+        return alleles;
+    }
+
+    // ---------------------------------------------------------------------------------------------------------
+    //
+    // Filter
+    //
+    // ---------------------------------------------------------------------------------------------------------
+
+    public Set<Object> getFilters() {
+        return filters;
+    }
+
+    public boolean isFiltered() {
+        return filters.size() > 0;
+    }
+
+    public boolean isNotFiltered() {
+        return ! isFiltered();
+    }
+
+    public void addFilter(Object filter) {
+        if ( filter == null ) throw new IllegalArgumentException("BUG: Attempting to add null filter " + this);
+        if ( getFilters().contains(filter) ) throw new IllegalArgumentException("BUG: Attempting to add duplicate filter " + filter + " at " + this);
+        filters.add(filter);
+    }
+
+    public void addFilters(Collection<? extends Object> filters) {
+        if ( filters == null ) throw new IllegalArgumentException("BUG: Attempting to add null filters at" + this);
+        for ( Object f : filters )
+            addFilter(f);
+    }
+
+    public void clearFilters() {
+        filters.clear();
+    }
+
+    public void setFilters(Collection<? extends Object> filters) {
+        clearFilters();
+        addFilters(filters);
+    }
+
     // ---------------------------------------------------------------------------------------------------------
     //
     // type operations
@@ -142,7 +225,6 @@ public class VariantContext extends AttributedObject {
      *   Assigned to variations that are multi-base variations of a single, common length
      *   GGA/AGT
      */
-
     public enum Type {
         NO_VARIATION,
         SNP,
@@ -172,6 +254,14 @@ public class VariantContext extends AttributedObject {
      */
     public boolean isSNP() { return getType() == Type.SNP; }
 
+    public BaseUtils.BaseSubstitutionType getSNPSubstitutionType() {
+        if ( ! isSNP() || ! isBiallelic() ) throw new IllegalStateException("Requested SNP substitution type for bialleic non-SNP " + this);
+        return BaseUtils.SNPSubstitutionType(getReference().getBases()[0], getAlternateAllele(0).getBases()[0]);
+    }
+
+    public boolean isTransition()       { return getSNPSubstitutionType() == BaseUtils.BaseSubstitutionType.TRANSITION; } 
+    public boolean isTransversion()     { return getSNPSubstitutionType() == BaseUtils.BaseSubstitutionType.TRANSVERSION; }
+
     /**
      * convenience method for variants
      *
@@ -207,6 +297,7 @@ public class VariantContext extends AttributedObject {
      */
     public boolean isMixed() { return getType() == Type.MIXED; }    
 
+
     // ---------------------------------------------------------------------------------------------------------
     //
     // Generic accessors
@@ -276,26 +367,6 @@ public class VariantContext extends AttributedObject {
         return false;
     }
 
-//    public int getMaxAlleleLength() {
-//        int max = 0;
-//
-//        for ( Allele a : getAlleles() ) {
-//            max = a.length() > max ? a.length() : max;
-//        }
-//
-//        return max;
-//    }
-//
-//    public int getMinAlleleLength() {
-//        int min = Integer.MAX_VALUE;
-//
-//        for ( Allele a : getAlleles() ) {
-//            min = a.length() < min ? a.length() : min;
-//        }
-//
-//        return min;
-//    }
-
     /**
      * Gets the alleles.  This method should return all of the alleles present at the location,
      * including the reference allele.  There are no constraints imposed on the ordering of alleles
@@ -334,7 +405,7 @@ public class VariantContext extends AttributedObject {
     }
 
 
-    public void setAlleles(Set<Allele> alleles) {
+    public void setAlleles(Collection<Allele> alleles) {
         this.alleles.clear();
         for ( Allele a : alleles )
             addAllele(a);
@@ -368,11 +439,28 @@ public class VariantContext extends AttributedObject {
      */
     public boolean hasGenotypes() { return genotypes.size() > 0; }
 
+    public boolean hasSingleSample() { return genotypes.size() == 1; }
+
     /**
      * @return set of all Genotypes associated with this context
      */
     public Map<String, Genotype> getGenotypes() { return genotypes; }
 
+    public Map<String, Genotype> getGenotypes(String sampleName) {
+        return getGenotypes(Arrays.asList(sampleName));
+    }
+
+    public Map<String, Genotype> getGenotypes(Collection<String> sampleNames) {
+        HashMap<String, Genotype> map = new HashMap<String, Genotype>();
+
+        for ( String name : sampleNames ) {
+            if ( map.containsKey(name) ) throw new IllegalArgumentException("Duplicate names detected in requested samples " + sampleNames);
+            map.put(name, getGenotype(name));
+        }
+
+        return map;
+    }
+
     /**
      * @return the set of all sample names in this context
      */
@@ -553,12 +641,11 @@ public class VariantContext extends AttributedObject {
         if ( ! alreadySeenRef )
             throw new IllegalArgumentException("No reference allele found in VariantContext");
 
-        if ( getType() == Type.INDEL ) {
+//        if ( getType() == Type.INDEL ) {
 //            if ( getReference().length() != (getLocation().size()-1) ) {
-            if ( (getReference().isNull() && getLocation().size() != 1 ) ||
+        if ( (getReference().isNull() && getLocation().size() != 1 ) ||
-                    (getReference().isNonNull() && getReference().length() != getLocation().size()) ) {
+                (getReference().isNonNull() && getReference().length() != getLocation().size()) ) {
-                throw new IllegalStateException("BUG: GenomeLoc " + getLocation() + " has a size == " + getLocation().size() + " but the variation reference allele has length " + getReference().length() + " this = " + this);
+            throw new IllegalStateException("BUG: GenomeLoc " + getLocation() + " has a size == " + getLocation().size() + " but the variation reference allele has length " + getReference().length() + " this = " + this);
-            }
         }
     }
 

java/src/org/broadinstitute/sting/oneoffprojects/variantcontext/VariantContextAdaptors.java

@@ -9,7 +9,7 @@ import java.util.*;
 
 
 public class VariantContextAdaptors {
-    public static VariantContext dbsnp2VariantContext(rodDbSNP dbsnp) {
+    public static VariantContext dbsnpToVariantContext(rodDbSNP dbsnp) {
         if ( dbsnp.isSNP() || dbsnp.isIndel() || dbsnp.varType.contains("mixed") ) {
             VariantContext vc = new VariantContext(dbsnp.getLocation());
 
@@ -37,7 +37,7 @@ public class VariantContextAdaptors {
             return null; // can't handle anything else
     }
 
-    public static VariantContext vcf2VariantContext(RodVCF vcf) {
+    public static VariantContext vcfToVariantContext(RodVCF vcf) {
         if ( vcf.isSNP() || vcf.isIndel() ) {
             VariantContext vc = new VariantContext(vcf.getLocation());
 
@@ -52,7 +52,7 @@ public class VariantContextAdaptors {
             vc.setNegLog10PError(vcf.getNegLog10PError());
             vc.setAttributes(vcf.getInfoValues());
             vc.putAttribute("ID", vcf.getID());
-            vc.putAttribute("FILTER", vcf.getFilterString());
+            if ( vcf.isFiltered() ) vc.setFilters(Arrays.asList(vcf.getFilteringCodes()));
 
             // add all of the alt alleles
             for ( String alt : vcf.getAlternateAlleleList() ) {

java/src/org/broadinstitute/sting/oneoffprojects/variantcontext/VariantContextUtils.java

@@ -13,9 +13,9 @@ public class VariantContextUtils {
      * throws an exception if there is a collision such that the same sample exists in both contexts
      * @return a context representing the merge of this context and the other provided context
      */
-    public VariantContext merge(VariantContext left, VariantContext other) {
+//    public VariantContext merge(VariantContext left, VariantContext other) {
-        return merge(left, other, false);
+//        return merge(left, other, false);
-    }
+//    }
 
     /**
      * @param other            another variant context
@@ -25,30 +25,30 @@ public class VariantContextUtils {
      *
      * @return a context representing the merge of this context and the other provided context
      */
-    public VariantContext merge(VariantContext left, VariantContext other, boolean uniquifySamples) {
+//    public VariantContext merge(VariantContext left, VariantContext other, boolean uniquifySamples) {
-        // todo -- make functional
+//        // todo -- make functional
-
+//
-        if ( !left.getLocation().equals(other.getLocation()) )
+//        if ( !left.getLocation().equals(other.getLocation()) )
-            throw new IllegalArgumentException("The locations must be identical for two contexts to be merged");
+//            throw new IllegalArgumentException("The locations must be identical for two contexts to be merged");
-
+//
-        Set<String> samples = left.getSampleNames();
+//        Set<String> samples = left.getSampleNames();
-        Set<Genotype> Gs = new HashSet<Genotype>(left.getGenotypes().values());
+//        Set<Genotype> Gs = new HashSet<Genotype>(left.getGenotypes().values());
-
+//
-        for ( Genotype g : other.getGenotypes().values() ) {
+//        for ( Genotype g : other.getGenotypes().values() ) {
-            if ( samples.contains(g.getSampleName()) ) {
+//            if ( samples.contains(g.getSampleName()) ) {
-                if ( uniquifySamples )
+//                if ( uniquifySamples )
-                    g.setSampleName(g.getSampleName() + UNIQUIFIED_SUFFIX);
+//                    g.setSampleName(g.getSampleName() + UNIQUIFIED_SUFFIX);
-                else
+//                else
-                    throw new IllegalStateException("The same sample name exists in both contexts when attempting to merge");
+//                    throw new IllegalStateException("The same sample name exists in both contexts when attempting to merge");
-            }
+//            }
-            Gs.add(g);
+//            Gs.add(g);
-        }
+//        }
-
+//
-        HashMap<Object, Object> attrs = new HashMap<Object, Object>(left.getAttributes());
+//        HashMap<Object, Object> attrs = new HashMap<Object, Object>(left.getAttributes());
-        attrs.putAll(other.getAttributes());
+//        attrs.putAll(other.getAttributes());
-
+//
-        return new VariantContext(left, Gs, attrs);
+//        return new VariantContext(left, Gs, attrs);
-    }
+//    }
 
 
     /**

java/src/org/broadinstitute/sting/playground/gatk/walkers/diagnostics/newvarianteval/SubstitutionEvaluation.java

@@ -44,9 +44,9 @@ public class SubstitutionEvaluation implements VariantEvaluation {
             for (char abase : BaseUtils.BASES) {
                 String altAllele = String.format("%c", abase);
 
-                if (BaseUtils.isTransition(rbase, abase)) {
+                if (BaseUtils.isTransition((byte)rbase, (byte)abase)) {
                     transitions += substitutions.get(refAllele, altAllele);
-                } else if (BaseUtils.isTransversion(rbase, abase)) {
+                } else if (BaseUtils.isTransversion((byte)rbase, (byte)abase)) {
                     transversions += substitutions.get(refAllele, altAllele);
                 } else {
                     unknown += substitutions.get(refAllele, altAllele);

java/src/org/broadinstitute/sting/playground/gatk/walkers/varianteval/TransitionTranversionAnalysis.java

@@ -28,7 +28,7 @@ public class TransitionTranversionAnalysis extends BasicVariantAnalysis implemen
     public String update(Variation eval, RefMetaDataTracker tracker, char ref, AlignmentContext context) {
         if (eval != null && eval.isSNP()) {
             char altBase = eval.getAlternativeBaseForSNP();
-            BaseUtils.BaseSubstitutionType subType = BaseUtils.SNPSubstitutionType(ref, altBase);
+            BaseUtils.BaseSubstitutionType subType = BaseUtils.SNPSubstitutionType((byte)ref, (byte)altBase);
             if (subType == BaseUtils.BaseSubstitutionType.TRANSITION)
                 nTransitions++;
             else

java/src/org/broadinstitute/sting/playground/gatk/walkers/variantoptimizer/AnnotationDataManager.java

@@ -83,7 +83,7 @@ public class AnnotationDataManager {
             final boolean isTrueVariant = false; //BUGBUG: Check truth input file to see if this variant is in the truth set
 
             // Decide if the variant is a transition or transversion
-            if( BaseUtils.isTransition( variant.getReferenceForSNP(), variant.getAlternativeBaseForSNP()) ) {
+            if( BaseUtils.isTransition( (byte)variant.getReferenceForSNP(), (byte)variant.getAlternativeBaseForSNP()) ) {
                 datum.incrementTi( isNovelVariant, isTrueVariant );
             } else {
                 datum.incrementTv( isNovelVariant, isTrueVariant );

java/src/org/broadinstitute/sting/utils/BaseUtils.java

@@ -24,20 +24,20 @@ public class BaseUtils {
      * @param base2
      * @return
      */
-    public static BaseSubstitutionType SNPSubstitutionType( char base1, char base2 ) {
+    public static BaseSubstitutionType SNPSubstitutionType( byte base1, byte base2 ) {
         BaseSubstitutionType t = isTransition(base1, base2) ? BaseSubstitutionType.TRANSITION : BaseSubstitutionType.TRANSVERSION;
         //System.out.printf("SNPSubstitutionType( char %c, char %c ) => %s%n", base1, base2, t);
         return t;
     }
 
-    public static boolean isTransition( char base1, char base2 ) {
+    public static boolean isTransition( byte  base1, byte  base2 ) {
         int b1 = simpleBaseToBaseIndex(base1);
         int b2 = simpleBaseToBaseIndex(base2);
         return b1 == 0 && b2 == 2 || b1 == 2 && b2 == 0 ||
                b1 == 1 && b2 == 3 || b1 == 3 && b2 == 1;
     }
 
-    public static boolean isTransversion( char base1, char base2 ) {
+    public static boolean isTransversion( byte  base1, byte  base2 ) {
         return ! isTransition(base1, base2);
     }
 

java/test/org/broadinstitute/sting/oneoffprojects/variantcontext/VariantContextTest.java

@@ -15,6 +15,8 @@ import org.junit.BeforeClass;
 
 import java.util.Arrays;
 import java.util.List;
+import java.util.Set;
+import java.util.Collection;
 import java.io.FileNotFoundException;
 import java.io.File;
 
@@ -37,7 +39,7 @@ public class VariantContextTest extends BaseTest {
     Allele del, delRef, ATC, ATCref;
 
     // A [ref] / T at 10
-    GenomeLoc snpLoc = GenomeLocParser.createGenomeLoc("chr1", 10, 11);
+    GenomeLoc snpLoc = GenomeLocParser.createGenomeLoc("chr1", 10, 10);
 
     // - / ATC [ref] from 20-23
     GenomeLoc delLoc = GenomeLocParser.createGenomeLoc("chr1", 20, 22);
@@ -80,6 +82,9 @@ public class VariantContextTest extends BaseTest {
         Assert.assertTrue(vc.isBiallelic());
         Assert.assertEquals(vc.getNAlleles(), 2);
 
+        Assert.assertTrue(vc.isTransversion());
+        Assert.assertFalse(vc.isTransition());
+
         Assert.assertEquals(vc.getReference(), Aref);
         Assert.assertEquals(vc.getAlleles().size(), 2);
         Assert.assertEquals(vc.getAlternateAlleles().size(), 1);
@@ -203,6 +208,19 @@ public class VariantContextTest extends BaseTest {
         new VariantContext(insLoc, Arrays.asList(Aref, A));
     }
 
+    @Test (expected = IllegalStateException.class)
+    public void testBadLoc1() {
+        logger.warn("testBadLoc1");
+        List<Allele> alleles = Arrays.asList(Aref, T, del);
+        VariantContext vc = new VariantContext(delLoc, alleles);
+    }
+
+
+    @Test (expected = IllegalStateException.class)
+    public void testBadTiTvRequest() {
+        logger.warn("testBadConstructorArgsDuplicateAlleles2");
+        new VariantContext(insLoc, Arrays.asList(Aref, ATC)).isTransition();
+    }
 
     @Test
     public void testAccessingSimpleSNPGenotypes() {
@@ -249,7 +267,7 @@ public class VariantContextTest extends BaseTest {
         logger.warn("testAccessingCompleteGenotypes");
 
         List<Allele> alleles = Arrays.asList(Aref, T, del);
-        VariantContext vc = new VariantContext(delLoc, alleles);
+        VariantContext vc = new VariantContext(snpLoc, alleles);
         logger.warn("vc = " + vc);
 
         Genotype g1 = new Genotype(Arrays.asList(Aref, Aref), "AA", 10);
@@ -266,6 +284,8 @@ public class VariantContextTest extends BaseTest {
         Assert.assertTrue(vc.isPolymorphic());
         Assert.assertEquals(vc.getGenotypes().size(), 6);
 
+        Assert.assertEquals(3, vc.getGenotypes(Arrays.asList("AA", "Td", "dd")).size());
+
         Assert.assertEquals(10, vc.getChromosomeCount());
         Assert.assertEquals(3, vc.getChromosomeCount(Aref));
         Assert.assertEquals(4, vc.getChromosomeCount(T));
@@ -302,6 +322,99 @@ public class VariantContextTest extends BaseTest {
         }
     }
 
+    @Test
+    public void testFilters() {
+        logger.warn("testFilters");
+
+        List<Allele> alleles = Arrays.asList(Aref, T, del);
+        Genotype g1 = new Genotype(Arrays.asList(Aref, Aref), "AA", 10);
+        Genotype g2 = new Genotype(Arrays.asList(Aref, T), "AT", 10);
+        VariantContext vc = new VariantContext(snpLoc, alleles, Arrays.asList(g1,g2));
+        logger.warn("vc = " + vc);
+
+        Assert.assertTrue(vc.isNotFiltered());
+        Assert.assertFalse(vc.isFiltered());
+        Assert.assertEquals(0, vc.getFilters().size());
+
+        vc.addFilter("BAD_SNP_BAD!");
+
+        Assert.assertFalse(vc.isNotFiltered());
+        Assert.assertTrue(vc.isFiltered());
+        Assert.assertEquals(1, vc.getFilters().size());
+
+        vc.addFilters(Arrays.asList("REALLY_BAD_SNP", "CHRIST_THIS_IS_TERRIBLE"));
+
+        Assert.assertFalse(vc.isNotFiltered());
+        Assert.assertTrue(vc.isFiltered());
+        Assert.assertEquals(3, vc.getFilters().size());
+
+        vc.clearFilters();
+
+        Assert.assertTrue(vc.isNotFiltered());
+        Assert.assertFalse(vc.isFiltered());
+        Assert.assertEquals(0, vc.getFilters().size());
+    }
+
+    @Test
+    public void testVCromGenotypes() {
+        logger.warn("testVCromGenotypes");
+
+        List<Allele> alleles = Arrays.asList(Aref, T, del);
+        Genotype g1 = new Genotype(Arrays.asList(Aref, Aref), "AA", 10);
+        Genotype g2 = new Genotype(Arrays.asList(Aref, T), "AT", 10);
+        Genotype g3 = new Genotype(Arrays.asList(T, T), "TT", 10);
+        Genotype g4 = new Genotype(Arrays.asList(Allele.NO_CALL, Allele.NO_CALL), "..", 10);
+        Genotype g5 = new Genotype(Arrays.asList(del, del), "--", 10);
+        VariantContext vc = new VariantContext(snpLoc, alleles, Arrays.asList(g1,g2,g3,g4,g5));
+        logger.warn("vc = " + vc);
+
+        VariantContext vc12 = vc.subContextFromGenotypes(Arrays.asList(g1,g2));
+        VariantContext vc1 = vc.subContextFromGenotypes(Arrays.asList(g1));
+        VariantContext vc23 = vc.subContextFromGenotypes(Arrays.asList(g2, g3));
+        VariantContext vc4 = vc.subContextFromGenotypes(Arrays.asList(g4));
+        VariantContext vc14 = vc.subContextFromGenotypes(Arrays.asList(g1, g4));
+        VariantContext vc5 = vc.subContextFromGenotypes(Arrays.asList(g5));
+
+        Assert.assertTrue(vc12.isPolymorphic());
+        Assert.assertTrue(vc23.isPolymorphic());
+        Assert.assertTrue(vc1.isMonomorphic());
+        Assert.assertTrue(vc4.isMonomorphic());
+        Assert.assertTrue(vc14.isMonomorphic());
+        Assert.assertTrue(vc5.isPolymorphic());
+
+        Assert.assertTrue(vc12.isSNP());
+        Assert.assertTrue(vc12.isVariant());
+        Assert.assertTrue(vc12.isBiallelic());
+
+        Assert.assertFalse(vc1.isSNP());
+        Assert.assertFalse(vc1.isVariant());
+        Assert.assertFalse(vc1.isBiallelic());
+
+        Assert.assertTrue(vc23.isSNP());
+        Assert.assertTrue(vc23.isVariant());
+        Assert.assertTrue(vc23.isBiallelic());
+
+        Assert.assertFalse(vc4.isSNP());
+        Assert.assertFalse(vc4.isVariant());
+        Assert.assertFalse(vc4.isBiallelic());
+
+        Assert.assertFalse(vc14.isSNP());
+        Assert.assertFalse(vc14.isVariant());
+        Assert.assertFalse(vc14.isBiallelic());
+
+        Assert.assertTrue(vc5.isIndel());
+        Assert.assertTrue(vc5.isDeletion());
+        Assert.assertTrue(vc5.isVariant());
+        Assert.assertTrue(vc5.isBiallelic());
+
+        Assert.assertEquals(3, vc12.getChromosomeCount(Aref));
+        Assert.assertEquals(1, vc23.getChromosomeCount(Aref));
+        Assert.assertEquals(2, vc1.getChromosomeCount(Aref));
+        Assert.assertEquals(0, vc4.getChromosomeCount(Aref));
+        Assert.assertEquals(2, vc14.getChromosomeCount(Aref));
+        Assert.assertEquals(0, vc5.getChromosomeCount(Aref));
+    }
+
 
     @Test
     public void testManipulatingAlleles() {

java/test/org/broadinstitute/sting/utils/BaseUtilsTest.java

@@ -27,27 +27,27 @@ public class BaseUtilsTest extends BaseTest {
     public void testTransitionTransversion() {
         logger.warn("Executing testTransitionTransversion");
 
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'G' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'G' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'C', 'A' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'C', (byte)'A' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'C', 'T' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'C', (byte)'T' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'C', 'G' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'C', (byte)'G' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'T', 'A' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'T', (byte)'A' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'T', 'C' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'T', (byte)'C' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'T', 'G' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'T', (byte)'G' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'G', 'A' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'G', (byte)'A' ) == BaseUtils.BaseSubstitutionType.TRANSITION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'G', 'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'G', (byte)'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'G', 'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'G', (byte)'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
 
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'a', 'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'a', (byte)'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'a', 'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'a', (byte)'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'T' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'C' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 't' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'t' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'A', 'c' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'A', (byte)'c' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'a', 't' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'a', (byte)'t' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
-        Assert.assertTrue( BaseUtils.SNPSubstitutionType( 'a', 'c' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
+        Assert.assertTrue( BaseUtils.SNPSubstitutionType( (byte)'a', (byte)'c' ) == BaseUtils.BaseSubstitutionType.TRANSVERSION );
     }
 
     @Test