Merge branch 'master' of ssh://gsa1/humgen/gsa-scr1/gsa-engineering/git/unstable

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/VariantEvalWalker.java

@@ -68,7 +68,8 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
     protected Set<String> SAMPLE_EXPRESSIONS;
 
     @Argument(shortName="knownName", doc="Name of ROD bindings containing variant sites that should be treated as known when splitting eval rods into known and novel subsets", required=false)
-    protected String[] KNOWN_NAMES = {};
+    protected HashSet<String> KNOWN_NAMES = new HashSet<String>();
+    List<RodBinding<VariantContext>> knowns = new ArrayList<RodBinding<VariantContext>>();
 
     // Stratification arguments
     @Argument(fullName="stratificationModule", shortName="ST", doc="One or more specific stratification modules to apply to the eval track(s) (in addition to the standard stratifications, unless -noS is specified)", required=false)
@@ -108,9 +109,6 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
 
     // Variables
     private Set<SortableJexlVCMatchExp> jexlExpressions = new TreeSet<SortableJexlVCMatchExp>();
-    private Set<String> compNames = new TreeSet<String>();
-    private Set<String> knownNames = new TreeSet<String>();
-    private Set<String> evalNames = new TreeSet<String>();
 
     private Set<String> sampleNamesForEvaluation = new TreeSet<String>();
     private Set<String> sampleNamesForStratification = new TreeSet<String>();
@@ -149,23 +147,24 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
         comps.addAll(compsProvided);
         if ( dbsnp.dbsnp.isBound() ) {
             comps.add(dbsnp.dbsnp);
-            knownNames.add(dbsnp.dbsnp.getName());
+            knowns.add(dbsnp.dbsnp);
         }
 
         // Add a dummy comp track if none exists
         if ( comps.size() == 0 )
             comps.add(new RodBinding<VariantContext>(VariantContext.class, "none", "UNBOUND", "", new Tags()));
 
-        // Cache the rod names
+        // Set up set of additional knowns
-        for ( RodBinding<VariantContext> compRod : comps )
+        for ( RodBinding<VariantContext> compRod : comps ) {
-            compNames.add(compRod.getName());
+            if ( KNOWN_NAMES.contains(compRod.getName()) )
+                knowns.add(compRod);
+        }
 
+        // Collect the eval rod names
+        Set<String> evalNames = new TreeSet<String>();
         for ( RodBinding<VariantContext> evalRod : evals )
             evalNames.add(evalRod.getName());
 
-        // Set up set of additional known names
-        knownNames.addAll(Arrays.asList(KNOWN_NAMES));
-
         // Now that we have all the rods categorized, determine the sample list from the eval rods.
         Map<String, VCFHeader> vcfRods = VCFUtils.getVCFHeadersFromRods(getToolkit(), evalNames);
         Set<String> vcfSamples = SampleUtils.getSampleList(vcfRods, VariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE);
@@ -462,15 +461,15 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
 
     public static String getAllSampleName() { return ALL_SAMPLE_NAME; }
 
-    public Set<String> getKnownNames() { return knownNames; }
+    public List<RodBinding<VariantContext>> getKnowns() { return knowns; }
 
-    public Set<String> getEvalNames() { return evalNames; }
+    public List<RodBinding<VariantContext>> getEvals() { return evals; }
 
     public Set<String> getSampleNamesForEvaluation() { return sampleNamesForEvaluation; }
 
     public Set<String> getSampleNamesForStratification() { return sampleNamesForStratification; }
 
-    public Set<String> getCompNames() { return compNames; }
+    public List<RodBinding<VariantContext>> getComps() { return comps; }
 
     public Set<SortableJexlVCMatchExp> getJexlExpressions() { return jexlExpressions; }
 

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/AlleleCount.java

@@ -1,23 +1,25 @@
 package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
+import org.broadinstitute.sting.commandline.RodBinding;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.exceptions.UserException;
 import org.broadinstitute.sting.utils.variantcontext.Allele;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
+import java.util.List;
 
 public class AlleleCount extends VariantStratifier {
     // needs to know the variant context
     private ArrayList<String> states = new ArrayList<String>();
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
+        List<RodBinding<VariantContext>> evals = getVariantEvalWalker().getEvals();
+
         // we can only work with a single eval VCF, and it must have genotypes
-        if ( evalNames.size() != 1 )
+        if ( evals.size() != 1 )
             throw new UserException.BadArgumentValue("AlleleCount", "AlleleCount stratification only works with a single eval vcf");
 
         // There are 2 x n sample chromosomes for diploids

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/AlleleFrequency.java

@@ -2,19 +2,17 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.MathUtils;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class AlleleFrequency extends VariantStratifier {
     // needs to know the variant context
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
         for( double a = 0.000; a <= 1.005; a += 0.005 ) {
             states.add(String.format("%.3f", a));

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/CompRod.java

@@ -1,24 +1,20 @@
 package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
+import org.broadinstitute.sting.commandline.RodBinding;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class CompRod extends VariantStratifier implements RequiredStratification {
-    // Needs to know the comp rods
-    private Set<String> compNames;
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
-        this.compNames = compNames;
-
         states = new ArrayList<String>();
-        states.addAll(compNames);
+        for ( RodBinding<VariantContext> rod : getVariantEvalWalker().getComps() )
+            states.add(rod.getName());
     }
 
     public ArrayList<String> getAllStates() {

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/Contig.java

@@ -2,20 +2,18 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class Contig extends VariantStratifier {
     // needs to know the variant context
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
-        states.addAll(contigNames);
+        states.addAll(getVariantEvalWalker().getContigNames());
         states.add("all");
     }
 

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/CpG.java

@@ -2,11 +2,9 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 /**
  * CpG is a stratification module for VariantEval that divides the input data by within/not within a CpG site
@@ -24,7 +22,7 @@ public class CpG extends VariantStratifier {
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
         states.add("all");
         states.add("CpG");
@@ -40,7 +38,7 @@ public class CpG extends VariantStratifier {
         if (ref != null && ref.getBases() != null) {
             String fwRefBases = new String(ref.getBases());
 
-            String leftFlank = fwRefBases.substring((fwRefBases.length()/2) - 1, (fwRefBases.length()/2) + 1);
+            //String leftFlank = fwRefBases.substring((fwRefBases.length()/2) - 1, (fwRefBases.length()/2) + 1);
             String rightFlank = fwRefBases.substring((fwRefBases.length()/2), (fwRefBases.length()/2) + 2);
 
             //if (leftFlank.equalsIgnoreCase("CG") || leftFlank.equalsIgnoreCase("GC") || rightFlank.equalsIgnoreCase("CG") || rightFlank.equalsIgnoreCase("GC")) {

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/Degeneracy.java

@@ -2,13 +2,11 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
 import java.util.HashMap;
 import java.util.HashSet;
-import java.util.Set;
 
 public class Degeneracy extends VariantStratifier {
     private ArrayList<String> states;
@@ -16,7 +14,7 @@ public class Degeneracy extends VariantStratifier {
     private HashMap<String, HashMap<Integer, String>> degeneracies;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
         states.add("1-fold");
         states.add("2-fold");

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/EvalRod.java

@@ -1,24 +1,20 @@
 package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
+import org.broadinstitute.sting.commandline.RodBinding;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class EvalRod extends VariantStratifier implements RequiredStratification {
-    // needs to know the eval rods
-    private Set<String> evalNames;
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
-        this.evalNames = evalNames;
-
         states = new ArrayList<String>();
-        states.addAll(evalNames);
+        for ( RodBinding<VariantContext> rod : getVariantEvalWalker().getEvals() )
+            states.add(rod.getName());
     }
 
     public ArrayList<String> getAllStates() {

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/Filter.java

@@ -2,18 +2,16 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class Filter extends VariantStratifier {
     // needs to know the variant context
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
         states.add("called");
         states.add("filtered");

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/FunctionalClass.java

@@ -2,18 +2,16 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class FunctionalClass extends VariantStratifier {
     // needs to know the variant context
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         states = new ArrayList<String>();
         states.add("all");
         states.add("silent");

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/JexlExpression.java

@@ -15,8 +15,8 @@ public class JexlExpression extends VariantStratifier implements StandardStratif
     private ArrayList<String> states;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
-        this.jexlExpressions = jexlExpressions;
+        jexlExpressions = getVariantEvalWalker().getJexlExpressions();
 
         states = new ArrayList<String>();
         states.add("none");

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/Novelty.java

@@ -1,21 +1,21 @@
 package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
+import org.broadinstitute.sting.commandline.RodBinding;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.*;
 
 public class Novelty extends VariantStratifier implements StandardStratification {
     // needs the variant contexts and known names
-    private Set<String> knownNames;
+    private List<RodBinding<VariantContext>> knowns;
     final private ArrayList<String> states = new ArrayList<String>(Arrays.asList("all", "known", "novel"));
 
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
-        this.knownNames = knownNames;
+        knowns = getVariantEvalWalker().getKnowns();
     }
 
     public ArrayList<String> getAllStates() {
@@ -24,13 +24,11 @@ public class Novelty extends VariantStratifier implements StandardStratification
 
     public ArrayList<String> getRelevantStates(ReferenceContext ref, RefMetaDataTracker tracker, VariantContext comp, String compName, VariantContext eval, String evalName, String sampleName) {
         if (tracker != null && eval != null) {
-            for (final String knownName : knownNames) {
+            final Collection<VariantContext> knownComps = tracker.getValues(knowns, ref.getLocus());
-                final Collection<VariantContext> knownComps = tracker.getValues(VariantContext.class, knownName, ref.getLocus());
+            for ( final VariantContext c : knownComps ) {
-                for ( final VariantContext c : knownComps ) {
+                // loop over sites, looking for something that matches the type eval
-                    // loop over sites, looking for something that matches the type eval
+                if ( eval.getType() == c.getType() ) {
-                    if ( eval.getType() == c.getType() ) {
+                    return new ArrayList<String>(Arrays.asList("all", "known"));
-                        return new ArrayList<String>(Arrays.asList("all", "known"));
-                    }
                 }
             }
         }

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/Sample.java

@@ -2,20 +2,18 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public class Sample extends VariantStratifier {
     // needs the sample names
     private ArrayList<String> samples;
 
     @Override
-    public void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames) {
+    public void initialize() {
         samples = new ArrayList<String>();
-        samples.addAll(sampleNames);
+        samples.addAll(getVariantEvalWalker().getSampleNamesForStratification());
     }
 
     public ArrayList<String> getAllStates() {

public/java/src/org/broadinstitute/sting/gatk/walkers/varianteval/stratifications/VariantStratifier.java

@@ -3,11 +3,9 @@ package org.broadinstitute.sting.gatk.walkers.varianteval.stratifications;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
 import org.broadinstitute.sting.gatk.walkers.varianteval.VariantEvalWalker;
-import org.broadinstitute.sting.gatk.walkers.varianteval.util.SortableJexlVCMatchExp;
 import org.broadinstitute.sting.utils.variantcontext.VariantContext;
 
 import java.util.ArrayList;
-import java.util.Set;
 
 public abstract class VariantStratifier implements Comparable {
     private VariantEvalWalker variantEvalWalker;
@@ -27,7 +25,7 @@ public abstract class VariantStratifier implements Comparable {
         this.variantEvalWalker = variantEvalWalker;
     }
 
-    public abstract void initialize(Set<SortableJexlVCMatchExp> jexlExpressions, Set<String> compNames, Set<String> knownNames, Set<String> evalNames, Set<String> sampleNames, Set<String> contigNames);
+    public abstract void initialize();
 
     public ArrayList<String> getAllStates() {
         return new ArrayList<String>();

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

@@ -103,7 +103,7 @@ public class VariantEvalUtils {
                 try {
                     VariantStratifier vs = c.newInstance();
                     vs.setVariantEvalWalker(variantEvalWalker);
-                    vs.initialize(variantEvalWalker.getJexlExpressions(), variantEvalWalker.getCompNames(), variantEvalWalker.getKnownNames(), variantEvalWalker.getEvalNames(), variantEvalWalker.getSampleNamesForStratification(), variantEvalWalker.getContigNames());
+                    vs.initialize();
 
                     strats.add(vs);
                 } catch (InstantiationException e) {