Added evaluation module to count functional classes (missense, nonsense, etc.). At the moment, it only understands Cancer's MAF annotations. Added integration test for the functional class counting. Added better description for VariantEval.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3595 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/walkers/varianteval/CountFunctionalClasses.java

@@ -0,0 +1,77 @@
+package org.broadinstitute.sting.gatk.walkers.varianteval;
+
+import org.broadinstitute.sting.playground.utils.report.tags.Analysis;
+import org.broadinstitute.sting.playground.utils.report.tags.DataPoint;
+import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
+import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
+import org.broadinstitute.sting.gatk.contexts.variantcontext.VariantContext;
+import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
+
+@Analysis(name = "Count Functional Classes", description = "Counts instances of different functional variant classes (provided the variants are annotated with that information)")
+public class CountFunctionalClasses extends VariantEvaluator {
+    // the following fields are in output order:
+    @DataPoint(description = "miRNA")
+    long nMiRNA= 0;
+
+    @DataPoint(description = "3'-UTR")
+    long nUTR3 = 0;
+
+    @DataPoint(description = "Intron")
+    long nIntron = 0;
+
+    @DataPoint(description = "Splice-site")
+    long nSpliceSite= 0;
+
+    @DataPoint(description = "Read-through")
+    long nReadThrough = 0;
+
+    @DataPoint(description = "Nonsense")
+    long nNonsense = 0;
+
+    @DataPoint(description = "Missense")
+    long nMissense = 0;
+
+    @DataPoint(description = "Synonymous")
+    long nSynonymous = 0;
+
+    @DataPoint(description = "5'-UTR")
+    long nUTR5= 0;
+
+    @DataPoint(description = "Promoter")
+    long nPromoter = 0;
+
+    public CountFunctionalClasses(VariantEvalWalker parent) {
+        super(parent);
+    }
+
+    public String getName() {
+        return "functionalclasses";
+    }
+
+    public boolean enabled() {
+        return false;
+    }
+
+    public int getComparisonOrder() {
+        return 1;
+    }
+
+    public String update1(VariantContext vc1, RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
+        String type = vc1.getAttributeAsString("type");
+
+        if (type != null) {
+            if      (type.equalsIgnoreCase("miRNA")) { nMiRNA++; }
+            else if (type.equalsIgnoreCase("3'-UTR")) { nUTR3++; }
+            else if (type.equalsIgnoreCase("Intron")) { nIntron++; }
+            else if (type.equalsIgnoreCase("Splice_site")) { nSpliceSite++; }
+            else if (type.equalsIgnoreCase("Read-through")) { nReadThrough++; }
+            else if (type.equalsIgnoreCase("Nonsense")) { nNonsense++; }
+            else if (type.equalsIgnoreCase("Missense")) { nMissense++; }
+            else if (type.equalsIgnoreCase("Synonymous")) { nSynonymous++; }
+            else if (type.equalsIgnoreCase("5'-UTR")) { nUTR5++; }
+            else if (type.equalsIgnoreCase("Promoter")) { nPromoter++; }
+        }
+
+        return null;
+    }
+}

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

@@ -103,7 +103,7 @@ import java.util.*;
 
 
 /**
- * Test routine for new VariantContext object
+ * General-purpose tool for variant evaluation (% in dbSNP, genotype concordance, Ts/Tv ratios, and a lot more)
  */
 @Reference(window=@Window(start=-50,stop=50))
 public class VariantEvalWalker extends RodWalker<Integer, Integer> {
@@ -389,6 +389,10 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> {
     private List<EvaluationContext> initializeEvaluationContexts(Set<String> evalNames, Set<String> compNames, List<VariantContextUtils.JexlVCMatchExp> selectExps) {
         List<EvaluationContext> contexts = new ArrayList<EvaluationContext>();
 
+        // todo -- add another for loop for each sample (be smart about the selection here -
+        // honor specifications of just one or a few samples), and put an "all" in here so
+        // that we don't lose multi-sample evaluations
+
         selectExps = append(selectExps, null);
         for ( String evalName : evalNames ) {
             for ( String compName : compNames ) {

java/test/org/broadinstitute/sting/gatk/walkers/varianteval/VariantEvalIntegrationTest.java

@@ -97,6 +97,21 @@ public class
         }
     }
 
+    @Test
+    public void testVEGenomicallyAnnotated() {
+        String vecmd = "-T VariantEval " +
+                       "-R " + oneKGLocation + "reference/human_b36_both.fasta " +
+                       "-L 21 " +
+                       "-D " + GATKDataLocation + "dbsnp_129_b36.rod " +
+                       "-E CountFunctionalClasses " +
+                       "-B eval,VCF," + validationDataLocation + "test.filtered.maf_annotated.vcf " +
+                       "-o %s";
+        String md5 = "d41d8cd98f00b204e9800998ecf8427e";
+
+        WalkerTestSpec spec = new WalkerTestSpec(vecmd, 1, Arrays.asList(md5));
+        executeTest("testVEGenomicallyAnnotated", spec);
+    }
+
     @Test
     public void testVEWriteVCF() {
         String extraArgs = "-L 1:1-10,000,000 -family NA19238+NA19239=NA19240 -MVQ 30";