Realignment filter annotates vcf with passing and failing read counts (broadinstitute#5328)

Author: davidbenjamin

src/main/java/org/broadinstitute/hellbender/tools/walkers/realignmentfilter/FilterAlignmentArtifacts.java

@@ -137,6 +137,7 @@ public void onTraversalStart() {
         final Set<VCFHeaderLine> headerLines = new HashSet<>(inputHeader.getMetaDataInSortedOrder());
         headerLines.add(GATKVCFHeaderLines.getFilterLine(GATKVCFConstants.ALIGNMENT_ARTIFACT_FILTER_NAME));
         headerLines.addAll(getDefaultToolVCFHeaderLines());
+        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.REALIGNMENT_COUNTS_VCF_ATTRIBUTE));
         final VCFHeader vcfHeader = new VCFHeader(headerLines, inputHeader.getGenotypeSamples());
         vcfWriter.writeHeader(vcfHeader);
     }
@@ -200,8 +201,10 @@ public void apply(final VariantContext vc, final ReadsContext readsContext, fina
         // if we haven't decided yet due to too few supporting reads, fail if there are more failures than successes
         passesFilter = passesFilter != Trilean.UNKNOWN ? passesFilter :
                 Trilean.of(failedRealignmentCount.intValue() <= succeededRealignmentCount.intValue());
+        final VariantContextBuilder vcb = new VariantContextBuilder(vc)
+                .attribute(GATKVCFConstants.REALIGNMENT_COUNTS_VCF_ATTRIBUTE, new int[] {succeededRealignmentCount.intValue(), failedRealignmentCount.intValue()});
 
-        vcfWriter.add(passesFilter == Trilean.TRUE ? vc : new VariantContextBuilder(vc).filter(GATKVCFConstants.ALIGNMENT_ARTIFACT_FILTER_NAME).make());
+        vcfWriter.add( (passesFilter == Trilean.TRUE ? vcb : vcb.filter(GATKVCFConstants.ALIGNMENT_ARTIFACT_FILTER_NAME)).make());
     }
 
     private static int pairScore(final Pair<BwaMemAlignment, BwaMemAlignment> pair) {

src/main/java/org/broadinstitute/hellbender/utils/variant/GATKVCFConstants.java

@@ -110,6 +110,7 @@ public final class GATKVCFConstants {
     public static final String NORMAL_ARTIFACT_LOD_ATTRIBUTE =      "N_ART_LOD";
     public static final String POPULATION_AF_VCF_ATTRIBUTE =        "POP_AF";
     public static final String GERMLINE_POSTERIORS_VCF_ATTRIBUTE =  "P_GERMLINE";
+    public static final String REALIGNMENT_COUNTS_VCF_ATTRIBUTE =   "RCNTS";
     public static final String POSTERIOR_PROB_OF_CONTAMINATION_ATTRIBUTE =  "P_CONTAM";
     public static final List<String> STANDARD_MUTECT_INFO_FIELDS = Arrays.asList(NORMAL_LOD_KEY, TUMOR_LOD_KEY, NORMAL_ARTIFACT_LOD_ATTRIBUTE,
             EVENT_COUNT_IN_HAPLOTYPE_KEY, IN_PON_VCF_ATTRIBUTE, POPULATION_AF_VCF_ATTRIBUTE, GERMLINE_POSTERIORS_VCF_ATTRIBUTE, POSTERIOR_PROB_OF_CONTAMINATION_ATTRIBUTE);

src/main/java/org/broadinstitute/hellbender/utils/variant/GATKVCFHeaderLines.java

@@ -202,6 +202,7 @@ private static void addFilterLine(final VCFFilterHeaderLine line) {
         // M2-related info lines
         addInfoLine(new VCFInfoHeaderLine(EVENT_COUNT_IN_HAPLOTYPE_KEY, 1, VCFHeaderLineType.Integer, "Number of events in this haplotype"));
         addInfoLine(new VCFInfoHeaderLine(NORMAL_LOD_KEY, VCFHeaderLineCount.A, VCFHeaderLineType.Float, "Normal LOD score"));
+        addInfoLine(new VCFInfoHeaderLine(REALIGNMENT_COUNTS_VCF_ATTRIBUTE, 2, VCFHeaderLineType.Integer, "Number of reads passing and failing realignment."));
         addInfoLine(new VCFInfoHeaderLine(TUMOR_LOD_KEY, VCFHeaderLineCount.A, VCFHeaderLineType.Float, "Log odds ratio score for variant"));
         addInfoLine(new VCFInfoHeaderLine(IN_PON_VCF_ATTRIBUTE, 0, VCFHeaderLineType.Flag, "site found in panel of normals"));
         addInfoLine(new VCFInfoHeaderLine(POPULATION_AF_VCF_ATTRIBUTE, VCFHeaderLineCount.A, VCFHeaderLineType.Float, "population allele frequencies of alt alleles"));