Funcotator - VCF ID for VCF Data Sources

src/main/java/org/broadinstitute/hellbender/tools/funcotator/dataSources/vcf/VcfFuncotationFactory.java

@@ -80,6 +80,12 @@ public class VcfFuncotationFactory extends DataSourceFuncotationFactory {
      */
     private final static String DUPLICATE_RECORD_DELIMITER = "|";
 
+    /**
+     * The name of the additional ID field to add to VCF annotations to preserve the ID of the original (data source)
+     * variant.
+     */
+    private static final String ID_FIELD_NAME = "ID";
+
     @VisibleForTesting
     int cacheHits = 0;
     @VisibleForTesting
@@ -147,12 +153,20 @@ private FuncotationMetadata createFuncotationMetadata(final Path sourceFilePath)
     List<VCFInfoHeaderLine> createFuncotationVcfInfoHeaderLines(final VCFHeader vcfHeader) {
         final List<VCFInfoHeaderLine> supportedVcfInfoHeaderLines = vcfHeader.getInfoHeaderLines().stream()
                 .filter(vcfInfoHeaderLine -> supportedFieldNames.contains(createFinalFieldName(name, vcfInfoHeaderLine.getID())))
+                .map(vcfInfoHeaderLine -> copyWithRename(vcfInfoHeaderLine, name))
                 .collect(Collectors.toList());
 
+        // Add in the ID field to the meta data:
+        final VCFInfoHeaderLine idHeaderLine = new VCFInfoHeaderLine(
+                createFinalFieldName(name, ID_FIELD_NAME),
+                VCFHeaderLineCount.A,
+                VCFHeaderLineType.String,
+                "ID of the variant from the data source creating this annotation."
+                );
+        supportedVcfInfoHeaderLines.add( idHeaderLine );
+
         // Make sure to rename the input VCF field names to the output funcotation field names for this funcotation factory.
-        return supportedVcfInfoHeaderLines.stream()
-                .map(vcfInfoHeaderLine -> copyWithRename(vcfInfoHeaderLine, name))
-                .collect(Collectors.toList());
+        return supportedVcfInfoHeaderLines;
     }
 
     private static VCFInfoHeaderLine copyWithRename(final VCFInfoHeaderLine vcfInfoHeaderLine, final String name) {
@@ -223,21 +237,21 @@ protected List<Funcotation> createFuncotationsOnVariant(final VariantContext var
 
             // Get rid of any null features.
             // By this point we know the feature type is correct, so we cast it:
-            final List<VariantContext> funcotationFactoryVariants = featureList.stream().filter(f -> f != null)
+            final List<VariantContext> featureVariantList = featureList.stream().filter(f -> f != null)
                     .map(f -> (VariantContext) f).collect(Collectors.toList());
 
             // Create a map that will keep the final outputs.  Default it to default funcotations for each alt allele in the
             //  query variant.
             final Map<Allele, Funcotation> outputOrderedMap = new LinkedHashMap<>();
 
-            for ( final VariantContext funcotationFactoryVariant : funcotationFactoryVariants ) {
+            for ( final VariantContext featureVariant : featureVariantList ) {
 
-                // The funcotationFactoryVariants already overlap the query variant in position, now get which
+                // The featureVariantList already overlaps the query variant in position, now get which
                 //  match in ref/alt as well.  And make sure to handle multiallelics in both the query variant and the
                 //  funcotation factory variant.
                 // matchIndices length will always be the same as the number of alt alleles in the variant (first parameter)
-                //  Note that this is not the same length as the funcotationFactoryVariant.
-                final int[] matchIndices = GATKVariantContextUtils.matchAllelesOnly(variant, funcotationFactoryVariant);
+                //  Note that this is not the same length as the featureVariant.
+                final int[] matchIndices = GATKVariantContextUtils.matchAllelesOnly(variant, featureVariant);
 
                 for (int i = 0; i < matchIndices.length; i++) {
                     final int matchIndex = matchIndices[i];
@@ -246,10 +260,13 @@ protected List<Funcotation> createFuncotationsOnVariant(final VariantContext var
 
                         final LinkedHashMap<String, Object> annotations = new LinkedHashMap<>(supportedFieldNamesAndDefaults);
 
-                        for (final Map.Entry<String, Object> entry : funcotationFactoryVariant.getAttributes().entrySet()) {
-                            populateAnnotationMap(funcotationFactoryVariant, variant, matchIndex, annotations, entry);
+                        for (final Map.Entry<String, Object> entry : featureVariant.getAttributes().entrySet()) {
+                            populateAnnotationMap(featureVariant, variant, matchIndex, annotations, entry);
                         }
 
+                        // Add the ID of the variant:
+                        annotations.put(createFinalFieldName(name, ID_FIELD_NAME), featureVariant.getID());
+
                         final TableFuncotation newFuncotation = TableFuncotation.create(annotations, queryAltAllele, name, supportedFieldMetadata);
                         outputOrderedMap.merge(queryAltAllele, newFuncotation, VcfFuncotationFactory::mergeDuplicateFuncotationFactoryVariant);
                     }
@@ -439,6 +456,10 @@ private void populateSupportedFieldNamesFromVcfFile() {
             }
             supportedFieldNames.add(createFinalFieldName(name, key));
         }
+
+        // Add our ID to the supported fields:
+        supportedFieldNamesAndDefaults.put(createFinalFieldName(name, ID_FIELD_NAME), "" );
+        supportedFieldNames.add(createFinalFieldName(name, ID_FIELD_NAME));
     }
 
     @VisibleForTesting

src/test/java/org/broadinstitute/hellbender/tools/funcotator/dataSources/vcf/VcfFuncotationFactoryUnitTest.java

@@ -104,6 +104,7 @@ public class VcfFuncotationFactoryUnitTest extends GATKBaseTest {
         FIELD_DEFAULT_MAP.put("WGT", "");
         FIELD_DEFAULT_MAP.put("WTD", "false");
         FIELD_DEFAULT_MAP.put("dbSNPBuildID", "");
+        FIELD_DEFAULT_MAP.put("ID", "");
     }
 
     //==================================================================================================================
@@ -155,14 +156,14 @@ private Object[][] provideForTestCreateFuncotationsOnVariant() {
                 helpProvideForTestCreateFuncotations("3", 61662, 61662, "T", "C",
                         Collections.singletonList(
                                 TableFuncotation.create(FIELD_DEFAULT_MAP.keySet().stream().map(s -> FACTORY_NAME + "_" + s).collect(Collectors.toList()),
-                                        Arrays.asList("true", "false", "0.9744,0.02556", "false", "false", "1", "false", "true", "false", "", "false", "true", "false", "true", "true", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "73009205", "61662", "false", "false", "0", "true", "0", "false", "0.954392,0.0456075", "false", "false", "false", "SNV", "true", "0x05010000000515043e000100", "1", "false", "130"),
+                                        Arrays.asList("true", "false", "0.9744,0.02556", "false", "false", "1", "false", "true", "false", "", "false", "true", "false", "true", "true", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "73009205", "61662", "false", "false", "0", "true", "0", "false", "0.954392,0.0456075", "false", "false", "false", "SNV", "true", "0x05010000000515043e000100", "1", "false", "130", "rs73009205"),
                                         Allele.create("C"), FACTORY_NAME, null)
                         )
                 ),
                 // No matching VCF features (three overlap by position only), since there are no indels in dbSNP (the test datasource), so the ground truth should be a default entry, which was constructed here manually:
                 helpProvideForTestCreateFuncotations("3", 64157, 64166, "AGAAAGGTCA", "TCTTTCCAGT",
                         Collections.singletonList(TableFuncotation.create(FIELD_DEFAULT_MAP.keySet().stream().map(s -> FACTORY_NAME + "_" + s).collect(Collectors.toList()),
-                                Arrays.asList("false", "false", "", "false", "false", "", "false", "false", "false", "", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "", "", "false", "false", "", "false", "", "false", "", "false", "false", "false", "", "false", "", "", "false", ""),
+                                Arrays.asList("false", "false", "", "false", "false", "", "false", "false", "false", "", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "false", "", "", "false", "false", "", "false", "", "false", "", "false", "false", "false", "", "false", "", "", "false", "", ""),
                                 Allele.create("TCTTTCCAGT"), FACTORY_NAME, null))
                 ),
         };