Adding in a specific test that the FILTER field exists.

Author: jonn-smith

src/test/java/org/broadinstitute/hellbender/tools/funcotator/FuncotatorIntegrationTest.java

@@ -60,7 +60,10 @@ public class FuncotatorIntegrationTest extends CommandLineProgramTest {
     private static final String  GERMLINE_DATASOURCES_FOLDER   = "funcotator_dataSources_germline_latest";
 
     private static final String XSV_CLINVAR_MULTIHIT_TEST_VCF = toolsTestDir + "funcotator" + File.separator + "clinvar_hg19_multihit_test.vcf";
+    private static final String FILTER_TEST_VCF               = toolsTestDir + "funcotator" + File.separator + "FILTER_test.vcf";
     private static final String DS_XSV_CLINVAR_TESTS          = largeFileTestDir + "funcotator" + File.separator + "small_ds_clinvar_hg19" + File.separator;
+    private static final String DS_FILTER_PARSE_TESTS         = largeFileTestDir + "funcotator" + File.separator + "small_ds_FILTER_test" + File.separator;
+
     private static final String NOT_M2_TEST_HG19 = toolsTestDir + "funcotator/NotM2_test_custom_maf_fields.vcf";
     private static final String M2_TEST_HG19 = toolsTestDir + "funcotator/M2_test_custom_maf_fields.vcf";
     private static final String NOT_M2_TEST_HG19_TUMOR_ONLY = toolsTestDir + "funcotator/NotM2_test_custom_maf_fields_tumor_only.vcf";
@@ -903,6 +906,41 @@ public void testCanAnnotateHg38ClinvarAndGencodeV28() {
                 .count(), NUM_CLINVAR_HITS);
     }
 
+    @Test
+    public void testFilterParsing() {
+
+        final File outputFile = createTempFile("tmpTestFilterParsing", "vcf");
+
+        final ArgumentsBuilder arguments = createBaselineArgumentsForFuncotator(
+                FILTER_TEST_VCF,
+                outputFile,
+                b37Reference,
+                DS_FILTER_PARSE_TESTS,
+                FuncotatorTestConstants.REFERENCE_VERSION_HG19,
+                FuncotatorArgumentDefinitions.OutputFormatType.VCF,
+                false);
+
+        arguments.addBooleanArgument(FuncotatorArgumentDefinitions.FORCE_B37_TO_HG19_REFERENCE_CONTIG_CONVERSION, true);
+
+        runCommandLine(arguments);
+
+        final Pair<VCFHeader, List<VariantContext>> tempVcf =  VariantContextTestUtils.readEntireVCFIntoMemory(outputFile.getAbsolutePath());
+        Assert.assertEquals( tempVcf.getRight().size(), 1 );
+
+        final String[] funcotatorKeys = FuncotatorUtils.extractFuncotatorKeysFromHeaderDescription(tempVcf.getLeft().getInfoHeaderLine(VcfOutputRenderer.FUNCOTATOR_VCF_FIELD_NAME).getDescription());
+
+        final VariantContext variantContext = tempVcf.getRight().get(0);
+        final Map<Allele, FuncotationMap> funcs = FuncotatorUtils.createAlleleToFuncotationMapFromFuncotationVcfAttribute(
+                funcotatorKeys, variantContext, "Gencode_19_annotationTranscript", "FAKE_SOURCE");
+
+        final String txId = funcs.get(variantContext.getAlternateAllele(0)).getTranscriptList().get(0);
+        Assert.assertEquals( funcs.get(variantContext.getAlternateAllele(0)).get(txId).size(), 1 );
+
+        final Funcotation funcotation = funcs.get(variantContext.getAlternateAllele(0)).get(txId).get(0);
+
+        Assert.assertEquals(funcotation.getField("dbSnp_FILTER"), "FILTER_8");
+    }
+
     @Test
     public void testExclusionFromDatasourceVcfToVcf() {
         // Clinvar datasource did  go through one round of preprocessing to make contig names "1" --> "chr1" (for example).  This is an issue with ClinVar, not GATK.

src/test/resources/large/funcotator/small_ds_FILTER_test/dbSnp/hg19/dbSnp.config

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e43b7f14d6f7836abef57eefb2d309ccd1f2c0b34ba5fea5759af3f3daac602
+size 1771

src/test/resources/large/funcotator/small_ds_FILTER_test/dbSnp/hg19/dbSnp.regressionTestSet.hg19.FILTER.vcf.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:987bf261c1152138adc618567d2ca689bac39c58058fcad0425a1fc43689ba10
+size 33359546

src/test/resources/large/funcotator/small_ds_FILTER_test/dbSnp/hg19/dbSnp.regressionTestSet.hg19.FILTER.vcf.gz.tbi

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a923868bb4e69c3919dcb1ff1d8e81c9f3d617c51e5b933df45ec410b3297992
+size 18844

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.config

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb604e5862217c3978cc2fb0abf2f6f2cb8b4cf34681a026fedcd891aa0aa6f0
+size 1679

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.v19.PIK3CA.gtf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7c09d309121a8664c3f465ca169fadbfe539a188ab541052c4e6719aa209a15
+size 23214

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.v19.PIK3CA.gtf.idx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd2aa8475b77a6c8e2c782a2de47c7ec8070bfd541303e548e96f787a8920c1d
+size 312

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.v19.PIK3CA_transcript.dict

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74883af09a5772034d53f2df884a82e34df415951081e5754bedf1517ee9faa9
+size 968

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.v19.PIK3CA_transcript.fasta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d3f582d1ca1ed259c77e4a82f4bfed8e9253b4b5c3aba9a63eb08f75f5cc68d
+size 10353

src/test/resources/large/funcotator/small_ds_FILTER_test/gencode_pik3ca/hg19/gencode.v19.PIK3CA_transcript.fasta.fai

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4e49f4d57d32943e1663cf9902460ff7a2e2c03dfaf0bd3c5fc7ad9f01c2d47
+size 441