Merge pull request #1030 from broadinstitute/rhl_incorrect_rbp

Merge if both GT are phased

protected/gatk-tools-protected/src/main/java/org/broadinstitute/gatk/tools/walkers/phasing/PhasingUtils.java

@@ -363,7 +363,7 @@ class PhasingUtils {
      *
      * @param gt1 genotype 1
      * @param gt2 genotype 2
-     * @return true if genotypes have the same number of chromosomes, haplotype, number of attributes
+     * @return true if genotypes have the same number of chromosomes, haplotype, phased, number of attributes
      * as chromosomes, and either genotype is homozygous, false otherwise
      */
     static boolean alleleSegregationIsKnown(Genotype gt1, Genotype gt2) {
@@ -371,6 +371,10 @@ class PhasingUtils {
         if (gt1.getPloidy() != gt2.getPloidy())
             return false;
 
+        // If gt1 or gt2 are not phased, then can not be merged.
+        if (!gt1.isPhased() || !gt2.isPhased())
+            return false;
+
         // If gt1 or gt2 are homozygous, then could be merged.
         if (gt1.isHom() || gt2.isHom())
             return true;

protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/phasing/PhasingUtilsUnitTest.java

@@ -81,6 +81,7 @@ public class PhasingUtilsUnitTest extends BaseTest {
     private ReferenceSequenceFile referenceFile;
     private Genotype genotype1;
     private Genotype genotype2;
+    private Genotype genotype2unphased;
     private String contig;
     private List<Allele> alleleList1;
     private List<Allele> alleleList2;
@@ -93,8 +94,9 @@ public class PhasingUtilsUnitTest extends BaseTest {
         genomeLocParser = new GenomeLocParser(referenceFile);
         alleleList1 = Arrays.asList(Allele.create("T", true), Allele.create("C", false));
         alleleList2 = Arrays.asList(Allele.create("G", true), Allele.create("A", false));
-        genotype1 = new GenotypeBuilder().name("sample").attribute("HP", new String[]{"10-1", "10-2"}).attribute("PQ", 100.0).alleles(alleleList1).make();
+        genotype1 = new GenotypeBuilder().name("sample").attribute("HP", new String[]{"10-1", "10-2"}).attribute("PQ", 100.0).alleles(alleleList1).phased(true).make();
-        genotype2 = new GenotypeBuilder().name("sample").attribute("HP", new String[]{"10-2", "10-1"}).attribute("PQ", 200.0).alleles(alleleList2).make();
+        genotype2 = new GenotypeBuilder().name("sample").attribute("HP", new String[]{"10-2", "10-1"}).attribute("PQ", 200.0).alleles(alleleList2).phased(true).make();
+        genotype2unphased = new GenotypeBuilder().name("sample").attribute("HP", new String[]{"10-2", "10-1"}).attribute("PQ", 200.0).alleles(alleleList2).phased(false).make();
         contig = new String("1");
         vc1 = new VariantContextBuilder().chr(contig).id("id1").source("TC").start(start).stop(start).alleles(alleleList1).genotypes(genotype1).make();
         vc2 = new VariantContextBuilder().chr(contig).id("id2").source("GA").start(start+1).stop(start+1).alleles(alleleList2).genotypes(genotype2).make();
@@ -184,6 +186,7 @@ public class PhasingUtilsUnitTest extends BaseTest {
     @Test
     public void TestAlleleSegregationIsKnown(){
         Assert.assertTrue(PhasingUtils.alleleSegregationIsKnown(genotype1, genotype2));
+        Assert.assertFalse(PhasingUtils.alleleSegregationIsKnown(genotype1, genotype2unphased));
     }
 
     @Test

protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/phasing/ReadBackedPhasingIntegrationTest.java

@@ -138,7 +138,21 @@ public class ReadBackedPhasingIntegrationTest extends WalkerTest {
                         + " -L chr20:332341-802503",
                 1,
                 Arrays.asList("ac41d1aa9c9a67c07d894f485c29c574"));
-        executeTest("Use trio-phased VCF, adding read-backed phasing infomration in HP tag (as is now standard for RBP) [TEST SEVEN]", spec);
+        executeTest("Use trio-phased VCF, adding read-backed phasing information in HP tag (as is now standard for RBP) [TEST SEVEN]", spec);
     }
 
+    @Test
+    public void testDoNotMergeUnphasedSNPs() {
+        WalkerTestSpec spec = new WalkerTestSpec(
+                "-T ReadBackedPhasing" +
+                " -R " + hg19Reference +
+                " -I " + privateTestDir + "S17C1-8.KRAS.bam" +
+                " --variant " + privateTestDir + "S17C1-8_bwa_mutect_filtered.vcf" +
+                "  --phaseQualityThresh 20.0 -enableMergeToMNP -maxDistMNP 3 -L 12:25398281-25398284" +
+                " -o %s" +
+                " --no_cmdline_in_header",
+                1,
+                Arrays.asList("59ee67d657ee955477bca94d07014ac3"));
+        executeTest("Do not merge unphased SNPs", spec);
+    }
 }