#243 only reference filtering precise variants when then nominal posi…

…tion is reference
PapenfussLab · Sep 18, 2019 · 715c594 · 715c594
1 parent aaf2ae9
commit 715c594
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 9 deletions.
diff --git a/src/main/java/au/edu/wehi/idsv/BreakpointSummary.java b/src/main/java/au/edu/wehi/idsv/BreakpointSummary.java
@@ -177,8 +177,13 @@ public boolean couldBeDeletionOfSize(int minSize, int maxSize) {
 	 * Determines whether this breakpoint could just be the reference allele
 	 * @return
 	 */
-	public boolean couldBeReferenceAllele() {
-		return couldBeDeletionOfSize(0, 0);
+	public boolean couldBeReferenceAllele(boolean isPrecise) {
+		if (!isPrecise) {
+			return couldBeDeletionOfSize(0, 0);
+		} else {
+			return new BreakpointSummary(referenceIndex, direction, nominal, referenceIndex2, direction2, nominal2)
+				.couldBeReferenceAllele(false);
+		}
 	}
 	/**
 	 * Gets the nominal position of the variant for variant calling purposes

diff --git a/src/main/java/au/edu/wehi/idsv/configuration/VariantCallingConfiguration.java b/src/main/java/au/edu/wehi/idsv/configuration/VariantCallingConfiguration.java
@@ -131,8 +131,8 @@ public List<VcfFilter> calculateBreakpointFilters(VariantContextDirectedBreakpoi
 			// we allow the default output 
 			filters.add(VcfFilter.SMALL_EVENT);
 		}
-		if (bp.couldBeReferenceAllele() && call.getUntemplatedSequence().length() == 0) {
-			// add condition: (IMPRECISE or (nominal is reference))
+		if (bp.couldBeReferenceAllele(call.isBreakendExact()) && call.getUntemplatedSequence().length() == 0) {
+			// unhandled edge case: precise only on one side
 			filters.add(VcfFilter.REFERENCE_ALLELE);
 		}
 		if (call.getBreakpointQual() < minScore) {

diff --git a/src/test/java/au/edu/wehi/idsv/BreakpointSummaryTest.java b/src/test/java/au/edu/wehi/idsv/BreakpointSummaryTest.java
@@ -218,4 +218,12 @@ public void isValid_should_check_both_breakends() {
 		assertTrue(new BreakpointSummary(0, BWD, 10, 1, FWD, 1).isValid(dict));
 		assertTrue(new BreakpointSummary(0, BWD, 1, 1, FWD, 10).isValid(dict));
 	}
+	@Test
+	public void couldBeReferenceAllele() {
+		assertTrue(new BreakpointSummary(0, FWD, 10, 0, BWD, 11).couldBeReferenceAllele(true));
+		assertTrue(new BreakpointSummary(0, FWD, 10, 0, BWD, 11).couldBeReferenceAllele(false));
+		// Can't be reference since the nominal position is non-reference
+		assertFalse(new BreakpointSummary(0, FWD, 6, 5,7, 0, BWD, 8, 7, 9).couldBeReferenceAllele(true));
+		assertTrue(new BreakpointSummary(0, FWD, 6, 5,7, 0, BWD, 8, 7, 9).couldBeReferenceAllele(false));
+	}
 }
diff --git a/src/test/java/au/edu/wehi/idsv/configuration/VariantCallingConfigurationTest.java b/src/test/java/au/edu/wehi/idsv/configuration/VariantCallingConfigurationTest.java
@@ -4,6 +4,7 @@
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertTrue;
 
+import au.edu.wehi.idsv.vcf.VcfSvConstants;
 import htsjdk.variant.variantcontext.VariantContextBuilder;
 import org.junit.Test;
 
@@ -18,17 +19,28 @@
 
 public class VariantCallingConfigurationTest extends TestHelper {
 	public VariantContextDirectedBreakpoint V(BreakpointSummary bs, String untemplated) {
+		return V(true, bs, untemplated);
+	}
+	public VariantContextDirectedBreakpoint V(boolean precise, BreakpointSummary bs, String untemplated) {
 		IdsvVariantContextBuilder builder = new IdsvVariantContextBuilder(getContext());
 		builder.breakpoint(bs, untemplated);
+		if (!precise) {
+			builder.attribute(VcfSvConstants.IMPRECISE_KEY, true);
+		}
 		return (VariantContextDirectedBreakpoint)builder.make();
 	}
 	@Test
-	public void breakpointFilters_should_filter_reference_allele() {
+	public void breakpointFilters_should_filter_reference_allele_imprecise() {
 		VariantCallingConfiguration p = getConfig().getVariantCalling();
-		assertTrue(p.calculateBreakpointFilters(V(new BreakpointSummary(0, FWD, 150, 100, 200, 0, BWD, 150, 100, 200), "")).contains(VcfFilter.REFERENCE_ALLELE));
-		assertTrue(p.calculateBreakpointFilters(V(new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
-		assertFalse(p.calculateBreakpointFilters(V(new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 3, 3, 3), "")).contains(VcfFilter.REFERENCE_ALLELE));
-		assertFalse(p.calculateBreakpointFilters(V(new BreakpointSummary(0, FWD, 1, 1, 1, 0, FWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertTrue(p.calculateBreakpointFilters(V(false, new BreakpointSummary(0, FWD, 150, 100, 200, 0, BWD, 150, 100, 200), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertTrue(p.calculateBreakpointFilters(V(false, new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertFalse(p.calculateBreakpointFilters(V(false, new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 3, 3, 3), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertFalse(p.calculateBreakpointFilters(V(false, new BreakpointSummary(0, FWD, 1, 1, 1, 0, FWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
+
+		assertFalse(p.calculateBreakpointFilters(V(true, new BreakpointSummary(0, FWD, 150, 100, 200, 0, BWD, 150, 100, 200), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertTrue(p.calculateBreakpointFilters(V(true, new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertFalse(p.calculateBreakpointFilters(V(true, new BreakpointSummary(0, FWD, 1, 1, 1, 0, BWD, 3, 3, 3), "")).contains(VcfFilter.REFERENCE_ALLELE));
+		assertFalse(p.calculateBreakpointFilters(V(true, new BreakpointSummary(0, FWD, 1, 1, 1, 0, FWD, 2, 2, 2), "")).contains(VcfFilter.REFERENCE_ALLELE));
 	}
 	@Test
 	public void breakpointFilters_should_filter_insertion() {