(SV) Representation change commit 3:

change how DUP variants with short duplicated ref region are represented in NEW CODE PATH ONLY
broadinstitute · Apr 20, 2018 · 5b8e571 · 5b8e571
1 parent a597b70
commit 5b8e571
Show file tree

Hide file tree

Showing 7 changed files with 263 additions and 93 deletions.
diff --git a/...nstitute/hellbender/tools/spark/sv/discovery/inference/NovelAdjacencyAndAltHaplotype.java b/...nstitute/hellbender/tools/spark/sv/discovery/inference/NovelAdjacencyAndAltHaplotype.java
@@ -232,10 +232,16 @@ public List<SvType> toSimpleOrBNDTypes(final ReferenceMultiSource reference, fin
                 final int svLength = this.getComplication().getInsertedSequenceForwardStrandRep().length();
                 return Collections.singletonList( new SimpleSVType.Insertion(this, svLength) );
             }
-            case SMALL_DUP_EXPANSION: // TODO: 4/16/18 make adjustments based on duplicated size
+            case SMALL_DUP_EXPANSION:
             {
                 final int svLength = getLengthForDupTandem(this);
-                return Collections.singletonList( new SimpleSVType.DuplicationTandem(this, svLength) );
+                final BreakpointComplications.SmallDuplicationWithPreciseDupRangeBreakpointComplications duplicationComplication =
+                        (BreakpointComplications.SmallDuplicationWithPreciseDupRangeBreakpointComplications) this.getComplication();
+                if (duplicationComplication.getDupSeqRepeatUnitRefSpan().size() < 50) {
+                    return Collections.singletonList( new SimpleSVType.Insertion(this, svLength));
+                } else {
+                    return Collections.singletonList( new SimpleSVType.DuplicationTandem(this, svLength) );
+                }
             }
             case DEL_DUP_CONTRACTION:
             {
@@ -244,13 +250,19 @@ public List<SvType> toSimpleOrBNDTypes(final ReferenceMultiSource reference, fin
             }
             case SMALL_DUP_CPX:
             {
-                if ( ((BreakpointComplications.SmallDuplicationWithImpreciseDupRangeBreakpointComplications)
-                        this.getComplication()).isDupContraction() ) {
+                final BreakpointComplications.SmallDuplicationWithImpreciseDupRangeBreakpointComplications duplicationComplication =
+                        (BreakpointComplications.SmallDuplicationWithImpreciseDupRangeBreakpointComplications)
+                        this.getComplication();
+                if ( duplicationComplication.isDupContraction() ) {
                     final int svLength = leftJustifiedLeftRefLoc.getEnd() - leftJustifiedRightRefLoc.getStart();
                     return Collections.singletonList( new SimpleSVType.Deletion(this, svLength) );
-                } else { // TODO: 4/16/18 make adjustments based on duplicated size
+                } else {
                     final int svLength = getLengthForDupTandem(this);
-                    return Collections.singletonList( new SimpleSVType.DuplicationTandem(this, svLength) );
+                    if (duplicationComplication.getDupSeqRepeatUnitRefSpan().size() < 50) {
+                        return Collections.singletonList( new SimpleSVType.Insertion(this, svLength));
+                    } else {
+                        return Collections.singletonList( new SimpleSVType.DuplicationTandem(this, svLength) );
+                    }
                 }
             }
             default:

diff --git a/.../broadinstitute/hellbender/tools/spark/sv/discovery/AnnotatedVariantProducerUnitTest.java b/.../broadinstitute/hellbender/tools/spark/sv/discovery/AnnotatedVariantProducerUnitTest.java
@@ -109,10 +109,10 @@ private Object[][] dataForEvidenceAnnotation() {
         data.add(new Object[]{forSimpleTanDupContraction_minus, new String[]{"60"}, new String[]{String.valueOf(40)}});
 
         // simple tandem dup expansion from 1 unit to 2 units
-        data.add(new Object[]{forSimpleTanDupExpansion_plus, new String[]{"60"}, new String[]{String.valueOf(50)}});
+        data.add(new Object[]{forSimpleTanDupExpansion_ins_plus, new String[]{"60"}, new String[]{String.valueOf(50)}});
 
         // simple tandem dup expansion from 1 unit to 2 units and novel insertion
-        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_minus, new String[]{"60"}, new String[]{String.valueOf(137)}});
+        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_dup_minus, new String[]{"60"}, new String[]{String.valueOf(137)}});
 
         // tandem dup expansion from 1 unit to 2 units with pseudo-homology
         data.add(new Object[]{forComplexTanDup_1to2_pseudoHom_plus, new String[]{"60"}, new String[]{String.valueOf(127)}});
@@ -206,12 +206,12 @@ private Object[][] dataForIntegrativeTest() {
                 broadcastCNVCalls, referenceBroadcast, refSeqDictBroadcast});
 
         // simple tandem dup expansion from 1 unit to 2 units
-        data.add(new Object[]{forSimpleTanDupExpansion_plus,
+        data.add(new Object[]{forSimpleTanDupExpansion_ins_plus,
                 Stream.concat( commonAttributes.stream(), Sets.newHashSet(DUP_TAN_EXPANSION_STRING, DUP_REPEAT_UNIT_REF_SPAN, DUP_SEQ_CIGARS, DUPLICATION_NUMBERS, DUP_ORIENTATIONS).stream()).sorted().collect(Collectors.toList()),
                 broadcastCNVCalls, referenceBroadcast, refSeqDictBroadcast});
 
         // simple tandem dup expansion from 1 unit to 2 units and novel insertion
-        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_minus,
+        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_dup_minus,
                 Stream.concat( commonAttributes.stream(), Sets.newHashSet(DUP_TAN_EXPANSION_STRING, DUP_REPEAT_UNIT_REF_SPAN, DUP_SEQ_CIGARS, DUPLICATION_NUMBERS, DUP_ORIENTATIONS, INSERTED_SEQUENCE, INSERTED_SEQUENCE_LENGTH).stream()).sorted().collect(Collectors.toList()),
                 broadcastCNVCalls, referenceBroadcast, refSeqDictBroadcast});
 

diff --git a/...bender/tools/spark/sv/discovery/DiscoverVariantsFromContigAlignmentsSAMSparkUnitTest.java b/...bender/tools/spark/sv/discovery/DiscoverVariantsFromContigAlignmentsSAMSparkUnitTest.java
@@ -99,10 +99,10 @@ private Object[][] forTypeInference() {
         data.add(new Object[]{forSimpleTanDupContraction_plus.biPathBubble, DEL.name(), ImmutableSet.of(DUP_TAN_CONTRACTION_STRING)});
 
         // simple tandem dup expansion from 1 unit to 2 units
-        data.add(new Object[]{forSimpleTanDupExpansion_minus.biPathBubble, DUP.name(), ImmutableSet.of(DUP_TAN_EXPANSION_STRING)});
+        data.add(new Object[]{forSimpleTanDupExpansion_ins_minus.biPathBubble, DUP.name(), ImmutableSet.of(DUP_TAN_EXPANSION_STRING)});
 
         // simple tandem dup expansion from 1 unit to 2 units and novel insertion
-        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_plus.biPathBubble, DUP.name(), ImmutableSet.of(DUP_TAN_EXPANSION_STRING)});
+        data.add(new Object[]{forSimpleTanDupExpansionWithNovelIns_dup_plus.biPathBubble, DUP.name(), ImmutableSet.of(DUP_TAN_EXPANSION_STRING)});
 
         // tandem dup expansion from 1 unit to 2 units with pseudo-homology
         data.add(new Object[]{forComplexTanDup_1to2_pseudoHom_minus.biPathBubble, DUP.name(), ImmutableSet.of(DUP_TAN_EXPANSION_STRING)});