(SV) trim down ref bases for CPX variants (#4970)

* before all bases from affected region is extracted, leading to bloated VCF, now only the anchor base * also fixes downstream CPX variant re-interpreter
broadinstitute · Jul 9, 2018 · fa5244f · fa5244f
1 parent 111c8ef
commit fa5244f
Show file tree

Hide file tree

Showing 5 changed files with 49 additions and 47 deletions.
diff --git a/...g/broadinstitute/hellbender/tools/spark/sv/discovery/inference/CpxVariantInterpreter.java b/...g/broadinstitute/hellbender/tools/spark/sv/discovery/inference/CpxVariantInterpreter.java
@@ -325,12 +325,13 @@ static VariantContext turnIntoVariantContext(final Tuple2<CpxVariantCanonicalRep
         return rawVariantContextBuilder.make();
     }
 
-    // TODO: 6/22/18 this leads to unnecessarily large file size, next PR trim it down to a single base like the case for DEL
     private static byte[] getRefBases(final ReferenceMultiSource reference, final CpxVariantCanonicalRepresentation cpxVariantCanonicalRepresentation)
             throws IOException {
         final SimpleInterval affectedRefRegion = cpxVariantCanonicalRepresentation.getAffectedRefRegion();
+        SimpleInterval refBase = new SimpleInterval(affectedRefRegion.getContig(), affectedRefRegion.getStart(),
+                                                                                   affectedRefRegion.getStart());
         return reference
-                .getReferenceBases(affectedRefRegion)
+                .getReferenceBases(refBase)
                 .getBases();
     }
 

diff --git a/...lbender/tools/spark/sv/discovery/inference/SegmentedCpxVariantSimpleVariantExtractor.java b/...lbender/tools/spark/sv/discovery/inference/SegmentedCpxVariantSimpleVariantExtractor.java
@@ -15,6 +15,7 @@
 import org.broadinstitute.hellbender.tools.spark.sv.discovery.SvDiscoveryInputMetaData;
 import org.broadinstitute.hellbender.tools.spark.sv.discovery.alignment.AlignedContig;
 import org.broadinstitute.hellbender.tools.spark.sv.discovery.alignment.AssemblyContigWithFineTunedAlignments;
+import org.broadinstitute.hellbender.tools.spark.sv.utils.GATKSVVCFConstants;
 import org.broadinstitute.hellbender.tools.spark.sv.utils.SVInterval;
 import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.read.GATKRead;
@@ -679,7 +680,7 @@ List<VariantContext> extract(final VariantContext complexVC, final ReferenceMult
                 }
 
                 // head and tail insertions only
-                extractFrontAndRearInsertions(complexVC, refSegments, altArrangement, result);
+                extractFrontAndRearInsertions(complexVC, refSegments, altArrangement, reference, result);
             }
 
             final String sourceID = complexVC.getID();
@@ -737,8 +738,8 @@ private static void whenAllSegmentsAppearAsIs(final VariantContext complexVC, fi
             }
             if ( idx + refSegments.size() - 1 < altArrangement.size() - 1 ) { // e.g. there's more after 1,2,3,4,..., there could be (that is, if long enough) front insertion
                 final SimpleInterval insertionPos = new SimpleInterval(complexVC.getContig(), complexVC.getEnd(), complexVC.getEnd());
-                final byte[] refBases = complexVC.getReference().getBases();
-                final Allele anchorBaseRefAlleleRear = Allele.create(refBases[refBases.length - 1], true);
+                final byte[] refBases = getReferenceBases(insertionPos, reference);
+                final Allele anchorBaseRefAlleleRear = Allele.create(refBases, true);
                 final VariantContextBuilder rearIns = getInsFromOneEnd(false, idx + refSegments.size() - 1, insertionPos,
                         anchorBaseRefAlleleRear, refSegmentLengths, altArrangement, true);
                 if (rearIns != null) result.add(rearIns);
@@ -805,9 +806,8 @@ static List<SimpleInterval> compactifyMissingSegments(final Set<SimpleInterval>
         }
 
         private void extractFrontAndRearInsertions(final VariantContext complexVC, final List<SimpleInterval> refSegmentIntervals,
-                                                   final List<String> altArrangement,
+                                                   final List<String> altArrangement, final ReferenceMultiSource reference,
                                                    final List<VariantContextBuilder> result) {
-            final byte[] refBases = complexVC.getReference().getBases();
             final List<Integer> refSegmentLengths = refSegmentIntervals.stream().map(SimpleInterval::size).collect(Collectors.toList());
             // index pointing to first appearance of ref segment (inverted or not) in altArrangement, from either side
             int firstRefSegmentIdx = 0; // first front
@@ -819,8 +819,8 @@ private void extractFrontAndRearInsertions(final VariantContext complexVC, final
                 }
             }
             if (firstRefSegmentIdx > 0) {
-                final Allele anchorBaseRefAlleleFront = Allele.create(refBases[0], true);
                 final SimpleInterval startAndStop = makeOneBpInterval(complexVC.getContig(), complexVC.getStart());
+                final Allele anchorBaseRefAlleleFront = Allele.create(getReferenceBases(startAndStop, reference), true);
                 final VariantContextBuilder frontIns = getInsFromOneEnd(true, firstRefSegmentIdx, startAndStop, anchorBaseRefAlleleFront, refSegmentLengths, altArrangement, true);
                 if (frontIns != null) result.add( frontIns );
             }
@@ -835,9 +835,10 @@ private void extractFrontAndRearInsertions(final VariantContext complexVC, final
             }
 
             if (firstRefSegmentIdx != altArrangement.size() - 1) {
-                final Allele anchorBaseRefAlleleRear = Allele.create(refBases[refBases.length - 2], true);
-                final SimpleInterval startAndStop = makeOneBpInterval(complexVC.getContig(), complexVC.getEnd());
-                final VariantContextBuilder rearIns = getInsFromOneEnd(false, firstRefSegmentIdx, startAndStop, anchorBaseRefAlleleRear, refSegmentLengths, altArrangement, true);
+                final int pos = complexVC.getEnd();
+                final SimpleInterval insertionPos = makeOneBpInterval(complexVC.getContig(), pos);
+                final Allele anchorBaseRefAlleleRear = Allele.create(getReferenceBases(insertionPos, reference), true);
+                final VariantContextBuilder rearIns = getInsFromOneEnd(false, firstRefSegmentIdx, insertionPos, anchorBaseRefAlleleRear, refSegmentLengths, altArrangement, true);
                 if (rearIns != null) result.add( rearIns );
             }
         }

diff --git a/...nstitute/hellbender/tools/spark/sv/discovery/inference/CpxVariantInterpreterUnitTest.java b/...nstitute/hellbender/tools/spark/sv/discovery/inference/CpxVariantInterpreterUnitTest.java
@@ -289,7 +289,7 @@ public void testTurnIntoVariantContext() throws IOException {
         final CpxVariantCanonicalRepresentation cpxVariantCanonicalRepresentation = new CpxVariantCanonicalRepresentation(tig13846_3);
         final Tuple2<CpxVariantCanonicalRepresentation, Iterable<CpxVariantInducingAssemblyContig>> tuple2 =
                 new Tuple2<>(cpxVariantCanonicalRepresentation, Arrays.asList(tig13846_3, tig28220_5));
-        final byte[] refBases = TestUtilsForAssemblyBasedSVDiscovery.b38_reference_chr20_chr21.getReferenceBases(new SimpleInterval("chr20", 54849491, 54849615)).getBases();
+        final byte[] refBases = TestUtilsForAssemblyBasedSVDiscovery.b38_reference_chr20_chr21.getReferenceBases(new SimpleInterval("chr20", 54849491, 54849491)).getBases();
 
         final VariantContextBuilder baseVariantContextBuilder = cpxVariantCanonicalRepresentation.toVariantContext(refBases);
         baseVariantContextBuilder.attribute(GATKSVVCFConstants.TOTAL_MAPPINGS, 2);