From 80b5c5261a097b1a97ebc63e397d3f0f7d13f2eb Mon Sep 17 00:00:00 2001 From: Eric Banks Date: Mon, 18 Jul 2011 13:42:45 -0400 Subject: [PATCH] CombineVariants no longer combines records of different types. So now when combining SNP and indel callsets, overlapping calls get their own records. Useful for Khalid in the pipeline. For those interested, it turns out the previous behavior was doing the wrong thing occasionally (and this was even captured in the integration tests). --- .../walkers/variantutils/CombineVariants.java | 19 +++++++++++++------ .../variantcontext/VariantContextUtils.java | 13 ++++++++++++- .../CombineVariantsIntegrationTest.java | 6 +++--- 3 files changed, 28 insertions(+), 10 deletions(-) diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariants.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariants.java index 3dc47caa2..6970431ff 100755 --- a/public/java/src/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariants.java +++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariants.java @@ -172,17 +172,24 @@ public class CombineVariants extends RodWalker { if (minimumN > 1 && (vcs.size() - numFilteredRecords < minimumN)) return 0; - VariantContext mergedVC; + List mergedVCs = new ArrayList(); if ( master ) { - mergedVC = VariantContextUtils.masterMerge(vcs, "master"); + mergedVCs.add(VariantContextUtils.masterMerge(vcs, "master")); } else { - mergedVC = VariantContextUtils.simpleMerge(getToolkit().getGenomeLocParser(),vcs, priority, filteredRecordsMergeType, - genotypeMergeOption, true, printComplexMerges, ref.getBase(), SET_KEY, filteredAreUncalled, MERGE_INFO_WITH_MAX_AC); + Map> VCsByType = VariantContextUtils.separateVariantContextsByType(vcs); + // iterate over the keys (and not the values) so that it's deterministic + for ( VariantContext.Type type : VCsByType.keySet() ) { + mergedVCs.add(VariantContextUtils.simpleMerge(getToolkit().getGenomeLocParser(), VCsByType.get(type), + priority, filteredRecordsMergeType, genotypeMergeOption, true, printComplexMerges, + ref.getBase(), SET_KEY, filteredAreUncalled, MERGE_INFO_WITH_MAX_AC)); + } } - //out.printf(" merged => %s%nannotated => %s%n", mergedVC, annotatedMergedVC); + for ( VariantContext mergedVC : mergedVCs ) { + // only operate at the start of events + if ( mergedVC == null ) + continue; - if ( mergedVC != null ) { // only operate at the start of events HashMap attributes = new HashMap(mergedVC.getAttributes()); // re-compute chromosome counts VariantContextUtils.calculateChromosomeCounts(mergedVC, attributes, false); diff --git a/public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContextUtils.java b/public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContextUtils.java index c1eb48b68..212600360 100755 --- a/public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContextUtils.java +++ b/public/java/src/org/broadinstitute/sting/utils/variantcontext/VariantContextUtils.java @@ -492,7 +492,7 @@ public class VariantContextUtils { if ( ! filteredAreUncalled || vc.isNotFiltered() ) VCs.add(VariantContext.createVariantContextWithPaddedAlleles(vc,inputRefBase,false)); } - if ( VCs.size() == 0 ) // everything is filtered out and we're filteredareUncalled + if ( VCs.size() == 0 ) // everything is filtered out and we're filteredAreUncalled return null; // establish the baseline info from the first VC @@ -637,6 +637,17 @@ public class VariantContextUtils { return merged; } + public static Map> separateVariantContextsByType(Collection VCs) { + HashMap> mappedVCs = new HashMap>(); + for ( VariantContext vc : VCs ) { + if ( !mappedVCs.containsKey(vc.getType()) ) + mappedVCs.put(vc.getType(), new ArrayList()); + mappedVCs.get(vc.getType()).add(vc); + } + + return mappedVCs; + } + private static class AlleleMapper { private VariantContext vc = null; private Map map = null; diff --git a/public/java/test/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariantsIntegrationTest.java b/public/java/test/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariantsIntegrationTest.java index 00ee44f75..c5fd9bcbe 100755 --- a/public/java/test/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariantsIntegrationTest.java +++ b/public/java/test/org/broadinstitute/sting/gatk/walkers/variantutils/CombineVariantsIntegrationTest.java @@ -80,9 +80,9 @@ public class CombineVariantsIntegrationTest extends WalkerTest { @Test public void combineTrioCalls() { combine2("CEU.trio.2010_03.genotypes.vcf.gz", "YRI.trio.2010_03.genotypes.vcf.gz", "", "1d5a021387a8a86554db45a29f66140f", false); } // official project VCF files in tabix format @Test public void combineTrioCallsMin() { combine2("CEU.trio.2010_03.genotypes.vcf.gz", "YRI.trio.2010_03.genotypes.vcf.gz", " -minimalVCF", "20163d60f18a46496f6da744ab5cc0f9", false); } // official project VCF files in tabix format - @Test public void combine2Indels() { combine2("CEU.dindel.vcf4.trio.2010_06.indel.genotypes.vcf", "CEU.dindel.vcf4.low_coverage.2010_06.indel.genotypes.vcf", "", "5b82f37df1f5ba40f0474d71c94142ec", false); } + @Test public void combine2Indels() { combine2("CEU.dindel.vcf4.trio.2010_06.indel.genotypes.vcf", "CEU.dindel.vcf4.low_coverage.2010_06.indel.genotypes.vcf", "", "cba8f749f2444d69a54553b15328ed47", false); } - @Test public void combineSNPsAndIndels() { combine2("CEU.trio.2010_03.genotypes.vcf.gz", "CEU.dindel.vcf4.low_coverage.2010_06.indel.genotypes.vcf", "", "c58dca482bf97069eac6d9f1a07a2cba", false); } + @Test public void combineSNPsAndIndels() { combine2("CEU.trio.2010_03.genotypes.vcf.gz", "CEU.dindel.vcf4.low_coverage.2010_06.indel.genotypes.vcf", "", "78b169cf9955c9fd01340292d5ba2dca", false); } @Test public void uniqueSNPs() { combine2("pilot2.snps.vcf4.genotypes.vcf", "yri.trio.gatk_glftrio.intersection.annotated.filtered.chr1.vcf", "", "89f55abea8f59e39d1effb908440548c", true); } @@ -100,7 +100,7 @@ public class CombineVariantsIntegrationTest extends WalkerTest { " -priority NA19240_BGI,NA19240_ILLUMINA,NA19240_WUGSC,denovoInfo" + " -genotypeMergeOptions UNIQUIFY -L 1"), 1, - Arrays.asList("8b78339ccf7a5a5a837f79e88a3a38e5")); + Arrays.asList("0e475c98d5152fb12eb17f3907b849a9")); executeTest("threeWayWithRefs", spec); }