Search in sources :

Example 6 with CopyNumberTriStateAllele

use of org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele in project gatk by broadinstitute.

the class EvaluateCopyNumberTriStateCallsIntegrationTest method checkOutputCallsWithOverlappingTruthConcordance.

private void checkOutputCallsWithOverlappingTruthConcordance(final File truthFile, final File callsFile, final File targetsFile, final File vcfOutput, final EvaluationFiltersArgumentCollection filteringOptions) {
    final List<VariantContext> truthVariants = readVCFFile(truthFile);
    final List<VariantContext> outputVariants = readVCFFile(vcfOutput);
    final List<VariantContext> callsVariants = readVCFFile(callsFile);
    final Set<String> outputSamples = outputVariants.get(0).getSampleNames();
    final TargetCollection<Target> targets = TargetArgumentCollection.readTargetCollection(targetsFile);
    for (final VariantContext truth : truthVariants) {
        final List<Target> overlappingTargets = targets.targets(truth);
        final List<VariantContext> overlappingOutput = outputVariants.stream().filter(vc -> new SimpleInterval(vc).overlaps(truth)).collect(Collectors.toList());
        final List<VariantContext> overlappingCalls = callsVariants.stream().filter(vc -> new SimpleInterval(vc).overlaps(truth)).collect(Collectors.toList());
        if (overlappingTargets.isEmpty()) {
            Assert.assertTrue(overlappingOutput.isEmpty());
            continue;
        }
        @SuppressWarnings("all") final VariantContext matchingOutput = overlappingOutput.stream().filter(vc -> new SimpleInterval(truth).equals(new SimpleInterval(vc))).findAny().get();
        final int[] sampleCallsCount = new int[CopyNumberTriStateAllele.ALL_ALLELES.size()];
        for (final String sample : outputSamples) {
            final Genotype outputGenotype = matchingOutput.getGenotype(sample);
            final List<Pair<VariantContext, Genotype>> sampleCalls = overlappingCalls.stream().map(vc -> new ImmutablePair<>(vc, vc.getGenotype(sample))).filter(p -> XHMMSegmentGenotyper.DISCOVERY_TRUE.equals(p.getRight().getExtendedAttribute(XHMMSegmentGenotyper.DISCOVERY_KEY))).filter(p -> callPassFilters(p.getLeft(), p.getRight(), targets, filteringOptions)).collect(Collectors.toList());
            final int[] expectedCounts = new int[CopyNumberTriStateAllele.ALL_ALLELES.size()];
            sampleCalls.forEach(p -> {
                expectedCounts[CopyNumberTriStateAllele.valueOf(p.getRight().getAllele(0)).index()]++;
            });
            final int[] actualCounts = GATKProtectedVariantContextUtils.getAttributeAsIntArray(outputGenotype, VariantEvaluationContext.CALLED_ALLELE_COUNTS_KEY, () -> new int[CopyNumberTriStateAllele.ALL_ALLELES.size()], 0);
            Assert.assertEquals(actualCounts, expectedCounts, Arrays.toString(actualCounts) + " " + Arrays.toString(expectedCounts));
            final int expectedTotalCount = (int) MathUtils.sum(expectedCounts);
            final int actualTotalCount = GATKProtectedVariantContextUtils.getAttributeAsInt(outputGenotype, VariantEvaluationContext.CALLED_SEGMENTS_COUNT_KEY, -1);
            Assert.assertEquals(actualTotalCount, expectedTotalCount);
            final int expectedTargetCount = sampleCalls.stream().mapToInt(p -> targets.targetCount(p.getLeft())).sum();
            final int observedTargetCount = GATKProtectedVariantContextUtils.getAttributeAsInt(outputGenotype, VariantEvaluationContext.CALLED_TARGET_COUNT_KEY, -1);
            Assert.assertEquals(observedTargetCount, expectedTargetCount);
            final Allele truthCallAllele = outputTruthAllele(outputGenotype);
            final boolean isMixed = IntStream.of(actualCounts).filter(i -> i > 0).count() > 1;
            final String evalClass = GATKProtectedVariantContextUtils.getAttributeAsString(outputGenotype, VariantEvaluationContext.EVALUATION_CLASS_KEY, null);
            if (sampleCalls.size() > 0 && !isMixed) {
                final Pair<VariantContext, Genotype> bestCall = sampleCalls.stream().sorted((p1, p2) -> -Double.compare(callGQ(p1.getRight()), callGQ(p2.getRight()))).findFirst().get();
                final CopyNumberTriStateAllele expectedCall = CopyNumberTriStateAllele.valueOf(bestCall.getRight().getAllele(0));
                final CopyNumberTriStateAllele actualCall = CopyNumberTriStateAllele.valueOf(outputGenotype.getAllele(0));
                Assert.assertEquals(actualCall, expectedCall);
                sampleCallsCount[expectedCall.index()]++;
                if (!truthCallAllele.isReference()) {
                    if (truthCallAllele.equals(actualCall)) {
                        Assert.assertEquals(evalClass, EvaluationClass.TRUE_POSITIVE.acronym);
                    } else if (!truthCallAllele.isNoCall()) {
                        Assert.assertEquals(evalClass, EvaluationClass.DISCORDANT_POSITIVE.acronym);
                    } else {
                        Assert.assertNull(evalClass);
                    }
                } else if (truthCallAllele.isReference()) {
                    Assert.assertEquals(evalClass, EvaluationClass.FALSE_POSITIVE.acronym);
                }
            } else {
                Assert.assertEquals(Allele.NO_CALL, outputGenotype.getAllele(0));
                if (sampleCalls.isEmpty()) {
                    Assert.assertEquals(evalClass, !truthCallAllele.isReference() && truthCallAllele.isCalled() ? EvaluationClass.FALSE_NEGATIVE.acronym : null);
                    Assert.assertEquals(GATKProtectedVariantContextUtils.getAttributeAsDouble(outputGenotype, VariantEvaluationContext.CALL_QUALITY_KEY, -1), 0.0);
                } else {
                    Assert.assertEquals(evalClass, !truthCallAllele.isReference() && truthCallAllele.isCalled() ? EvaluationClass.MIXED_POSITIVE.acronym : EvaluationClass.FALSE_POSITIVE.acronym);
                    final Pair<VariantContext, Genotype> bestCall = sampleCalls.stream().sorted((p1, p2) -> -Double.compare(callGQ(p1.getRight()), callGQ(p2.getRight()))).findFirst().get();
                    Assert.assertEquals(GATKProtectedVariantContextUtils.getAttributeAsDouble(outputGenotype, VariantEvaluationContext.CALL_QUALITY_KEY, -1), callGQ(bestCall.getRight()));
                }
            }
        }
        final int expectedAN = (int) MathUtils.sum(sampleCallsCount);
        final int observedAN = matchingOutput.getAttributeAsInt(VariantEvaluationContext.CALLS_ALLELE_NUMBER_KEY, -1);
        Assert.assertEquals(observedAN, expectedAN);
        final double[] expectedAF = Arrays.copyOfRange(IntStream.of(sampleCallsCount).mapToDouble(i -> expectedAN > 0 ? i / (double) expectedAN : 0.0).toArray(), 1, sampleCallsCount.length);
        final double[] observedAF = GATKProtectedVariantContextUtils.getAttributeAsDoubleArray(matchingOutput, VariantEvaluationContext.CALLS_ALLELE_FREQUENCY_KEY, () -> new double[matchingOutput.getAlternateAlleles().size()], 0.0);
        Assert.assertNotNull(observedAF);
        assertEquals(observedAF, expectedAF, 0.01);
    }
}
Also used : Genotype(htsjdk.variant.variantcontext.Genotype) IntStream(java.util.stream.IntStream) Allele(htsjdk.variant.variantcontext.Allele) java.util(java.util) DataProvider(org.testng.annotations.DataProvider) VCFFileReader(htsjdk.variant.vcf.VCFFileReader) Argument(org.broadinstitute.barclay.argparser.Argument) StandardArgumentDefinitions(org.broadinstitute.hellbender.cmdline.StandardArgumentDefinitions) Test(org.testng.annotations.Test) TargetArgumentCollection(org.broadinstitute.hellbender.tools.exome.TargetArgumentCollection) Pair(org.apache.commons.lang3.tuple.Pair) Assert(org.testng.Assert) GATKProtectedVariantContextUtils(org.broadinstitute.hellbender.utils.GATKProtectedVariantContextUtils) TargetCollection(org.broadinstitute.hellbender.tools.exome.TargetCollection) VCFConstants(htsjdk.variant.vcf.VCFConstants) IOException(java.io.IOException) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Field(java.lang.reflect.Field) CommandLineProgramTest(org.broadinstitute.hellbender.CommandLineProgramTest) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) Collectors(java.util.stream.Collectors) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) File(java.io.File) XHMMSegmentGenotyper(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentGenotyper) MathUtils(org.broadinstitute.hellbender.utils.MathUtils) Target(org.broadinstitute.hellbender.tools.exome.Target) VariantContext(htsjdk.variant.variantcontext.VariantContext) VariantContext(htsjdk.variant.variantcontext.VariantContext) Genotype(htsjdk.variant.variantcontext.Genotype) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Target(org.broadinstitute.hellbender.tools.exome.Target) Allele(htsjdk.variant.variantcontext.Allele) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) Pair(org.apache.commons.lang3.tuple.Pair) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair)

Example 7 with CopyNumberTriStateAllele

use of org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele in project gatk by broadinstitute.

the class EvaluateCopyNumberTriStateCalls method buildAndAnnotateTruthOverlappingGenotype.

private Genotype buildAndAnnotateTruthOverlappingGenotype(final String sample, final TargetCollection<Target> targets, final Genotype truthGenotype, final int truthCopyNumber, final CopyNumberTriStateAllele truthAllele, final List<Pair<VariantContext, Genotype>> calls) {
    final Set<CopyNumberTriStateAllele> calledAlleles = calls.stream().map(pair -> CopyNumberTriStateAllele.valueOf(pair.getRight().getAllele(0))).collect(Collectors.toSet());
    final Allele calledAllele = calledAlleles.size() == 1 ? calledAlleles.iterator().next() : Allele.NO_CALL;
    final GenotypeBuilder builder = new GenotypeBuilder(sample);
    // Set the call allele.
    builder.alleles(Collections.singletonList(calledAllele));
    // Set the truth allele.
    builder.attribute(VariantEvaluationContext.TRUTH_GENOTYPE_KEY, CopyNumberTriStateAllele.ALL_ALLELES.indexOf(truthAllele));
    // Annotate the genotype with the number of calls.
    builder.attribute(VariantEvaluationContext.CALLED_SEGMENTS_COUNT_KEY, calls.size());
    // When there is more than one qualified type of event we indicate how many.
    builder.attribute(VariantEvaluationContext.CALLED_ALLELE_COUNTS_KEY, CopyNumberTriStateAllele.ALL_ALLELES.stream().mapToInt(allele -> (int) calls.stream().filter(pair -> pair.getRight().getAllele(0).equals(allele, true)).count()).toArray());
    // Calculate the length in targets of the call as the sum across all calls.
    builder.attribute(VariantEvaluationContext.CALLED_TARGET_COUNT_KEY, calls.stream().mapToInt(pair -> getTargetCount(targets, pair.getLeft(), pair.getRight())).sum());
    // Calculate call quality-- if there is more than one overlapping call we take the maximum qual one.
    builder.attribute(VariantEvaluationContext.CALL_QUALITY_KEY, calls.stream().mapToDouble(pair -> GATKProtectedVariantContextUtils.calculateGenotypeQualityFromPLs(pair.getRight())).max().orElse(0.0));
    // Calculate the truth copy fraction.
    builder.attribute(VariantEvaluationContext.TRUTH_COPY_FRACTION_KEY, truthGenotype.getExtendedAttribute(GS_COPY_NUMBER_FRACTION_KEY));
    // Calculate the truth call quality.
    final double truthQuality = calculateTruthQuality(truthGenotype, truthCopyNumber);
    builder.attribute(VariantEvaluationContext.TRUTH_QUALITY_KEY, truthQuality);
    // Set genotype filters:
    final boolean truthPassQualityMinimum = truthQuality >= filterArguments.minimumTruthSegmentQuality;
    builder.filter(truthPassQualityMinimum ? EvaluationFilter.PASS : EvaluationFilter.LowQuality.acronym);
    // Calculate the evaluation class (TP, FN, etc.). Only if there is actually either a truth or a call that is not ref.
    if (calledAlleles.contains(CopyNumberTriStateAllele.DEL) || calledAlleles.contains(CopyNumberTriStateAllele.DUP) || truthAllele != CopyNumberTriStateAllele.REF) {
        final EvaluationClass evaluationClass;
        if (calledAlleles.isEmpty() || (calledAlleles.size() == 1 && calledAlleles.contains(CopyNumberTriStateAllele.REF))) {
            evaluationClass = EvaluationClass.FALSE_NEGATIVE;
        } else if (calledAlleles.size() == 1) {
            evaluationClass = calledAlleles.contains(truthAllele) ? EvaluationClass.TRUE_POSITIVE : truthAllele == CopyNumberTriStateAllele.REF ? EvaluationClass.FALSE_POSITIVE : /* else */
            EvaluationClass.DISCORDANT_POSITIVE;
        } else {
            evaluationClass = truthAllele == CopyNumberTriStateAllele.REF ? EvaluationClass.FALSE_POSITIVE : EvaluationClass.MIXED_POSITIVE;
        }
        builder.attribute(VariantEvaluationContext.EVALUATION_CLASS_KEY, evaluationClass.acronym);
    }
    return builder.make();
}
Also used : Genotype(htsjdk.variant.variantcontext.Genotype) DocumentedFeature(org.broadinstitute.barclay.help.DocumentedFeature) Allele(htsjdk.variant.variantcontext.Allele) htsjdk.variant.vcf(htsjdk.variant.vcf) CommandLineProgramProperties(org.broadinstitute.barclay.argparser.CommandLineProgramProperties) java.util(java.util) CopyNumberProgramGroup(org.broadinstitute.hellbender.cmdline.programgroups.CopyNumberProgramGroup) Argument(org.broadinstitute.barclay.argparser.Argument) VariantContextWriterBuilder(htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder) StandardArgumentDefinitions(org.broadinstitute.hellbender.cmdline.StandardArgumentDefinitions) ArgumentCollection(org.broadinstitute.barclay.argparser.ArgumentCollection) TargetArgumentCollection(org.broadinstitute.hellbender.tools.exome.TargetArgumentCollection) Function(java.util.function.Function) Pair(org.apache.commons.lang3.tuple.Pair) StreamSupport(java.util.stream.StreamSupport) TargetCollection(org.broadinstitute.hellbender.tools.exome.TargetCollection) org.broadinstitute.hellbender.utils(org.broadinstitute.hellbender.utils) Locatable(htsjdk.samtools.util.Locatable) CommandLineProgram(org.broadinstitute.hellbender.cmdline.CommandLineProgram) GenotypeBuilder(htsjdk.variant.variantcontext.GenotypeBuilder) IOException(java.io.IOException) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Collectors(java.util.stream.Collectors) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) File(java.io.File) HMMPostProcessor(org.broadinstitute.hellbender.utils.hmm.segmentation.HMMPostProcessor) XHMMSegmentGenotyper(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentGenotyper) Stream(java.util.stream.Stream) UserException(org.broadinstitute.hellbender.exceptions.UserException) VariantContextWriter(htsjdk.variant.variantcontext.writer.VariantContextWriter) Target(org.broadinstitute.hellbender.tools.exome.Target) XHMMSegmentCaller(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentCaller) VariantContext(htsjdk.variant.variantcontext.VariantContext) BufferedReader(java.io.BufferedReader) FileReader(java.io.FileReader) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Allele(htsjdk.variant.variantcontext.Allele) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) GenotypeBuilder(htsjdk.variant.variantcontext.GenotypeBuilder)

Example 8 with CopyNumberTriStateAllele

use of org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele in project gatk by broadinstitute.

the class EvaluateCopyNumberTriStateCalls method buildAndAnnotateTruthOverlappingGenotype.

private Genotype buildAndAnnotateTruthOverlappingGenotype(final String sample, final VariantContext truth, final List<VariantContext> calls, final TargetCollection<Target> targets) {
    final Genotype truthGenotype = truth.getGenotype(sample);
    // if there is no truth genotype for that sample, we output the "empty" genotype.
    if (truthGenotype == null) {
        return GenotypeBuilder.create(sample, Collections.emptyList());
    }
    final int truthCopyNumber = GATKProtectedVariantContextUtils.getAttributeAsInt(truthGenotype, GS_COPY_NUMBER_FORMAT_KEY, truthNeutralCopyNumber);
    final CopyNumberTriStateAllele truthAllele = copyNumberToTrueAllele(truthCopyNumber);
    final List<Pair<VariantContext, Genotype>> allCalls = calls.stream().map(vc -> new ImmutablePair<>(vc, vc.getGenotype(sample))).filter(pair -> pair.getRight() != null).filter(pair -> GATKProtectedVariantContextUtils.getAttributeAsString(pair.getRight(), XHMMSegmentGenotyper.DISCOVERY_KEY, XHMMSegmentGenotyper.DISCOVERY_FALSE).equals(XHMMSegmentGenotyper.DISCOVERY_TRUE)).collect(Collectors.toList());
    final List<Pair<VariantContext, Genotype>> qualifiedCalls = composeQualifyingCallsList(targets, allCalls);
    return buildAndAnnotateTruthOverlappingGenotype(sample, targets, truthGenotype, truthCopyNumber, truthAllele, qualifiedCalls);
}
Also used : Genotype(htsjdk.variant.variantcontext.Genotype) DocumentedFeature(org.broadinstitute.barclay.help.DocumentedFeature) Allele(htsjdk.variant.variantcontext.Allele) htsjdk.variant.vcf(htsjdk.variant.vcf) CommandLineProgramProperties(org.broadinstitute.barclay.argparser.CommandLineProgramProperties) java.util(java.util) CopyNumberProgramGroup(org.broadinstitute.hellbender.cmdline.programgroups.CopyNumberProgramGroup) Argument(org.broadinstitute.barclay.argparser.Argument) VariantContextWriterBuilder(htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder) StandardArgumentDefinitions(org.broadinstitute.hellbender.cmdline.StandardArgumentDefinitions) ArgumentCollection(org.broadinstitute.barclay.argparser.ArgumentCollection) TargetArgumentCollection(org.broadinstitute.hellbender.tools.exome.TargetArgumentCollection) Function(java.util.function.Function) Pair(org.apache.commons.lang3.tuple.Pair) StreamSupport(java.util.stream.StreamSupport) TargetCollection(org.broadinstitute.hellbender.tools.exome.TargetCollection) org.broadinstitute.hellbender.utils(org.broadinstitute.hellbender.utils) Locatable(htsjdk.samtools.util.Locatable) CommandLineProgram(org.broadinstitute.hellbender.cmdline.CommandLineProgram) GenotypeBuilder(htsjdk.variant.variantcontext.GenotypeBuilder) IOException(java.io.IOException) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Collectors(java.util.stream.Collectors) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) File(java.io.File) HMMPostProcessor(org.broadinstitute.hellbender.utils.hmm.segmentation.HMMPostProcessor) XHMMSegmentGenotyper(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentGenotyper) Stream(java.util.stream.Stream) UserException(org.broadinstitute.hellbender.exceptions.UserException) VariantContextWriter(htsjdk.variant.variantcontext.writer.VariantContextWriter) Target(org.broadinstitute.hellbender.tools.exome.Target) XHMMSegmentCaller(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentCaller) VariantContext(htsjdk.variant.variantcontext.VariantContext) BufferedReader(java.io.BufferedReader) FileReader(java.io.FileReader) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) Genotype(htsjdk.variant.variantcontext.Genotype) Pair(org.apache.commons.lang3.tuple.Pair) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair)

Example 9 with CopyNumberTriStateAllele

use of org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele in project gatk by broadinstitute.

the class EvaluateCopyNumberTriStateCallsIntegrationTest method checkOutputTruthConcordance.

private void checkOutputTruthConcordance(final File truthFile, final File targetsFile, final File vcfOutput, final EvaluationFiltersArgumentCollection filteringOptions) {
    final List<VariantContext> truthVariants = readVCFFile(truthFile);
    final List<VariantContext> outputVariants = readVCFFile(vcfOutput);
    final Set<String> outputSamples = outputVariants.get(0).getSampleNames();
    final TargetCollection<Target> targets = TargetArgumentCollection.readTargetCollection(targetsFile);
    for (final VariantContext truth : truthVariants) {
        final List<Target> overlappingTargets = targets.targets(truth);
        final List<VariantContext> overlappingOutput = outputVariants.stream().filter(vc -> new SimpleInterval(vc).overlaps(truth)).collect(Collectors.toList());
        if (overlappingTargets.isEmpty()) {
            Assert.assertTrue(overlappingOutput.isEmpty());
            continue;
        }
        Assert.assertFalse(overlappingOutput.isEmpty());
        Assert.assertEquals(overlappingOutput.stream().filter(vc -> new SimpleInterval(truth).equals(new SimpleInterval(vc))).count(), 1);
        @SuppressWarnings("all") final Optional<VariantContext> prospectiveMatchingOutput = overlappingOutput.stream().filter(vc -> new SimpleInterval(truth).equals(new SimpleInterval(vc))).findFirst();
        Assert.assertTrue(prospectiveMatchingOutput.isPresent());
        final VariantContext matchingOutput = prospectiveMatchingOutput.get();
        final int[] truthAC = calculateACFromTruth(truth);
        final long truthAN = MathUtils.sum(truthAC);
        final double[] truthAF = IntStream.of(Arrays.copyOfRange(truthAC, 1, truthAC.length)).mapToDouble(d -> d / (double) truthAN).toArray();
        Assert.assertEquals(matchingOutput.getAttributeAsInt(VariantEvaluationContext.TRUTH_ALLELE_NUMBER_KEY, -1), truthAN);
        assertEquals(GATKProtectedVariantContextUtils.getAttributeAsDoubleArray(matchingOutput, VariantEvaluationContext.TRUTH_ALLELE_FREQUENCY_KEY, () -> new double[2], 0.0), truthAF, 0.001);
        assertOutputVariantFilters(filteringOptions, overlappingTargets, matchingOutput, truthAF);
        for (final String sample : outputSamples) {
            final Genotype outputGenotype = matchingOutput.getGenotype(sample);
            final Genotype truthGenotype = truth.getGenotype(sample);
            final int truthCN = GATKProtectedVariantContextUtils.getAttributeAsInt(truthGenotype, ConvertGSVariantsToSegments.GS_COPY_NUMBER_FORMAT, -1);
            final int truthGT = GATKProtectedVariantContextUtils.getAttributeAsInt(outputGenotype, VariantEvaluationContext.TRUTH_GENOTYPE_KEY, -1);
            final Object truthQualObject = outputGenotype.getAnyAttribute(VariantEvaluationContext.TRUTH_QUALITY_KEY);
            Assert.assertNotNull(truthQualObject, "" + truthGenotype);
            final double truthQual = Double.parseDouble(String.valueOf(truthQualObject));
            if (truthQual < filteringOptions.minimumTruthSegmentQuality) {
                Assert.assertEquals(outputGenotype.getFilters(), EvaluationFilter.LowQuality.acronym);
            } else {
                Assert.assertTrue(outputGenotype.getFilters() == null || outputGenotype.getFilters().equals(VCFConstants.PASSES_FILTERS_v4));
            }
            final double[] truthPosteriors = GATKProtectedVariantContextUtils.getAttributeAsDoubleArray(truthGenotype, ConvertGSVariantsToSegments.GS_COPY_NUMBER_POSTERIOR, () -> new double[0], Double.NEGATIVE_INFINITY);
            final double truthDelPosterior = MathUtils.log10SumLog10(truthPosteriors, 0, EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT);
            final double truthRefPosterior = truthPosteriors[EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT];
            final double truthDupPosterior = truthPosteriors.length < EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT ? Double.NEGATIVE_INFINITY : MathUtils.log10SumLog10(truthPosteriors, EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT + 1, truthPosteriors.length);
            final CopyNumberTriStateAllele truthAllele = truthGT == -1 ? null : CopyNumberTriStateAllele.ALL_ALLELES.get(truthGT);
            if (truthCN < EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT) {
                Assert.assertEquals(truthAllele, CopyNumberTriStateAllele.DEL);
                Assert.assertEquals(truthQual * -.1, MathUtils.log10SumLog10(new double[] { truthRefPosterior, truthDupPosterior }), 0.01);
            } else if (truthCN > EvaluateCopyNumberTriStateCalls.REFERENCE_COPY_NUMBER_DEFAULT) {
                Assert.assertEquals(truthAllele, CopyNumberTriStateAllele.DUP);
                Assert.assertEquals(truthQual * -.1, MathUtils.log10SumLog10(new double[] { truthDelPosterior, truthRefPosterior }), 0.01, "" + truthGenotype + " " + outputGenotype);
            } else {
                Assert.assertEquals(truthAllele, CopyNumberTriStateAllele.REF);
                Assert.assertEquals(truthQual * -.1, MathUtils.log10SumLog10(new double[] { truthDelPosterior, truthDupPosterior }), 0.01);
            }
            final double outputTruthFraction = GATKProtectedVariantContextUtils.getAttributeAsDouble(outputGenotype, VariantEvaluationContext.TRUTH_COPY_FRACTION_KEY, -1);
            final double inputTruthFraction = GATKProtectedVariantContextUtils.getAttributeAsDouble(truthGenotype, ConvertGSVariantsToSegments.GS_COPY_NUMBER_FRACTION, -1);
            Assert.assertEquals(outputTruthFraction, inputTruthFraction, 0.01);
        }
    }
}
Also used : Genotype(htsjdk.variant.variantcontext.Genotype) IntStream(java.util.stream.IntStream) Allele(htsjdk.variant.variantcontext.Allele) java.util(java.util) DataProvider(org.testng.annotations.DataProvider) VCFFileReader(htsjdk.variant.vcf.VCFFileReader) Argument(org.broadinstitute.barclay.argparser.Argument) StandardArgumentDefinitions(org.broadinstitute.hellbender.cmdline.StandardArgumentDefinitions) Test(org.testng.annotations.Test) TargetArgumentCollection(org.broadinstitute.hellbender.tools.exome.TargetArgumentCollection) Pair(org.apache.commons.lang3.tuple.Pair) Assert(org.testng.Assert) GATKProtectedVariantContextUtils(org.broadinstitute.hellbender.utils.GATKProtectedVariantContextUtils) TargetCollection(org.broadinstitute.hellbender.tools.exome.TargetCollection) VCFConstants(htsjdk.variant.vcf.VCFConstants) IOException(java.io.IOException) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Field(java.lang.reflect.Field) CommandLineProgramTest(org.broadinstitute.hellbender.CommandLineProgramTest) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) Collectors(java.util.stream.Collectors) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) File(java.io.File) XHMMSegmentGenotyper(org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentGenotyper) MathUtils(org.broadinstitute.hellbender.utils.MathUtils) Target(org.broadinstitute.hellbender.tools.exome.Target) VariantContext(htsjdk.variant.variantcontext.VariantContext) VariantContext(htsjdk.variant.variantcontext.VariantContext) Genotype(htsjdk.variant.variantcontext.Genotype) CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele) Target(org.broadinstitute.hellbender.tools.exome.Target) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval)

Example 10 with CopyNumberTriStateAllele

use of org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele in project gatk by broadinstitute.

the class VariantEvaluationContextBuilder method validateAlleleStrings.

private void validateAlleleStrings(final List<String> allelesString) {
    Utils.containsNoNull(allelesString, "the input cannot contain null strings");
    Utils.nonEmpty(allelesString, "the allele list cannot be empty");
    Utils.validateArg(allelesString.get(0).equals(CopyNumberTriStateAllele.REF.getDisplayString()), "the first allele must be the reference allele: " + CopyNumberTriStateAllele.REF);
    final Set<CopyNumberTriStateAllele> allelesFound = new HashSet<>();
    allelesFound.add(CopyNumberTriStateAllele.REF);
    for (int i = 1; i < allelesString.size(); i++) {
        final String alleleString = allelesString.get(i);
        final CopyNumberTriStateAllele allele;
        try {
            allele = CopyNumberTriStateAllele.valueOf(Allele.create(alleleString, false));
        } catch (final IllegalArgumentException ex) {
            throw new IllegalArgumentException(String.format("unknown allele with string: '%s'", alleleString));
        }
        if (!allelesFound.add(allele)) {
            throw new IllegalArgumentException("an allele cannot be listed twice.");
        }
    }
}
Also used : CopyNumberTriStateAllele(org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele)

Aggregations

CopyNumberTriStateAllele (org.broadinstitute.hellbender.tools.exome.germlinehmm.CopyNumberTriStateAllele)10 Allele (htsjdk.variant.variantcontext.Allele)8 Genotype (htsjdk.variant.variantcontext.Genotype)8 VariantContext (htsjdk.variant.variantcontext.VariantContext)8 File (java.io.File)8 IOException (java.io.IOException)8 java.util (java.util)8 Collectors (java.util.stream.Collectors)8 ImmutablePair (org.apache.commons.lang3.tuple.ImmutablePair)8 Pair (org.apache.commons.lang3.tuple.Pair)8 Argument (org.broadinstitute.barclay.argparser.Argument)8 StandardArgumentDefinitions (org.broadinstitute.hellbender.cmdline.StandardArgumentDefinitions)8 Target (org.broadinstitute.hellbender.tools.exome.Target)8 TargetArgumentCollection (org.broadinstitute.hellbender.tools.exome.TargetArgumentCollection)8 TargetCollection (org.broadinstitute.hellbender.tools.exome.TargetCollection)8 XHMMSegmentGenotyper (org.broadinstitute.hellbender.tools.exome.germlinehmm.xhmm.XHMMSegmentGenotyper)8 Locatable (htsjdk.samtools.util.Locatable)4 GenotypeBuilder (htsjdk.variant.variantcontext.GenotypeBuilder)4 VariantContextWriter (htsjdk.variant.variantcontext.writer.VariantContextWriter)4 VariantContextWriterBuilder (htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder)4