use of org.broadinstitute.hellbender.tools.exome.sexgenotyper.SexGenotypeDataCollection in project gatk-protected by broadinstitute.
the class GermlineCNVCallerIntegrationTest method reportCopyNumberSummaryStatistics.
/* Shame on me for using {@link ReadCountCollection} to store copy numbers! */
private void reportCopyNumberSummaryStatistics(@Nonnull final File posteriorsOutputPath, @Nonnull final File truthCopyNumberFile, @Nonnull final List<Target> targets, @Nonnull final SexGenotypeDataCollection sexGenotypeDataCollection) {
final ReadCountCollection truthCopyNumberCollection = loadTruthCopyNumberTable(truthCopyNumberFile, targets);
final RealMatrix calledCopyNumberMatrix = Nd4jApacheAdapterUtils.convertINDArrayToApacheMatrix(Nd4jIOUtils.readNDArrayMatrixFromTextFile(new File(posteriorsOutputPath, CoverageModelGlobalConstants.COPY_RATIO_VITERBI_FILENAME)));
final ReadCountCollection calledCopyNumberCollection = new ReadCountCollection(targets, truthCopyNumberCollection.columnNames(), calledCopyNumberMatrix);
final int numSamples = calledCopyNumberCollection.columnNames().size();
final List<String> sampleSexGenotypes = truthCopyNumberCollection.columnNames().stream().map(sampleName -> sexGenotypeDataCollection.getSampleSexGenotypeData(sampleName).getSexGenotype()).collect(Collectors.toList());
final List<SampleCopyNumberSummaryStatistics> sampleSummaryStatisticsList = IntStream.range(0, numSamples).mapToObj(si -> calculateSampleCopyNumberConcordance(truthCopyNumberCollection, calledCopyNumberCollection, si, sampleSexGenotypes.get(si))).collect(Collectors.toList());
/* calculation various summary statistics */
final AbstractUnivariateStatistic calculator = new Mean();
final ConfusionRates homDelMedianRates = ConfusionMatrix.getConfusionRates(sampleSummaryStatisticsList.stream().map(ss -> ss.homozygousDeletionConfusionMatrix).collect(Collectors.toList()), calculator);
final ConfusionRates hetDelMedianRates = ConfusionMatrix.getConfusionRates(sampleSummaryStatisticsList.stream().map(ss -> ss.heterozygousDeletionConfusionMatrix).collect(Collectors.toList()), calculator);
final ConfusionRates dupMedianRates = ConfusionMatrix.getConfusionRates(sampleSummaryStatisticsList.stream().map(ss -> ss.duplicationConfusionMatrix).collect(Collectors.toList()), calculator);
final double absoluteConcordance = Concordance.getCollectionConcordance(sampleSummaryStatisticsList.stream().map(ss -> ss.absoluteCopyNumberConcordance).collect(Collectors.toList()), calculator);
/* log */
logger.info("Homozygous deletion statistics: " + homDelMedianRates.toString());
logger.info("Heterozygous deletion statistics: " + hetDelMedianRates.toString());
logger.info("Duplication statistics: " + dupMedianRates.toString());
logger.info(String.format("Absolute copy number calling concordance: %f", absoluteConcordance));
}
Aggregations