use of org.broadinstitute.hellbender.tools.exome.Target in project gatk-protected by broadinstitute.
the class SparkGenomeReadCountsIntegrationTest method testSparkGenomeReadCountsBigBins.
@Test
public void testSparkGenomeReadCountsBigBins() throws IOException {
final File outputFile = createTempFile(BAM_FILE.getName(), ".cov");
final String[] arguments = { "--disableSequenceDictionaryValidation", "-" + StandardArgumentDefinitions.REFERENCE_SHORT_NAME, REFERENCE_FILE.getAbsolutePath(), "-" + StandardArgumentDefinitions.INPUT_SHORT_NAME, BAM_FILE.getAbsolutePath(), "-" + SparkGenomeReadCounts.OUTPUT_FILE_SHORT_NAME, outputFile.getAbsolutePath(), "-" + SparkGenomeReadCounts.BINSIZE_SHORT_NAME, "16000" };
runCommandLine(arguments);
Assert.assertTrue(outputFile.exists());
Assert.assertTrue(outputFile.length() > 0);
final ReadCountCollection coverage = ReadCountCollectionUtils.parse(outputFile);
final File targetsFile = new File(outputFile.getAbsolutePath() + ".targets.tsv");
Assert.assertTrue(targetsFile.exists());
Assert.assertTrue(targetsFile.length() > 0);
final List<Target> targets = TargetTableReader.readTargetFile(targetsFile);
Assert.assertEquals(targets.size(), 4);
Assert.assertEquals(targets.get(1).getEnd(), 16000);
Assert.assertEquals(targets.get(2).getName(), "target_3_1_16000");
Assert.assertEquals(coverage.targets().size(), targets.size());
}
use of org.broadinstitute.hellbender.tools.exome.Target in project gatk-protected by broadinstitute.
the class SparkGenomeReadCountsIntegrationTest method testSparkGenomeReadCountsSubContig.
@Test
public void testSparkGenomeReadCountsSubContig() {
final File outputFile = createTempFile(BAM_FILE.getName(), ".cov");
final String[] arguments = { "--disableSequenceDictionaryValidation", "-" + StandardArgumentDefinitions.REFERENCE_SHORT_NAME, REFERENCE_FILE.getAbsolutePath(), "-" + StandardArgumentDefinitions.INPUT_SHORT_NAME, BAM_FILE.getAbsolutePath(), "-" + SparkGenomeReadCounts.OUTPUT_FILE_SHORT_NAME, outputFile.getAbsolutePath(), "-" + SparkGenomeReadCounts.BINSIZE_SHORT_NAME, "100", "-L", "1:1-500" };
runCommandLine(arguments);
final File targetsFile = new File(outputFile.getAbsolutePath() + ".targets.tsv");
final List<Target> targets = TargetTableReader.readTargetFile(targetsFile);
Assert.assertTrue(targets.stream().allMatch(t -> t.getContig().equals("1")));
Assert.assertEquals(targets.size(), 5);
for (int i = 0; i < targets.size(); i++) {
Assert.assertEquals(targets.get(i).getStart(), i * 100 + 1);
Assert.assertEquals(targets.get(i).getEnd(), (i + 1) * 100);
}
}
use of org.broadinstitute.hellbender.tools.exome.Target in project gatk 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));
}
use of org.broadinstitute.hellbender.tools.exome.Target in project gatk by broadinstitute.
the class GermlineCNVCallerIntegrationTest method runLearningAndCallingTest.
private void runLearningAndCallingTest(final String... extraArgs) {
runCommandLine(getLearningArgs(extraArgs));
final List<Target> modelledTargets = TargetTableReader.readTargetFile(new File(LEARNING_MODEL_OUTPUT_PATH, CoverageModelGlobalConstants.TARGET_LIST_OUTPUT_FILE));
reportCopyNumberSummaryStatistics(LEARNING_POSTERIORS_OUTPUT_PATH, TEST_LEARNING_COMBINED_COPY_NUMBER_FILE, modelledTargets, LEARNING_SEX_GENOTYPES_DATA);
logger.info("Copy number concordance test passed for simultaneous learning and calling");
}
use of org.broadinstitute.hellbender.tools.exome.Target in project gatk by broadinstitute.
the class GermlineCNVCallerIntegrationTest method runCaseSampleCallingTestOnExactModelParams.
private void runCaseSampleCallingTestOnExactModelParams(final String... extraArgs) {
runCommandLine(getCallingOnExactModelArgs(ArrayUtils.addAll(new String[] { "--" + GermlineCNVCaller.INPUT_MODEL_PATH_LONG_NAME, TEST_TRUTH_SIM_MODEL.getAbsolutePath() }, extraArgs)));
final List<Target> callingTargets = TargetTableReader.readTargetFile(new File(CALLING_POSTERIORS_OUTPUT_PATH, CoverageModelGlobalConstants.TARGET_LIST_OUTPUT_FILE));
reportCopyNumberSummaryStatistics(CALLING_POSTERIORS_OUTPUT_PATH, TEST_CALLING_COMBINED_COPY_NUMBER_FILE, callingTargets, CALLING_SEX_GENOTYPES_DATA);
logger.info("Copy number concordance test passed for case sample calling");
}
Aggregations