Search in sources :

Example 1 with FittedRegion

use of com.hartwig.hmftools.common.purple.region.FittedRegion in project hmftools by hartwigmedical.

the class LoadSomaticVariants method main.

public static void main(@NotNull final String[] args) throws ParseException, IOException, SQLException {
    final Options options = createBasicOptions();
    final CommandLine cmd = createCommandLine(args, options);
    final String vcfFileLocation = cmd.getOptionValue(VCF_FILE);
    final String highConfidenceBed = cmd.getOptionValue(HIGH_CONFIDENCE_BED);
    final String fastaFileLocation = cmd.getOptionValue(REF_GENOME);
    final String sample = cmd.getOptionValue(SAMPLE);
    final DatabaseAccess dbAccess = databaseAccess(cmd);
    final CompoundFilter filter = new CompoundFilter(true);
    if (cmd.hasOption(PASS_FILTER)) {
        filter.add(new PassingVariantFilter());
    }
    if (cmd.hasOption(SOMATIC_FILTER)) {
        filter.add(new SomaticFilter());
    }
    LOGGER.info("Reading somatic VCF File");
    final List<SomaticVariant> variants = SomaticVariantFactory.filteredInstance(filter).fromVCFFile(sample, vcfFileLocation);
    LOGGER.info("Reading high confidence bed file");
    final Multimap<String, GenomeRegion> highConfidenceRegions = BEDFileLoader.fromBedFile(highConfidenceBed);
    LOGGER.info("Loading indexed fasta reference file");
    IndexedFastaSequenceFile indexedFastaSequenceFile = new IndexedFastaSequenceFile(new File(fastaFileLocation));
    LOGGER.info("Querying purple database");
    final PurityContext purityContext = dbAccess.readPurityContext(sample);
    if (purityContext == null) {
        LOGGER.warn("Unable to retrieve purple data. Enrichment may be incomplete.");
    }
    final PurityAdjuster purityAdjuster = purityContext == null ? new PurityAdjuster(Gender.FEMALE, 1, 1) : new PurityAdjuster(purityContext.gender(), purityContext.bestFit().purity(), purityContext.bestFit().normFactor());
    final Multimap<String, PurpleCopyNumber> copyNumbers = Multimaps.index(dbAccess.readCopynumbers(sample), PurpleCopyNumber::chromosome);
    final Multimap<String, FittedRegion> copyNumberRegions = Multimaps.index(dbAccess.readCopyNumberRegions(sample), FittedRegion::chromosome);
    LOGGER.info("Incorporating purple purity");
    final PurityAdjustedSomaticVariantFactory purityAdjustmentFactory = new PurityAdjustedSomaticVariantFactory(purityAdjuster, copyNumbers, copyNumberRegions);
    final List<PurityAdjustedSomaticVariant> purityAdjustedVariants = purityAdjustmentFactory.create(variants);
    final double clonalPloidy = ClonalityCutoffKernel.clonalCutoff(purityAdjustedVariants);
    LOGGER.info("Enriching variants");
    final EnrichedSomaticVariantFactory enrichedSomaticVariantFactory = new EnrichedSomaticVariantFactory(highConfidenceRegions, indexedFastaSequenceFile, new ClonalityFactory(purityAdjuster, clonalPloidy), CanonicalTranscriptFactory.create(HmfGenePanelSupplier.allGeneList()));
    final List<EnrichedSomaticVariant> enrichedVariants = enrichedSomaticVariantFactory.enrich(purityAdjustedVariants);
    LOGGER.info("Persisting variants to database");
    dbAccess.writeSomaticVariants(sample, enrichedVariants);
    LOGGER.info("Complete");
}
Also used : Options(org.apache.commons.cli.Options) EnrichedSomaticVariant(com.hartwig.hmftools.common.variant.EnrichedSomaticVariant) PassingVariantFilter(htsjdk.variant.variantcontext.filter.PassingVariantFilter) FittedRegion(com.hartwig.hmftools.common.purple.region.FittedRegion) PurpleCopyNumber(com.hartwig.hmftools.common.purple.copynumber.PurpleCopyNumber) IndexedFastaSequenceFile(htsjdk.samtools.reference.IndexedFastaSequenceFile) DatabaseAccess(com.hartwig.hmftools.patientdb.dao.DatabaseAccess) SomaticFilter(com.hartwig.hmftools.common.variant.filter.SomaticFilter) PurityAdjuster(com.hartwig.hmftools.common.purple.PurityAdjuster) PurityAdjustedSomaticVariant(com.hartwig.hmftools.common.variant.PurityAdjustedSomaticVariant) SomaticVariant(com.hartwig.hmftools.common.variant.SomaticVariant) EnrichedSomaticVariant(com.hartwig.hmftools.common.variant.EnrichedSomaticVariant) PurityAdjustedSomaticVariantFactory(com.hartwig.hmftools.common.variant.PurityAdjustedSomaticVariantFactory) CompoundFilter(htsjdk.variant.variantcontext.filter.CompoundFilter) PurityAdjustedSomaticVariant(com.hartwig.hmftools.common.variant.PurityAdjustedSomaticVariant) CommandLine(org.apache.commons.cli.CommandLine) GenomeRegion(com.hartwig.hmftools.common.region.GenomeRegion) PurityContext(com.hartwig.hmftools.common.purple.purity.PurityContext) EnrichedSomaticVariantFactory(com.hartwig.hmftools.common.variant.EnrichedSomaticVariantFactory) ClonalityFactory(com.hartwig.hmftools.common.variant.ClonalityFactory) File(java.io.File) IndexedFastaSequenceFile(htsjdk.samtools.reference.IndexedFastaSequenceFile)

Example 2 with FittedRegion

use of com.hartwig.hmftools.common.purple.region.FittedRegion in project hmftools by hartwigmedical.

the class CopyNumberDAO method readCopyNumberRegions.

@NotNull
List<FittedRegion> readCopyNumberRegions(@NotNull final String sample) {
    List<FittedRegion> fittedRegions = Lists.newArrayList();
    Result<Record> result = context.select().from(COPYNUMBERREGION).where(COPYNUMBERREGION.SAMPLEID.eq(sample)).fetch();
    for (Record record : result) {
        fittedRegions.add(ImmutableFittedRegion.builder().chromosome(record.getValue(COPYNUMBERREGION.CHROMOSOME)).start(record.getValue(COPYNUMBERREGION.START)).end(record.getValue(COPYNUMBERREGION.END)).status(GermlineStatus.valueOf(record.getValue(COPYNUMBERREGION.GERMLINESTATUS))).svCluster(record.getValue(COPYNUMBERREGION.SVCLUSTER) != 0).ratioSupport(record.getValue(COPYNUMBERREGION.RATIOSUPPORT) != 0).support(SegmentSupport.valueOf(record.getValue(COPYNUMBERREGION.SEGMENTSTARTSUPPORT))).bafCount(record.getValue(COPYNUMBERREGION.BAFCOUNT)).observedBAF(record.getValue(COPYNUMBERREGION.OBSERVEDBAF)).observedTumorRatio(record.getValue(COPYNUMBERREGION.OBSERVEDTUMORRATIO)).observedNormalRatio(record.getValue(COPYNUMBERREGION.OBSERVEDNORMALRATIO)).observedTumorRatioCount(record.getValue(COPYNUMBERREGION.OBSERVEDTUMORRATIOCOUNT)).gcContent(record.getValue(COPYNUMBERREGION.GCCONTENT)).modelPloidy(record.getValue(COPYNUMBERREGION.MODELPLOIDY)).modelBAF(record.getValue(COPYNUMBERREGION.MODELBAF)).modelTumorRatio(record.getValue(COPYNUMBERREGION.MODELTUMORRATIO)).tumorBAF(record.getValue(COPYNUMBERREGION.ACTUALTUMORBAF)).tumorCopyNumber(record.getValue(COPYNUMBERREGION.ACTUALTUMORCOPYNUMBER)).refNormalisedCopyNumber(record.getValue(COPYNUMBERREGION.REFNORMALISEDTUMORCOPYNUMBER)).cnvDeviation(record.getValue(COPYNUMBERREGION.CNVDEVIATION)).bafDeviation(record.getValue(COPYNUMBERREGION.BAFDEVIATION)).deviation(record.getValue(COPYNUMBERREGION.TOTALDEVIATION)).segmentBAF(record.getValue(COPYNUMBERREGION.FITTEDBAF)).segmentTumorCopyNumber(record.getValue(COPYNUMBERREGION.FITTEDCOPYNUMBER)).ploidyPenalty(record.getValue(COPYNUMBERREGION.PLOIDYPENALTY)).build());
    }
    Collections.sort(fittedRegions);
    return fittedRegions;
}
Also used : Record(org.jooq.Record) ImmutableFittedRegion(com.hartwig.hmftools.common.purple.region.ImmutableFittedRegion) FittedRegion(com.hartwig.hmftools.common.purple.region.FittedRegion) NotNull(org.jetbrains.annotations.NotNull)

Example 3 with FittedRegion

use of com.hartwig.hmftools.common.purple.region.FittedRegion in project hmftools by hartwigmedical.

the class CopyNumberDAO method writeCopyNumberRegions.

void writeCopyNumberRegions(@NotNull final String sample, @NotNull List<FittedRegion> regions) {
    Timestamp timestamp = new Timestamp(new Date().getTime());
    context.delete(COPYNUMBERREGION).where(COPYNUMBERREGION.SAMPLEID.eq(sample)).execute();
    for (List<FittedRegion> splitRegions : Iterables.partition(regions, DB_BATCH_INSERT_SIZE)) {
        InsertValuesStepN inserter = context.insertInto(COPYNUMBERREGION, COPYNUMBERREGION.SAMPLEID, COPYNUMBERREGION.CHROMOSOME, COPYNUMBERREGION.START, COPYNUMBERREGION.END, COPYNUMBERREGION.GERMLINESTATUS, COPYNUMBERREGION.SVCLUSTER, COPYNUMBERREGION.RATIOSUPPORT, COPYNUMBERREGION.SEGMENTSTARTSUPPORT, COPYNUMBERREGION.BAFCOUNT, COPYNUMBERREGION.OBSERVEDBAF, COPYNUMBERREGION.OBSERVEDTUMORRATIO, COPYNUMBERREGION.OBSERVEDNORMALRATIO, COPYNUMBERREGION.OBSERVEDTUMORRATIOCOUNT, COPYNUMBERREGION.GCCONTENT, COPYNUMBERREGION.MODELPLOIDY, COPYNUMBERREGION.MODELBAF, COPYNUMBERREGION.MODELTUMORRATIO, COPYNUMBERREGION.ACTUALTUMORBAF, COPYNUMBERREGION.ACTUALTUMORCOPYNUMBER, COPYNUMBERREGION.REFNORMALISEDTUMORCOPYNUMBER, COPYNUMBERREGION.CNVDEVIATION, COPYNUMBERREGION.BAFDEVIATION, COPYNUMBERREGION.TOTALDEVIATION, COPYNUMBERREGION.PLOIDYPENALTY, COPYNUMBERREGION.FITTEDBAF, COPYNUMBERREGION.FITTEDCOPYNUMBER, COPYNUMBERREGION.MODIFIED);
        splitRegions.forEach(x -> addCopynumberRecord(timestamp, inserter, sample, x));
        inserter.execute();
    }
}
Also used : InsertValuesStepN(org.jooq.InsertValuesStepN) ImmutableFittedRegion(com.hartwig.hmftools.common.purple.region.ImmutableFittedRegion) FittedRegion(com.hartwig.hmftools.common.purple.region.FittedRegion) Timestamp(java.sql.Timestamp) Date(java.util.Date)

Example 4 with FittedRegion

use of com.hartwig.hmftools.common.purple.region.FittedRegion in project hmftools by hartwigmedical.

the class CopyNumberDeviationTest method testMaxCopyNumberDeviation.

@Test
public void testMaxCopyNumberDeviation() {
    final FittedRegion firstWithSV = create(1, 1000, SegmentSupport.MULTIPLE);
    final FittedRegion firstWithoutSV = create(1, 1000, SegmentSupport.NONE);
    final FittedRegion secondWithSV = create(1001, 2000, SegmentSupport.MULTIPLE);
    final FittedRegion secondWithoutSV = create(1001, 2000, SegmentSupport.NONE);
    final FittedRegion thirdWithSV = create(2001, 3000, SegmentSupport.MULTIPLE);
    final FittedRegion thirdWithoutSV = create(2001, 3000, SegmentSupport.NONE);
    assertEquals(HC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithSV, firstWithSV), EPSILON);
    assertEquals(HC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithSV, firstWithoutSV), EPSILON);
    assertEquals(HC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithSV, thirdWithSV), EPSILON);
    assertEquals(LC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithSV, thirdWithoutSV), EPSILON);
    assertEquals(LC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithoutSV, firstWithSV), EPSILON);
    assertEquals(LC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithoutSV, firstWithoutSV), EPSILON);
    assertEquals(HC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithoutSV, thirdWithSV), EPSILON);
    assertEquals(LC_MAX_COPY_NUMBER_TOLERANCE, CopyNumberDeviation.maxCopyNumberDeviation(secondWithoutSV, thirdWithoutSV), EPSILON);
}
Also used : FittedRegion(com.hartwig.hmftools.common.purple.region.FittedRegion) PurpleDatamodelTest(com.hartwig.hmftools.common.purple.PurpleDatamodelTest) Test(org.junit.Test)

Example 5 with FittedRegion

use of com.hartwig.hmftools.common.purple.region.FittedRegion in project hmftools by hartwigmedical.

the class ExtendDiploidTest method testDubiousRegionGetsIncludedFromLeft.

@Test
public void testDubiousRegionGetsIncludedFromLeft() {
    final FittedRegion somaticLeft = createValidSomatic(1, 10000, 3.0, 50, SegmentSupport.BND);
    final FittedRegion dubious = createDubiousRegion(10001, 20000, 2.0, 10);
    final FittedRegion somaticRight = createValidSomatic(20001, 30000, 3.0, 40, SegmentSupport.NONE);
    final List<CombinedRegion> result = PURE_VICTIM.extendDiploid(Lists.newArrayList(somaticLeft, dubious, somaticRight));
    assertEquals(1, result.size());
    assertRegion(1, 30000, 3, result.get(0));
}
Also used : FittedRegion(com.hartwig.hmftools.common.purple.region.FittedRegion) PurpleDatamodelTest(com.hartwig.hmftools.common.purple.PurpleDatamodelTest) Test(org.junit.Test)

Aggregations

FittedRegion (com.hartwig.hmftools.common.purple.region.FittedRegion)25 PurpleDatamodelTest (com.hartwig.hmftools.common.purple.PurpleDatamodelTest)14 Test (org.junit.Test)14 NotNull (org.jetbrains.annotations.NotNull)6 ImmutableFittedRegion (com.hartwig.hmftools.common.purple.region.ImmutableFittedRegion)2 ModifiableFittedRegion (com.hartwig.hmftools.common.purple.region.ModifiableFittedRegion)2 PurityAdjuster (com.hartwig.hmftools.common.purple.PurityAdjuster)1 PurpleCopyNumber (com.hartwig.hmftools.common.purple.copynumber.PurpleCopyNumber)1 PurityContext (com.hartwig.hmftools.common.purple.purity.PurityContext)1 ObservedRegion (com.hartwig.hmftools.common.purple.region.ObservedRegion)1 GenomeRegion (com.hartwig.hmftools.common.region.GenomeRegion)1 ClonalityFactory (com.hartwig.hmftools.common.variant.ClonalityFactory)1 EnrichedSomaticVariant (com.hartwig.hmftools.common.variant.EnrichedSomaticVariant)1 EnrichedSomaticVariantFactory (com.hartwig.hmftools.common.variant.EnrichedSomaticVariantFactory)1 PurityAdjustedSomaticVariant (com.hartwig.hmftools.common.variant.PurityAdjustedSomaticVariant)1 PurityAdjustedSomaticVariantFactory (com.hartwig.hmftools.common.variant.PurityAdjustedSomaticVariantFactory)1 SomaticVariant (com.hartwig.hmftools.common.variant.SomaticVariant)1 SomaticFilter (com.hartwig.hmftools.common.variant.filter.SomaticFilter)1 DatabaseAccess (com.hartwig.hmftools.patientdb.dao.DatabaseAccess)1 IndexedFastaSequenceFile (htsjdk.samtools.reference.IndexedFastaSequenceFile)1