Examples with StructuralVariantAnalysis com.hartwig.hmftools.svannotation.analysis.StructuralVariantAnalysis used on opensource projects

Example 1 with StructuralVariantAnalysis

use of com.hartwig.hmftools.svannotation.analysis.StructuralVariantAnalysis in project hmftools by hartwigmedical.

the class PatientReporter method run.

@NotNull
public SequencedPatientReport run(@NotNull final String runDirectory, @Nullable final String comments) throws IOException {
    final RunContext run = ProductionRunContextFactory.fromRunDirectory(runDirectory);
    final GenomeAnalysis genomeAnalysis = analyseGenomeData(run.tumorSample(), runDirectory);
    assert run.isSomaticRun() && run.tumorSample().equals(genomeAnalysis.sample());
    final String tumorSample = genomeAnalysis.sample();
    final VariantAnalysis variantAnalysis = genomeAnalysis.variantAnalysis();
    final PurpleAnalysis purpleAnalysis = genomeAnalysis.purpleAnalysis();
    final StructuralVariantAnalysis structuralVariantAnalysis = genomeAnalysis.structuralVariantAnalysis();
    final List<GeneFusionData> reportableFusions = structuralVariantAnalysis.reportableFusions().stream().sorted(fusionComparator()).map(GeneFusionData::from).collect(Collectors.toList());
    final List<GeneDisruptionData> reportableDisruptions = structuralVariantAnalysis.reportableDisruptions().stream().sorted(disruptionComparator(reporterData().panelGeneModel().transcriptMap())).map(GeneDisruptionData::from).collect(Collectors.toList());
    final int passedVariantCount = variantAnalysis.passedVariants().size();
    final int mutationalLoad = variantAnalysis.mutationalLoad();
    final int consequentialVariantCount = variantAnalysis.consequentialVariants().size();
    final int structuralVariantCount = structuralVariantAnalysis.annotations().size();
    final String cancerType = PatientReporterHelper.extractCancerType(baseReporterData().patientsCancerTypes(), tumorSample);
    final TumorLocationDoidMapping doidMapping = TumorLocationDoidMapping.fromResource("/tumor_location_doid_mapping.csv");
    final List<Alteration> alterations = civicAnalyzer().run(variantAnalysis.findings(), purpleAnalysis.reportableGeneCopyNumbers(), reportableDisruptions, reportableFusions, reporterData().panelGeneModel(), doidMapping.doidsForTumorType(cancerType));
    LOGGER.info(" Printing analysis results:");
    LOGGER.info("  Number of passed variants : " + Integer.toString(passedVariantCount));
    LOGGER.info("  Number of missense variants (mutational load) : " + Integer.toString(mutationalLoad));
    LOGGER.info("  Number of consequential variants to report : " + Integer.toString(consequentialVariantCount));
    LOGGER.info(" Determined copy number stats for " + Integer.toString(purpleAnalysis.genePanelSize()) + " genes which led to " + Integer.toString(purpleAnalysis.reportableGeneCopyNumbers().size()) + " copy numbers.");
    LOGGER.info("  Number of unreported structural variants : " + Integer.toString(structuralVariantCount));
    LOGGER.info("  Number of gene fusions to report : " + Integer.toString(reportableFusions.size()));
    LOGGER.info("  Number of gene disruptions to report : " + Integer.toString(reportableDisruptions.size()));
    LOGGER.info("  Number of CIViC alterations to report : " + alterations.size());
    LOGGER.info("  Microsatellite analysis results: " + variantAnalysis.indelsPerMb() + " indels per MB");
    final Lims lims = baseReporterData().limsModel();
    final Double tumorPercentage = lims.tumorPercentageForSample(tumorSample);
    final List<VariantReport> purpleEnrichedVariants = purpleAnalysis.enrichSomaticVariants(variantAnalysis.findings());
    final String sampleRecipient = baseReporterData().centerModel().getAddresseeStringForSample(tumorSample);
    final SampleReport sampleReport = ImmutableSampleReport.of(tumorSample, cancerType, tumorPercentage, lims.arrivalDateForSample(tumorSample), lims.arrivalDateForSample(run.refSample()), lims.labProceduresForSample(tumorSample), sampleRecipient);
    return ImmutableSequencedPatientReport.of(sampleReport, purpleEnrichedVariants, mutationalLoad, variantAnalysis.indelsPerMb(), purpleAnalysis.reportableGeneCopyNumbers(), reportableDisruptions, reportableFusions, purpleAnalysis.purityString(), alterations, PatientReporterHelper.findCircosPlotPath(runDirectory, tumorSample), Optional.ofNullable(comments), baseReporterData().signaturePath());
}

Example 2 with StructuralVariantAnalysis

use of com.hartwig.hmftools.svannotation.analysis.StructuralVariantAnalysis in project hmftools by hartwigmedical.

the class LoadStructuralVariants method main.

public static void main(@NotNull final String[] args) throws ParseException, IOException, SQLException {
    final Options options = createBasicOptions();
    final CommandLine cmd = createCommandLine(args, options);
    boolean loadFromDB = cmd.hasOption(LOAD_FROM_DB);
    final String tumorSample = cmd.getOptionValue(SAMPLE);
    boolean runClustering = cmd.hasOption(CLUSTER_SVS);
    boolean createFilteredPON = cmd.hasOption(WRITE_FILTERED_SVS);
    boolean reannotateFromVCFs = cmd.hasOption(REANNOTATE_FROM_VCFS);
    final DatabaseAccess dbAccess = cmd.hasOption(DB_URL) ? databaseAccess(cmd) : null;
    if (cmd.hasOption(LOG_DEBUG)) {
        Configurator.setRootLevel(Level.DEBUG);
    }
    if (createFilteredPON) {
        LOGGER.info("reading VCF files including filtered SVs");
        FilteredSVWriter filteredSvWriter = new FilteredSVWriter(cmd.getOptionValue(VCF_FILE), cmd.getOptionValue(DATA_OUTPUT_PATH));
        filteredSvWriter.processVcfFiles();
        LOGGER.info("reads complete");
        return;
    }
    if (reannotateFromVCFs) {
        LOGGER.info("reading VCF files to re-annotate");
        // for now just re-read the VCFs and write out new annotations to file
        // may later on turn into update SQL once clustering does the same
        SvVCFAnnotator vcfAnnotator = new SvVCFAnnotator(cmd.getOptionValue(VCF_FILE), cmd.getOptionValue(DATA_OUTPUT_PATH));
        vcfAnnotator.processVcfFiles();
        return;
    }
    StructuralVariantClustering svClusterer = null;
    if (runClustering) {
        LOGGER.info("will run clustering logic");
        SvClusteringConfig clusteringConfig = new SvClusteringConfig();
        clusteringConfig.setOutputCsvPath(cmd.getOptionValue(DATA_OUTPUT_PATH));
        clusteringConfig.setBaseDistance(Integer.parseInt(cmd.getOptionValue(CLUSTER_BASE_DISTANCE, "0")));
        clusteringConfig.setUseCombinedOutputFile(tumorSample.equals("*"));
        clusteringConfig.setSvPONFile(cmd.getOptionValue(SV_PON_FILE, ""));
        clusteringConfig.setFragileSiteFile(cmd.getOptionValue(FRAGILE_SITE_FILE, ""));
        clusteringConfig.setLineElementFile(cmd.getOptionValue(LINE_ELEMENT_FILE, ""));
        clusteringConfig.setExternalAnnotationsFile(cmd.getOptionValue(EXTERNAL_ANNOTATIONS, ""));
        svClusterer = new StructuralVariantClustering(clusteringConfig);
    }
    if (createFilteredPON) {
        LOGGER.info("reading VCF file including filtered SVs");
        FilteredSVWriter filteredSvWriter = new FilteredSVWriter(cmd.getOptionValue(VCF_FILE), cmd.getOptionValue(DATA_OUTPUT_PATH));
        filteredSvWriter.processVcfFiles();
        LOGGER.info("reads complete");
        return;
    }
    if (!loadFromDB) {
        boolean skipAnnotations = cmd.hasOption(SKIP_ANNOTATIONS);
        LOGGER.info("reading VCF File");
        final List<StructuralVariant> variants = readFromVcf(cmd.getOptionValue(VCF_FILE), true);
        LOGGER.info("enriching structural variants based on purple data");
        final List<EnrichedStructuralVariant> enrichedVariantWithoutPrimaryId = enrichStructuralVariants(variants, dbAccess, tumorSample);
        LOGGER.info("persisting variants to database");
        dbAccess.writeStructuralVariants(tumorSample, enrichedVariantWithoutPrimaryId);
        // NEVA: We read after we write to populate the primaryId field
        final List<EnrichedStructuralVariant> enrichedVariants = dbAccess.readStructuralVariants(tumorSample);
        LOGGER.info("initialising MqSql annotator");
        final VariantAnnotator annotator = MySQLAnnotator.make("jdbc:" + cmd.getOptionValue(ENSEMBL_DB));
        LOGGER.info("loading Cosmic Fusion data");
        final CosmicFusionModel cosmicGeneFusions = CosmicFusions.readFromCSV(cmd.getOptionValue(FUSION_CSV));
        final StructuralVariantAnalyzer analyzer = new StructuralVariantAnalyzer(annotator, HmfGenePanelSupplier.hmfPanelGeneList(), cosmicGeneFusions);
        LOGGER.info("analyzing structural variants for impact via disruptions and fusions");
        final StructuralVariantAnalysis analysis = analyzer.run(enrichedVariants, skipAnnotations);
        if (runClustering) {
            svClusterer.loadFromEnrichedSVs(tumorSample, enrichedVariants);
            svClusterer.runClustering();
        }
        LOGGER.info("persisting annotations to database");
        final StructuralVariantAnnotationDAO annotationDAO = new StructuralVariantAnnotationDAO(dbAccess.context());
        annotationDAO.write(analysis);
    } else {
        // KODU: Below assert feels somewhat risky!?
        assert runClustering;
        List<String> samplesList = Lists.newArrayList();
        if (tumorSample.isEmpty() || tumorSample.equals("*")) {
            samplesList = getStructuralVariantSamplesList(dbAccess);
        } else if (tumorSample.contains(",")) {
            String[] tumorList = tumorSample.split(",");
            samplesList = Arrays.stream(tumorList).collect(Collectors.toList());
        } else {
            samplesList.add(tumorSample);
        }
        int count = 0;
        for (final String sample : samplesList) {
            ++count;
            LOGGER.info("clustering for sample({}), total({})", sample, count);
            List<SvClusterData> svClusterData = queryStructuralVariantData(dbAccess, sample);
            svClusterer.loadFromDatabase(sample, svClusterData);
            // LOGGER.info("data loaded", sample, count);
            svClusterer.runClustering();
        // LOGGER.info("clustering complete", sample, count);
        // if(count > 10)
        // break;
        }
    }
    svClusterer.close();
    LOGGER.info("run complete");
}

Example 3 with StructuralVariantAnalysis

use of com.hartwig.hmftools.svannotation.analysis.StructuralVariantAnalysis in project hmftools by hartwigmedical.

the class PatientReporter method analyseGenomeData.

@NotNull
private GenomeAnalysis analyseGenomeData(@NotNull final String sample, @NotNull final String runDirectory) throws IOException {
    LOGGER.info(" Loading somatic snv and indels...");
    final List<SomaticVariant> variants = PatientReporterHelper.loadPassedSomaticVariants(sample, runDirectory);
    LOGGER.info("  " + variants.size() + " somatic passed snps, mnps and indels loaded for sample " + sample);
    LOGGER.info(" Analyzing somatic snp/mnp and indels....");
    final VariantAnalysis variantAnalysis = variantAnalyzer().run(variants);
    LOGGER.info(" Loading purity numbers...");
    final PurityContext context = PatientReporterHelper.loadPurity(runDirectory, sample);
    if (context.status().equals(FittedPurityStatus.NO_TUMOR)) {
        LOGGER.warn("PURPLE DID NOT DETECT A TUMOR. Proceed with utmost caution!");
    }
    final FittedPurity purity = context.bestFit();
    final FittedPurityScore purityScore = context.score();
    final List<PurpleCopyNumber> purpleCopyNumbers = PatientReporterHelper.loadPurpleCopyNumbers(runDirectory, sample);
    final List<GeneCopyNumber> panelGeneCopyNumbers = PatientReporterHelper.loadPurpleGeneCopyNumbers(runDirectory, sample).stream().filter(x -> reporterData().panelGeneModel().panel().contains(x.gene())).collect(Collectors.toList());
    LOGGER.info("  " + purpleCopyNumbers.size() + " purple copy number regions loaded for sample " + sample);
    LOGGER.info(" Analyzing purple somatic copy numbers...");
    final PurpleAnalysis purpleAnalysis = ImmutablePurpleAnalysis.builder().gender(context.gender()).status(context.status()).fittedPurity(purity).fittedScorePurity(purityScore).copyNumbers(purpleCopyNumbers).panelGeneCopyNumbers(panelGeneCopyNumbers).build();
    final Path structuralVariantVCF = PatientReporterHelper.findStructuralVariantVCF(runDirectory);
    LOGGER.info(" Loading structural variants...");
    final List<StructuralVariant> structuralVariants = StructuralVariantFileLoader.fromFile(structuralVariantVCF.toString(), true);
    LOGGER.info(" Enriching structural variants with purple data.");
    final List<EnrichedStructuralVariant> enrichedStructuralVariants = purpleAnalysis.enrichStructuralVariants(structuralVariants);
    LOGGER.info(" Analysing structural variants...");
    final StructuralVariantAnalysis structuralVariantAnalysis = structuralVariantAnalyzer().run(enrichedStructuralVariants, false);
    return ImmutableGenomeAnalysis.of(sample, variantAnalysis, purpleAnalysis, structuralVariantAnalysis);
}