Examples with SomaticRunMetadata com.hartwig.pipeline.metadata.SomaticRunMetadata used on opensource projects

Example 1 with SomaticRunMetadata

use of com.hartwig.pipeline.metadata.SomaticRunMetadata in project pipeline5 by hartwigmedical.

the class LinxGermline method persistedOutput.

@Override
public LinxGermlineOutput persistedOutput(final SomaticRunMetadata metadata) {
    String disruptionsTsv = metadata.sampleName() + GERMLINE_DISRUPTION_TSV;
    String driverCatalogTsv = metadata.sampleName() + GERMLINE_DRIVER_CATALOG_TSV;
    return LinxGermlineOutput.builder().status(PipelineStatus.PERSISTED).maybeLinxGermlineOutputLocations(LinxGermlineOutputLocations.builder().disruptions(persistedOrDefault(metadata, DataType.LINX_GERMLINE_DISRUPTIONS, disruptionsTsv)).driverCatalog(persistedOrDefault(metadata, DataType.LINX_GERMLINE_DRIVER_CATALOG, driverCatalogTsv)).outputDirectory(persistedOrDefault(metadata, DataType.LINX_DRIVER_CATALOG, driverCatalogTsv).transform(f -> new File(f).getParent()).asDirectory()).build()).build();
}

Example 2 with SomaticRunMetadata

use of com.hartwig.pipeline.metadata.SomaticRunMetadata in project pipeline5 by hartwigmedical.

the class LinxSomatic method persistedOutput.

@Override
public LinxSomaticOutput persistedOutput(final SomaticRunMetadata metadata) {
    String breakendTsv = metadata.tumor().sampleName() + BREAKEND_TSV;
    String driverCatalogTsv = metadata.tumor().sampleName() + DRIVER_CATALOG_TSV;
    String fusionsTsv = metadata.tumor().sampleName() + FUSION_TSV;
    String driversTsv = metadata.tumor().sampleName() + DRIVERS_TSV;
    String clustersTsv = metadata.tumor().sampleName() + CLUSTERS_TSV;
    String svAnnotationsTsv = metadata.tumor().sampleName() + SV_ANNOTATIONS_TSV;
    return LinxSomaticOutput.builder().status(PipelineStatus.PERSISTED).maybeLinxOutputLocations(LinxSomaticOutputLocations.builder().breakends(persistedOrDefault(metadata, DataType.LINX_BREAKENDS, breakendTsv)).driverCatalog(persistedOrDefault(metadata, DataType.LINX_DRIVER_CATALOG, driverCatalogTsv)).drivers(persistedOrDefault(metadata, DataType.LINX_DRIVERS, driversTsv)).fusions(persistedOrDefault(metadata, DataType.LINX_FUSIONS, fusionsTsv)).svAnnotations(persistedOrDefault(metadata, DataType.LINX_SV_ANNOTATIONS, svAnnotationsTsv)).clusters(persistedOrDefault(metadata, DataType.LINX_CLUSTERS, clustersTsv)).outputDirectory(persistedOrDefault(metadata, DataType.LINX_DRIVER_CATALOG, driverCatalogTsv).transform(f -> new File(f).getParent()).asDirectory()).build()).build();
}

Example 3 with SomaticRunMetadata

use of com.hartwig.pipeline.metadata.SomaticRunMetadata in project pipeline5 by hartwigmedical.

the class StagedOutputPublisher method publish.

public void publish(final PipelineState state, final SomaticRunMetadata metadata) {
    if (state.status() != PipelineStatus.FAILED && run.isPresent()) {
        List<AddDatatype> addDatatypes = state.stageOutputs().stream().map(StageOutput::datatypes).flatMap(List::stream).collect(Collectors.toList());
        SampleSet set = setResolver.resolve(metadata.set(), useOnlyDBSets);
        Optional<String> tumorSampleName = metadata.maybeTumor().map(SingleSampleRunMetadata::sampleName);
        Optional<String> refSampleName = metadata.maybeReference().map(SingleSampleRunMetadata::sampleName);
        ImmutableAnalysis.Builder alignedReadsAnalysis = eventBuilder(Type.ALIGNMENT);
        ImmutableAnalysis.Builder somaticAnalysis = eventBuilder(Type.SOMATIC);
        ImmutableAnalysis.Builder germlineAnalysis = eventBuilder(Type.GERMLINE);
        OutputIterator.from(blob -> {
            Optional<AddDatatype> dataType = addDatatypes.stream().filter(d -> blob.getName().endsWith(d.path())).findFirst();
            Blob blobWithMd5 = sourceBucket.get(blob.getName());
            if (isSecondary(blobWithMd5)) {
                alignedReadsAnalysis.addOutput(createBlob(tumorSampleName, refSampleName, dataType, blobWithMd5));
            } else {
                if (isGermline(blobWithMd5)) {
                    germlineAnalysis.addOutput(createBlob(tumorSampleName, refSampleName, dataType, blobWithMd5));
                } else if (notSecondary(blobWithMd5)) {
                    somaticAnalysis.addOutput(createBlob(tumorSampleName, refSampleName, dataType, blobWithMd5));
                }
            }
        }, sourceBucket).iterate(metadata);
        publish(PipelineComplete.builder().pipeline(ImmutablePipeline.builder().sample(tumorSampleName.orElseGet(() -> refSampleName.orElseThrow())).bucket(sourceBucket.getName()).runId(run.get().getId()).setId(set.getId()).context(context).addAnalyses(alignedReadsAnalysis.build(), somaticAnalysis.build(), germlineAnalysis.build()).version(Versions.pipelineMajorMinorVersion()).build()).build());
    }
}

Example 4 with SomaticRunMetadata

use of com.hartwig.pipeline.metadata.SomaticRunMetadata in project pipeline5 by hartwigmedical.

the class SageCaller method outputBuilder.

protected ImmutableSageOutput.Builder outputBuilder(final SomaticRunMetadata metadata, final PipelineStatus jobStatus, final RuntimeBucket bucket, final ResultsDirectory resultsDirectory) {
    final String filteredOutputFile = sageConfiguration.filteredTemplate().apply(metadata);
    final String unfilteredOutputFile = sageConfiguration.unfilteredTemplate().apply(metadata);
    final String geneCoverageFile = sageConfiguration.geneCoverageTemplate().apply(metadata);
    final Optional<String> somaticRefSampleBqrPlot = referenceSampleBqrPlot(metadata);
    final Optional<String> somaticTumorSampleBqrPlot = tumorSampleBqrPlot(metadata);
    final ImmutableSageOutput.Builder builder = SageOutput.builder(namespace()).status(jobStatus);
    somaticRefSampleBqrPlot.ifPresent(s -> builder.maybeSomaticRefSampleBqrPlot(GoogleStorageLocation.of(bucket.name(), resultsDirectory.path(s))).addReportComponents(bqrComponent("png", bucket, resultsDirectory, metadata.reference().sampleName())).addReportComponents(bqrComponent("tsv", bucket, resultsDirectory, metadata.reference().sampleName())));
    somaticTumorSampleBqrPlot.ifPresent(s -> builder.maybeSomaticTumorSampleBqrPlot(GoogleStorageLocation.of(bucket.name(), resultsDirectory.path(s))).addReportComponents(bqrComponent("png", bucket, resultsDirectory, metadata.tumor().sampleName())).addReportComponents(bqrComponent("tsv", bucket, resultsDirectory, metadata.tumor().sampleName())));
    return builder.addFailedLogLocations(GoogleStorageLocation.of(bucket.name(), RunLogComponent.LOG_FILE)).maybeGermlineGeneCoverage(GoogleStorageLocation.of(bucket.name(), resultsDirectory.path(geneCoverageFile))).maybeSomaticTumorSampleBqrPlot(somaticTumorSampleBqrPlot.map(t -> GoogleStorageLocation.of(bucket.name(), resultsDirectory.path(t)))).maybeVariants(GoogleStorageLocation.of(bucket.name(), resultsDirectory.path(filteredOutputFile))).addReportComponents(bqrComponent("png", bucket, resultsDirectory, metadata.sampleName())).addReportComponents(bqrComponent("tsv", bucket, resultsDirectory, metadata.sampleName())).addReportComponents(vcfComponent(unfilteredOutputFile, bucket, resultsDirectory)).addReportComponents(vcfComponent(filteredOutputFile, bucket, resultsDirectory)).addReportComponents(singleFileComponent(geneCoverageFile, bucket, resultsDirectory)).addReportComponents(new RunLogComponent(bucket, namespace(), Folder.root(), resultsDirectory)).addReportComponents(new StartupScriptComponent(bucket, namespace(), Folder.root())).addDatatypes(new AddDatatype(sageConfiguration.vcfDatatype(), metadata.barcode(), new ArchivePath(Folder.root(), namespace(), filteredOutputFile))).addDatatypes(new AddDatatype(sageConfiguration.geneCoverageDatatype(), metadata.barcode(), new ArchivePath(Folder.root(), namespace(), geneCoverageFile))).addAllDatatypes(somaticRefSampleBqrPlot.stream().map(r -> new AddDatatype(sageConfiguration.tumorSampleBqrPlot(), metadata.barcode(), new ArchivePath(Folder.root(), namespace(), r))).collect(Collectors.toList())).addAllDatatypes(somaticTumorSampleBqrPlot.stream().map(t -> new AddDatatype(sageConfiguration.refSampleBqrPlot(), metadata.barcode(), new ArchivePath(Folder.root(), namespace(), t))).collect(Collectors.toList()));
}

Example 5 with SomaticRunMetadata

use of com.hartwig.pipeline.metadata.SomaticRunMetadata in project pipeline5 by hartwigmedical.

the class PipelineMain method start.

public PipelineState start(final Arguments arguments) {
    LOGGER.info("Arguments are [{}]", arguments);
    Versions.printAll();
    try {
        GoogleCredentials credentials = CredentialProvider.from(arguments).get();
        Storage storage = StorageProvider.from(arguments, credentials).get();
        Publisher turquoisePublisher = PublisherProvider.from(arguments, credentials).get("turquoise.events");
        Publisher pipelinePublisher = PublisherProvider.from(arguments, credentials).get(PipelineComplete.TOPIC);
        SomaticMetadataApi somaticMetadataApi = SomaticMetadataApiProvider.from(arguments, storage, pipelinePublisher).get();
        SingleSampleEventListener referenceEventListener = new SingleSampleEventListener();
        SingleSampleEventListener tumorEventListener = new SingleSampleEventListener();
        SomaticRunMetadata somaticRunMetadata = somaticMetadataApi.get();
        InputMode mode = new ModeResolver().apply(somaticRunMetadata);
        LOGGER.info("Starting pipeline in [{}] mode", mode);
        String ini = somaticRunMetadata.isSingleSample() ? "single_sample" : arguments.shallow() ? "shallow" : "somatic";
        PipelineProperties eventSubjects = PipelineProperties.builder().sample(somaticRunMetadata.maybeTumor().map(SingleSampleRunMetadata::sampleName).orElseGet(() -> somaticRunMetadata.reference().sampleName())).runId(arguments.sbpApiRunId()).set(somaticRunMetadata.set()).referenceBarcode(somaticRunMetadata.maybeReference().map(SingleSampleRunMetadata::barcode)).tumorBarcode(somaticRunMetadata.maybeTumor().map(SingleSampleRunMetadata::barcode)).type(ini).build();
        somaticMetadataApi.start();
        startedEvent(eventSubjects, turquoisePublisher, arguments.publishToTurquoise());
        BlockingQueue<BamMetricsOutput> referenceBamMetricsOutputQueue = new ArrayBlockingQueue<>(1);
        BlockingQueue<BamMetricsOutput> tumorBamMetricsOutputQueue = new ArrayBlockingQueue<>(1);
        BlockingQueue<FlagstatOutput> referenceFlagstatOutputQueue = new ArrayBlockingQueue<>(1);
        BlockingQueue<FlagstatOutput> tumorFlagstatOutputQueue = new ArrayBlockingQueue<>(1);
        BlockingQueue<GermlineCallerOutput> germlineCallerOutputQueue = new ArrayBlockingQueue<>(1);
        StartingPoint startingPoint = new StartingPoint(arguments);
        PersistedDataset persistedDataset = arguments.biopsy().<PersistedDataset>map(b -> new ApiPersistedDataset(SbpRestApi.newInstance(arguments.sbpApiUrl()), ObjectMappers.get(), b, arguments.project())).orElse(new NoopPersistedDataset());
        PipelineState state = new FullPipeline(singleSamplePipeline(arguments, credentials, storage, referenceEventListener, somaticRunMetadata, referenceBamMetricsOutputQueue, germlineCallerOutputQueue, referenceFlagstatOutputQueue, startingPoint, persistedDataset, mode), singleSamplePipeline(arguments, credentials, storage, tumorEventListener, somaticRunMetadata, tumorBamMetricsOutputQueue, germlineCallerOutputQueue, tumorFlagstatOutputQueue, startingPoint, persistedDataset, mode), somaticPipeline(arguments, credentials, storage, somaticRunMetadata, referenceBamMetricsOutputQueue, tumorBamMetricsOutputQueue, referenceFlagstatOutputQueue, tumorFlagstatOutputQueue, startingPoint, persistedDataset, mode), Executors.newCachedThreadPool(), referenceEventListener, tumorEventListener, somaticMetadataApi, CleanupProvider.from(arguments, storage).get()).run();
        completedEvent(eventSubjects, turquoisePublisher, state.status().toString(), arguments.publishToTurquoise());
        VmExecutionLogSummary.ofFailedStages(storage, state);
        return state;
    } catch (Exception e) {
        throw new RuntimeException(e);
    }
}