Search in sources :

Example 1 with InputDownload

use of com.hartwig.pipeline.execution.vm.InputDownload in project pipeline5 by hartwigmedical.

the class GridssRerun method execute.

@Override
public VirtualMachineJobDefinition execute(final InputBundle inputs, final RuntimeBucket runtimeBucket, final BashStartupScript commands, final RuntimeFiles executionFlags) {
    // Inputs
    final ResourceFiles resourceFiles = ResourceFilesFactory.buildResourceFiles(RefGenomeVersion.V37);
    final String set = inputs.get("set").inputValue();
    final String tumorSampleName = inputs.get("tumor_sample").inputValue();
    final String referenceSampleName = inputs.get("reference_sample").inputValue();
    final InputFileDescriptor remoteTumorFile = inputs.get("tumor_cram");
    final InputFileDescriptor remoteReferenceFile = inputs.get("ref_cram");
    final InputFileDescriptor runData = inputs.get();
    final RemoteLocationsApi locationsApi = new RemoteLocationsApi(runData.billedProject(), tumorSampleName);
    InputDownload tumorBamDownload = new InputDownload(locationsApi.getTumorAlignment());
    InputDownload tumorBamIndexDownload = new InputDownload(locationsApi.getTumorAlignmentIndex());
    InputDownload referenceBamDownload = new InputDownload(locationsApi.getReferenceAlignment());
    InputDownload referenceBamIndexDownload = new InputDownload(locationsApi.getReferenceAlignmentIndex());
    final InputFileDescriptor remoteTumorIndex = remoteTumorFile.index();
    final InputFileDescriptor remoteReferenceIndex = remoteReferenceFile.index();
    final String localTumorFile = localFilename(remoteTumorFile);
    final String localReferenceFile = localFilename(remoteReferenceFile);
    final String tumorBamPath = localTumorFile.replace("cram", "bam");
    final String refBamPath = localReferenceFile.replace("cram", "bam");
    Driver driver = new Driver(resourceFiles, VmDirectories.outputFile(tumorSampleName + ".assembly.bam")).tumorSample(tumorSampleName, tumorBamPath).referenceSample(referenceSampleName, refBamPath);
    GridssAnnotation viralAnnotation = new GridssAnnotation(resourceFiles, false);
    SubStageInputOutput unfilteredVcfOutput = driver.andThen(viralAnnotation).apply(SubStageInputOutput.empty(tumorSampleName));
    final OutputFile unfilteredVcf = unfilteredVcfOutput.outputFile();
    final OutputFile unfilteredVcfIndex = unfilteredVcf.index(".tbi");
    final GoogleStorageLocation unfilteredVcfRemoteLocation = remoteUnfilteredVcfArchivePath(set, tumorSampleName);
    final GoogleStorageLocation unfilteredVcfIndexRemoteLocation = index(unfilteredVcfRemoteLocation, ".tbi");
    // COMMANDS
    commands.addCommand(new ExportPathCommand(new BwaCommand()));
    commands.addCommand(new ExportPathCommand(new SamtoolsCommand()));
    commands.addCommand(() -> remoteTumorFile.toCommandForm(localTumorFile));
    commands.addCommand(() -> remoteTumorIndex.toCommandForm(localFilename(remoteTumorIndex)));
    commands.addCommand(() -> remoteReferenceFile.toCommandForm(localReferenceFile));
    commands.addCommand(() -> remoteReferenceIndex.toCommandForm(localFilename(remoteReferenceIndex)));
    if (!localTumorFile.equals(tumorBamPath)) {
        commands.addCommands(cramToBam(localTumorFile));
    }
    if (!localReferenceFile.equals(refBamPath)) {
        commands.addCommands(cramToBam(localReferenceFile));
    }
    commands.addCommands(unfilteredVcfOutput.bash());
    commands.addCommand(() -> unfilteredVcf.copyToRemoteLocation(unfilteredVcfRemoteLocation));
    commands.addCommand(() -> unfilteredVcfIndex.copyToRemoteLocation(unfilteredVcfIndexRemoteLocation));
    commands.addCommand(new OutputUpload(GoogleStorageLocation.of(runtimeBucket.name(), "gridss"), executionFlags));
    return VirtualMachineJobDefinition.structuralCalling(commands, ResultsDirectory.defaultDirectory());
}
Also used : OutputFile(com.hartwig.pipeline.execution.vm.OutputFile) ExportPathCommand(com.hartwig.pipeline.execution.vm.unix.ExportPathCommand) InputFileDescriptor(com.hartwig.batch.input.InputFileDescriptor) Driver(com.hartwig.pipeline.calling.structural.gridss.stage.Driver) SubStageInputOutput(com.hartwig.pipeline.stages.SubStageInputOutput) GridssAnnotation(com.hartwig.pipeline.calling.structural.gridss.stage.GridssAnnotation) RemoteLocationsApi(com.hartwig.batch.api.RemoteLocationsApi) BwaCommand(com.hartwig.pipeline.calling.command.BwaCommand) ResourceFiles(com.hartwig.pipeline.resource.ResourceFiles) SamtoolsCommand(com.hartwig.pipeline.calling.command.SamtoolsCommand) OutputUpload(com.hartwig.pipeline.execution.vm.OutputUpload) InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) GoogleStorageLocation(com.hartwig.pipeline.storage.GoogleStorageLocation)

Example 2 with InputDownload

use of com.hartwig.pipeline.execution.vm.InputDownload in project pipeline5 by hartwigmedical.

the class LocalLocations method get.

private String get(Supplier<GoogleStorageLocation> remotePath) {
    if (commandsGenerated) {
        throw new IllegalStateException("Cannot request new files after generating download commands");
    }
    GoogleStorageLocation location = remotePath.get();
    InputDownload result = new InputDownload(location);
    commands.add(result.asBash());
    return result.getLocalTargetPath();
}
Also used : InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) GoogleStorageLocation(com.hartwig.pipeline.storage.GoogleStorageLocation)

Example 3 with InputDownload

use of com.hartwig.pipeline.execution.vm.InputDownload in project pipeline5 by hartwigmedical.

the class LocalLocations method getDirectory.

private String getDirectory(Supplier<GoogleStorageLocation> remotePath, final String localPath) {
    if (commandsGenerated) {
        throw new IllegalStateException("Cannot request new files after generating download commands");
    }
    GoogleStorageLocation location = remotePath.get();
    InputDownload result = new InputDownload(location, localPath);
    commands.add("mkdir -p " + localPath);
    commands.add(result.asBash());
    return result.getLocalTargetPath();
}
Also used : InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) GoogleStorageLocation(com.hartwig.pipeline.storage.GoogleStorageLocation)

Example 4 with InputDownload

use of com.hartwig.pipeline.execution.vm.InputDownload in project pipeline5 by hartwigmedical.

the class BwaAligner method run.

public AlignmentOutput run(final SingleSampleRunMetadata metadata) throws Exception {
    StageTrace trace = new StageTrace(NAMESPACE, metadata.sampleName(), StageTrace.ExecutorType.COMPUTE_ENGINE).start();
    RuntimeBucket rootBucket = RuntimeBucket.from(storage, NAMESPACE, metadata, arguments, labels);
    Sample sample = sampleSource.sample(metadata);
    if (sample.bam().isPresent()) {
        String noPrefix = sample.bam().orElseThrow().replace("gs://", "");
        int firstSlash = noPrefix.indexOf("/");
        String bucket = noPrefix.substring(0, firstSlash);
        String path = noPrefix.substring(firstSlash + 1);
        return AlignmentOutput.builder().sample(metadata.sampleName()).status(PipelineStatus.PROVIDED).maybeAlignments(GoogleStorageLocation.of(bucket, path)).build();
    }
    final ResourceFiles resourceFiles = buildResourceFiles(arguments);
    sampleUpload.run(sample, rootBucket);
    List<Future<PipelineStatus>> futures = new ArrayList<>();
    List<GoogleStorageLocation> perLaneBams = new ArrayList<>();
    List<ReportComponent> laneLogComponents = new ArrayList<>();
    List<GoogleStorageLocation> laneFailedLogs = new ArrayList<>();
    for (Lane lane : sample.lanes()) {
        RuntimeBucket laneBucket = RuntimeBucket.from(storage, laneNamespace(lane), metadata, arguments, labels);
        BashStartupScript bash = BashStartupScript.of(laneBucket.name());
        InputDownload first = new InputDownload(GoogleStorageLocation.of(rootBucket.name(), fastQFileName(sample.name(), lane.firstOfPairPath())));
        InputDownload second = new InputDownload(GoogleStorageLocation.of(rootBucket.name(), fastQFileName(sample.name(), lane.secondOfPairPath())));
        bash.addCommand(first).addCommand(second);
        bash.addCommands(OverrideReferenceGenomeCommand.overrides(arguments));
        SubStageInputOutput alignment = new LaneAlignment(arguments.sbpApiRunId().isPresent(), resourceFiles.refGenomeFile(), first.getLocalTargetPath(), second.getLocalTargetPath(), metadata.sampleName(), lane).apply(SubStageInputOutput.empty(metadata.sampleName()));
        perLaneBams.add(GoogleStorageLocation.of(laneBucket.name(), resultsDirectory.path(alignment.outputFile().fileName())));
        bash.addCommands(alignment.bash()).addCommand(new OutputUpload(GoogleStorageLocation.of(laneBucket.name(), resultsDirectory.path()), RuntimeFiles.typical()));
        futures.add(executorService.submit(() -> runWithRetries(metadata, laneBucket, VirtualMachineJobDefinition.alignment(laneId(lane).toLowerCase(), bash, resultsDirectory))));
        laneLogComponents.add(new RunLogComponent(laneBucket, laneNamespace(lane), Folder.from(metadata), resultsDirectory));
        laneFailedLogs.add(GoogleStorageLocation.of(laneBucket.name(), RunLogComponent.LOG_FILE));
    }
    AlignmentOutput output;
    if (lanesSuccessfullyComplete(futures)) {
        List<InputDownload> laneBams = perLaneBams.stream().map(InputDownload::new).collect(Collectors.toList());
        BashStartupScript mergeMarkdupsBash = BashStartupScript.of(rootBucket.name());
        laneBams.forEach(mergeMarkdupsBash::addCommand);
        SubStageInputOutput merged = new MergeMarkDups(laneBams.stream().map(InputDownload::getLocalTargetPath).filter(path -> path.endsWith("bam")).collect(Collectors.toList())).apply(SubStageInputOutput.empty(metadata.sampleName()));
        mergeMarkdupsBash.addCommands(merged.bash());
        mergeMarkdupsBash.addCommand(new OutputUpload(GoogleStorageLocation.of(rootBucket.name(), resultsDirectory.path()), RuntimeFiles.typical()));
        PipelineStatus status = runWithRetries(metadata, rootBucket, VirtualMachineJobDefinition.mergeMarkdups(mergeMarkdupsBash, resultsDirectory));
        ImmutableAlignmentOutput.Builder outputBuilder = AlignmentOutput.builder().sample(metadata.sampleName()).status(status).maybeAlignments(GoogleStorageLocation.of(rootBucket.name(), resultsDirectory.path(merged.outputFile().fileName()))).addAllReportComponents(laneLogComponents).addAllFailedLogLocations(laneFailedLogs).addFailedLogLocations(GoogleStorageLocation.of(rootBucket.name(), RunLogComponent.LOG_FILE)).addReportComponents(new RunLogComponent(rootBucket, Aligner.NAMESPACE, Folder.from(metadata), resultsDirectory));
        if (!arguments.outputCram()) {
            outputBuilder.addReportComponents(new SingleFileComponent(rootBucket, Aligner.NAMESPACE, Folder.from(metadata), bam(metadata.sampleName()), bam(metadata.sampleName()), resultsDirectory), new SingleFileComponent(rootBucket, Aligner.NAMESPACE, Folder.from(metadata), bai(bam(metadata.sampleName())), bai(bam(metadata.sampleName())), resultsDirectory)).addDatatypes(new AddDatatype(DataType.ALIGNED_READS, metadata.barcode(), new ArchivePath(Folder.from(metadata), BwaAligner.NAMESPACE, bam(metadata.sampleName()))), new AddDatatype(DataType.ALIGNED_READS_INDEX, metadata.barcode(), new ArchivePath(Folder.from(metadata), BwaAligner.NAMESPACE, bai(metadata.sampleName()))));
        }
        output = outputBuilder.build();
    } else {
        output = AlignmentOutput.builder().sample(metadata.sampleName()).status(PipelineStatus.FAILED).build();
    }
    trace.stop();
    executorService.shutdown();
    return output;
}
Also used : Arguments(com.hartwig.pipeline.Arguments) StageTrace(com.hartwig.pipeline.trace.StageTrace) SubStageInputOutput(com.hartwig.pipeline.stages.SubStageInputOutput) Aligner(com.hartwig.pipeline.alignment.Aligner) InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) ArrayList(java.util.ArrayList) VirtualMachineJobDefinition(com.hartwig.pipeline.execution.vm.VirtualMachineJobDefinition) Future(java.util.concurrent.Future) RuntimeBucket(com.hartwig.pipeline.storage.RuntimeBucket) PipelineStatus(com.hartwig.pipeline.execution.PipelineStatus) ExecutorService(java.util.concurrent.ExecutorService) BashStartupScript(com.hartwig.pipeline.execution.vm.BashStartupScript) DataType(com.hartwig.pipeline.datatypes.DataType) FileTypes.bai(com.hartwig.pipeline.datatypes.FileTypes.bai) ImmutableAlignmentOutput(com.hartwig.pipeline.alignment.ImmutableAlignmentOutput) GoogleStorageLocation(com.hartwig.pipeline.storage.GoogleStorageLocation) Lane(com.hartwig.patient.Lane) ArchivePath(com.hartwig.pipeline.metadata.ArchivePath) SampleUpload(com.hartwig.pipeline.storage.SampleUpload) Folder(com.hartwig.pipeline.report.Folder) ResultsDirectory(com.hartwig.pipeline.ResultsDirectory) OutputUpload(com.hartwig.pipeline.execution.vm.OutputUpload) DefaultBackoffPolicy(com.hartwig.pipeline.failsafe.DefaultBackoffPolicy) Collectors(java.util.stream.Collectors) String.format(java.lang.String.format) File(java.io.File) SingleFileComponent(com.hartwig.pipeline.report.SingleFileComponent) Failsafe(net.jodah.failsafe.Failsafe) ResourceFilesFactory.buildResourceFiles(com.hartwig.pipeline.resource.ResourceFilesFactory.buildResourceFiles) ExecutionException(java.util.concurrent.ExecutionException) List(java.util.List) Sample(com.hartwig.patient.Sample) AddDatatype(com.hartwig.pipeline.metadata.AddDatatype) AlignmentOutput(com.hartwig.pipeline.alignment.AlignmentOutput) OverrideReferenceGenomeCommand(com.hartwig.pipeline.resource.OverrideReferenceGenomeCommand) RuntimeFiles(com.hartwig.pipeline.execution.vm.RuntimeFiles) SingleSampleRunMetadata(com.hartwig.pipeline.metadata.SingleSampleRunMetadata) Storage(com.google.cloud.storage.Storage) Labels(com.hartwig.pipeline.labels.Labels) FileTypes.bam(com.hartwig.pipeline.datatypes.FileTypes.bam) ResourceFiles(com.hartwig.pipeline.resource.ResourceFiles) ComputeEngine(com.hartwig.pipeline.execution.vm.ComputeEngine) ReportComponent(com.hartwig.pipeline.report.ReportComponent) RunLogComponent(com.hartwig.pipeline.report.RunLogComponent) SampleSource(com.hartwig.pipeline.alignment.sample.SampleSource) RunLogComponent(com.hartwig.pipeline.report.RunLogComponent) PipelineStatus(com.hartwig.pipeline.execution.PipelineStatus) ArrayList(java.util.ArrayList) ReportComponent(com.hartwig.pipeline.report.ReportComponent) AddDatatype(com.hartwig.pipeline.metadata.AddDatatype) ArchivePath(com.hartwig.pipeline.metadata.ArchivePath) BashStartupScript(com.hartwig.pipeline.execution.vm.BashStartupScript) InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) SingleFileComponent(com.hartwig.pipeline.report.SingleFileComponent) Sample(com.hartwig.patient.Sample) Lane(com.hartwig.patient.Lane) SubStageInputOutput(com.hartwig.pipeline.stages.SubStageInputOutput) ImmutableAlignmentOutput(com.hartwig.pipeline.alignment.ImmutableAlignmentOutput) StageTrace(com.hartwig.pipeline.trace.StageTrace) ResourceFilesFactory.buildResourceFiles(com.hartwig.pipeline.resource.ResourceFilesFactory.buildResourceFiles) ResourceFiles(com.hartwig.pipeline.resource.ResourceFiles) OutputUpload(com.hartwig.pipeline.execution.vm.OutputUpload) ImmutableAlignmentOutput(com.hartwig.pipeline.alignment.ImmutableAlignmentOutput) AlignmentOutput(com.hartwig.pipeline.alignment.AlignmentOutput) RuntimeBucket(com.hartwig.pipeline.storage.RuntimeBucket) Future(java.util.concurrent.Future) GoogleStorageLocation(com.hartwig.pipeline.storage.GoogleStorageLocation)

Example 5 with InputDownload

use of com.hartwig.pipeline.execution.vm.InputDownload in project pipeline5 by hartwigmedical.

the class TeloBatch method execute.

@Override
public VirtualMachineJobDefinition execute(final InputBundle inputs, final RuntimeBucket runtimeBucket, final BashStartupScript commands, final RuntimeFiles executionFlags) {
    final String sampleId = inputs.get("sampleId").inputValue();
    Optional<String> specificChromosome = Optional.empty();
    try {
        specificChromosome = Optional.of(inputs.get("specificChromosome").inputValue());
    } catch (IllegalArgumentException ignored) {
    }
    final InputFileDescriptor runData = inputs.get();
    final RemoteLocationsApi locationsApi = new RemoteLocationsApi(runData.billedProject(), sampleId);
    // download the telo.jar
    // InputDownload teloJarDownload = new InputDownload(GoogleStorageLocation.of(teloToolsBucket, teloToolsPath + "/telo.jar"), VmDirectories.TOOLS);
    // InputDownload teloJarDownload = downloadExperimentalVersion();
    // commands.addCommand(teloJarDownload);
    commands.addCommand(downloadExperimentalVersion());
    /*() -> format("gsutil -u hmf-crunch cp gs://%s/%s/%s %s",
                COMMON_RESOURCES, TELO_DIR, TELO_JAR, VmDirectories.TOOLS));*/
    // ref genome
    final ResourceFiles resourceFiles = ResourceFilesFactory.buildResourceFiles(RefGenomeVersion.V37);
    InputDownload tumorBamDownload = new InputDownload(locationsApi.getTumorAlignment());
    InputDownload tumorBamIndexDownload = new InputDownload(locationsApi.getTumorAlignmentIndex());
    // download the tumour and reference bam / index files
    commands.addCommand(tumorBamDownload);
    commands.addCommand(tumorBamIndexDownload);
    commands.addCommand(makeTeloRunCommand(sampleId, "somatic", tumorBamDownload.getLocalTargetPath(), resourceFiles.refGenomeFile(), specificChromosome));
    // delete the tumor bam file to save disk space
    commands.addCommand(() -> format("rm -f %s", tumorBamDownload.getLocalTargetPath()));
    commands.addCommand(() -> format("rm -f %s", tumorBamIndexDownload.getLocalTargetPath()));
    InputDownload referenceBamDownload = new InputDownload(locationsApi.getReferenceAlignment());
    InputDownload referenceBamIndexDownload = new InputDownload(locationsApi.getReferenceAlignmentIndex());
    commands.addCommand(referenceBamDownload);
    commands.addCommand(referenceBamIndexDownload);
    commands.addCommand(makeTeloRunCommand(sampleId, "germline", referenceBamDownload.getLocalTargetPath(), resourceFiles.refGenomeFile(), specificChromosome));
    // JavaJarCommand jarCommand = new JavaJarCommand("telo", TELO_VERSION, "telo.jar", "16G", teloArgs);
    // commands.addCommand(jarCommand);
    // Store output
    commands.addCommand(new OutputUpload(GoogleStorageLocation.of(runtimeBucket.name(), sampleId), executionFlags));
    return ImmutableVirtualMachineJobDefinition.builder().name("telo").startupCommand(commands).namespacedResults(ResultsDirectory.defaultDirectory()).workingDiskSpaceGb(500).performanceProfile(VirtualMachinePerformanceProfile.custom(16, MEMORY_GB)).build();
}
Also used : ResourceFiles(com.hartwig.pipeline.resource.ResourceFiles) OutputUpload(com.hartwig.pipeline.execution.vm.OutputUpload) InputFileDescriptor(com.hartwig.batch.input.InputFileDescriptor) InputDownload(com.hartwig.pipeline.execution.vm.InputDownload) RemoteLocationsApi(com.hartwig.batch.api.RemoteLocationsApi)

Aggregations

InputDownload (com.hartwig.pipeline.execution.vm.InputDownload)6 GoogleStorageLocation (com.hartwig.pipeline.storage.GoogleStorageLocation)5 ResourceFiles (com.hartwig.pipeline.resource.ResourceFiles)4 OutputUpload (com.hartwig.pipeline.execution.vm.OutputUpload)3 RemoteLocationsApi (com.hartwig.batch.api.RemoteLocationsApi)2 InputFileDescriptor (com.hartwig.batch.input.InputFileDescriptor)2 SubStageInputOutput (com.hartwig.pipeline.stages.SubStageInputOutput)2 Storage (com.google.cloud.storage.Storage)1 Lane (com.hartwig.patient.Lane)1 Sample (com.hartwig.patient.Sample)1 Arguments (com.hartwig.pipeline.Arguments)1 ResultsDirectory (com.hartwig.pipeline.ResultsDirectory)1 Aligner (com.hartwig.pipeline.alignment.Aligner)1 AlignmentOutput (com.hartwig.pipeline.alignment.AlignmentOutput)1 ImmutableAlignmentOutput (com.hartwig.pipeline.alignment.ImmutableAlignmentOutput)1 SampleSource (com.hartwig.pipeline.alignment.sample.SampleSource)1 BwaCommand (com.hartwig.pipeline.calling.command.BwaCommand)1 SamtoolsCommand (com.hartwig.pipeline.calling.command.SamtoolsCommand)1 Driver (com.hartwig.pipeline.calling.structural.gridss.stage.Driver)1 GridssAnnotation (com.hartwig.pipeline.calling.structural.gridss.stage.GridssAnnotation)1