Examples with ParallelProcessor cc.redberry.pipe.blocks.ParallelProcessor used on opensource projects

Example 1 with ParallelProcessor

use of cc.redberry.pipe.blocks.ParallelProcessor in project mixcr by milaboratory.

the class ActionAlign method go.

@Override
@SuppressWarnings("unchecked")
public void go(ActionHelper helper) throws Exception {
    // FIXME remove in 2.2
    if (actionParameters.printNonFunctionalWarnings())
        System.out.println("WARNING: -wf / --non-functional-warnings option is deprecated, will be removed in 2.2 " + "release. Use --verbose instead.");
    // Saving initial timestamp
    long beginTimestamp = System.currentTimeMillis();
    // Getting aligner parameters
    VDJCAlignerParameters alignerParameters = actionParameters.getAlignerParameters();
    // FIXME remove in 2.3
    if (actionParameters.getSaveOriginalReads()) {
        System.out.println("WARNING: -g / --save-reads option is deprecated, will be removed in 2.3 " + "release. Use -OsaveOriginalReads=true.");
        alignerParameters.setSaveOriginalReads(true);
    }
    // FIXME remove in 2.3
    if (actionParameters.getSaveReadDescription()) {
        System.out.println("WARNING: -a / --save-description option is deprecated, will be removed in 2.3 " + "release. Use -OsaveOriginalReads=true.");
        alignerParameters.setSaveOriginalReads(true);
    }
    if (!actionParameters.overrides.isEmpty()) {
        // Perform parameters overriding
        alignerParameters = JsonOverrider.override(alignerParameters, VDJCAlignerParameters.class, actionParameters.overrides);
        if (alignerParameters == null)
            throw new ProcessException("Failed to override some parameter.");
    }
    // FIXME remove in 2.2
    if (actionParameters.allowDifferentVJLoci != null && actionParameters.allowDifferentVJLoci) {
        System.out.println("Warning: usage of --diff-loci is deprecated. Use -OallowChimeras=true instead.");
        alignerParameters.setAllowChimeras(true);
    }
    // Creating aligner
    VDJCAligner aligner = VDJCAligner.createAligner(alignerParameters, actionParameters.isInputPaired(), !actionParameters.getNoMerge());
    // Detect if automatic featureToAlign correction is required
    int totalV = 0, totalVErrors = 0, hasVRegion = 0;
    GeneFeature correctingFeature = alignerParameters.getVAlignerParameters().getGeneFeatureToAlign().hasReversedRegions() ? GeneFeature.VRegionWithP : GeneFeature.VRegion;
    VDJCLibrary library = VDJCLibraryRegistry.getDefault().getLibrary(actionParameters.library, actionParameters.species);
    System.out.println("Reference library: " + library.getLibraryId());
    // Printing library level warnings, if specified for the library
    if (!library.getWarnings().isEmpty()) {
        System.out.println("Library warnings:");
        for (String l : library.getWarnings()) System.out.println(l);
    }
    // Printing citation notice, if specified for the library
    if (!library.getCitations().isEmpty()) {
        System.out.println("Please cite:");
        for (String l : library.getCitations()) System.out.println(l);
    }
    for (VDJCGene gene : library.getGenes(actionParameters.getChains())) {
        if (gene.getGeneType() == GeneType.Variable) {
            totalV++;
            if (!alignerParameters.containsRequiredFeature(gene)) {
                totalVErrors++;
                if (gene.getPartitioning().isAvailable(correctingFeature))
                    hasVRegion++;
            }
        }
    }
    // Performing V featureToAlign correction if needed
    if (totalVErrors > totalV * 0.9 && hasVRegion > totalVErrors * 0.8) {
        System.out.println("WARNING: forcing -OvParameters.geneFeatureToAlign=" + GeneFeature.encode(correctingFeature) + " since current gene feature (" + GeneFeature.encode(alignerParameters.getVAlignerParameters().getGeneFeatureToAlign()) + ") is absent in " + Util.PERCENT_FORMAT.format(100.0 * totalVErrors / totalV) + "% of V genes.");
        alignerParameters.getVAlignerParameters().setGeneFeatureToAlign(correctingFeature);
    }
    int numberOfExcludedNFGenes = 0;
    int numberOfExcludedFGenes = 0;
    for (VDJCGene gene : library.getGenes(actionParameters.getChains())) {
        NucleotideSequence featureSequence = alignerParameters.extractFeatureToAlign(gene);
        // exclusionReason is null ==> gene is not excluded
        String exclusionReason = null;
        if (featureSequence == null)
            exclusionReason = "absent " + GeneFeature.encode(alignerParameters.getFeatureToAlign(gene.getGeneType()));
        else if (featureSequence.containsWildcards())
            exclusionReason = "wildcard symbols in " + GeneFeature.encode(alignerParameters.getFeatureToAlign(gene.getGeneType()));
        if (exclusionReason == null)
            // If there are no reasons to exclude the gene, adding it to aligner
            aligner.addGene(gene);
        else {
            if (gene.isFunctional()) {
                ++numberOfExcludedFGenes;
                if (actionParameters.verbose())
                    System.out.println("WARNING: Functional gene " + gene.getName() + " excluded due to " + exclusionReason);
            } else
                ++numberOfExcludedNFGenes;
        }
    }
    if (actionParameters.printWarnings() && numberOfExcludedFGenes > 0)
        System.out.println("WARNING: " + numberOfExcludedFGenes + " functional genes were excluded, re-run " + "with --verbose option to see the list of excluded genes and exclusion reason.");
    if (actionParameters.verbose() && numberOfExcludedNFGenes > 0)
        System.out.println("WARNING: " + numberOfExcludedNFGenes + " non-functional genes excluded.");
    if (aligner.getVGenesToAlign().isEmpty())
        throw new ProcessException("No V genes to align. Aborting execution. See warnings for more info " + "(turn on verbose warnings by adding --verbose option).");
    if (aligner.getJGenesToAlign().isEmpty())
        throw new ProcessException("No J genes to align. Aborting execution. See warnings for more info " + "(turn on verbose warnings by adding --verbose option).");
    AlignerReport report = new AlignerReport();
    report.setStartMillis(beginTimestamp);
    report.setInputFiles(actionParameters.getInputsForReport());
    report.setOutputFiles(actionParameters.getOutputsForReport());
    report.setCommandLine(helper.getCommandLineArguments());
    // Attaching report to aligner
    aligner.setEventsListener(report);
    try (SequenceReaderCloseable<? extends SequenceRead> reader = actionParameters.createReader();
        VDJCAlignmentsWriter writer = actionParameters.getOutputName().equals(".") ? null : new VDJCAlignmentsWriter(actionParameters.getOutputName());
        SequenceWriter notAlignedWriter = actionParameters.failedReadsR1 == null ? null : (actionParameters.isInputPaired() ? new PairedFastqWriter(actionParameters.failedReadsR1, actionParameters.failedReadsR2) : new SingleFastqWriter(actionParameters.failedReadsR1))) {
        if (writer != null)
            writer.header(aligner);
        OutputPort<? extends SequenceRead> sReads = reader;
        CanReportProgress progress = (CanReportProgress) reader;
        if (actionParameters.limit != 0) {
            sReads = new CountLimitingOutputPort<>(sReads, actionParameters.limit);
            progress = SmartProgressReporter.extractProgress((CountLimitingOutputPort<?>) sReads);
        }
        final boolean writeAllResults = actionParameters.getWriteAllResults();
        EnumMap<GeneType, VDJCHit[]> emptyHits = new EnumMap<>(GeneType.class);
        for (GeneType gt : GeneType.values()) if (alignerParameters.getGeneAlignerParameters(gt) != null)
            emptyHits.put(gt, new VDJCHit[0]);
        final PairedEndReadsLayout readsLayout = alignerParameters.getReadsLayout();
        SmartProgressReporter.startProgressReport("Alignment", progress);
        OutputPort<Chunk<? extends SequenceRead>> mainInputReads = CUtils.buffered((OutputPort) chunked(sReads, 64), 16);
        OutputPort<VDJCAlignmentResult> alignments = unchunked(new ParallelProcessor(mainInputReads, chunked(aligner), actionParameters.threads));
        for (VDJCAlignmentResult result : CUtils.it(new OrderedOutputPort<>(alignments, new Indexer<VDJCAlignmentResult>() {

            @Override
            public long getIndex(VDJCAlignmentResult o) {
                return o.read.getId();
            }
        }))) {
            VDJCAlignments alignment = result.alignment;
            SequenceRead read = result.read;
            if (alignment == null) {
                if (writeAllResults) // Creating empty alignment object if alignment for current read failed
                {
                    Target target = readsLayout.createTargets(read)[0];
                    alignment = new VDJCAlignments(emptyHits, target.targets, SequenceHistory.RawSequence.of(read.getId(), target), alignerParameters.isSaveOriginalReads() ? new SequenceRead[] { read } : null);
                } else {
                    if (notAlignedWriter != null)
                        notAlignedWriter.write(result.read);
                    continue;
                }
            }
            if (alignment.isChimera())
                report.onChimera();
            if (writer != null)
                writer.write(alignment);
        }
        if (writer != null)
            writer.setNumberOfProcessedReads(reader.getNumberOfReads());
    }
    report.setFinishMillis(System.currentTimeMillis());
    // Writing report to stout
    System.out.println("============= Report ==============");
    Util.writeReportToStdout(report);
    if (actionParameters.report != null)
        Util.writeReport(actionParameters.report, report);
    if (actionParameters.jsonReport != null)
        Util.writeJsonReport(actionParameters.jsonReport, report);
}

Example 2 with ParallelProcessor

use of cc.redberry.pipe.blocks.ParallelProcessor in project mixcr by milaboratory.

the class ActionAssembleContigs method go.

@Override
public void go(ActionHelper helper) throws Exception {
    // TODO FIX!!!!!!!!!!!!!
    if (parameters.threads > 1)
        throw new ParameterException("Multithreaded processing is not supported yet.");
    long beginTimestamp = System.currentTimeMillis();
    FullSeqAssemblerParameters p = FullSeqAssemblerParameters.getByName("default");
    if (!parameters.overrides.isEmpty()) {
        // Perform parameters overriding
        p = JsonOverrider.override(p, FullSeqAssemblerParameters.class, parameters.overrides);
        if (p == null)
            throw new ProcessException("Failed to override some parameter.");
    }
    final FullSeqAssemblerReport report = new FullSeqAssemblerReport();
    FullSeqAssemblerParameters assemblerParameters = p;
    int totalClonesCount = 0;
    List<VDJCGene> genes;
    VDJCAlignerParameters alignerParameters;
    CloneAssemblerParameters cloneAssemblerParameters;
    try (ClnAReader reader = new ClnAReader(parameters.getInputFileName(), VDJCLibraryRegistry.getDefault());
        PrimitivO tmpOut = new PrimitivO(new BufferedOutputStream(new FileOutputStream(parameters.getOutputFileName())))) {
        final CloneFactory cloneFactory = new CloneFactory(reader.getAssemblerParameters().getCloneFactoryParameters(), reader.getAssemblingFeatures(), reader.getGenes(), reader.getAlignerParameters().getFeaturesToAlignMap());
        alignerParameters = reader.getAlignerParameters();
        cloneAssemblerParameters = reader.getAssemblerParameters();
        genes = reader.getGenes();
        IOUtil.registerGeneReferences(tmpOut, genes, alignerParameters);
        ClnAReader.CloneAlignmentsPort cloneAlignmentsPort = reader.clonesAndAlignments();
        SmartProgressReporter.startProgressReport("Assembling", cloneAlignmentsPort);
        OutputPort<Clone[]> parallelProcessor = new ParallelProcessor<>(cloneAlignmentsPort, cloneAlignments -> {
            FullSeqAssembler fullSeqAssembler = new FullSeqAssembler(cloneFactory, assemblerParameters, cloneAlignments.clone, alignerParameters);
            fullSeqAssembler.setReport(report);
            FullSeqAssembler.RawVariantsData rawVariantsData = fullSeqAssembler.calculateRawData(() -> {
                try {
                    return cloneAlignments.alignments();
                } catch (IOException e) {
                    throw new RuntimeException(e);
                }
            });
            return fullSeqAssembler.callVariants(rawVariantsData);
        }, parameters.threads);
        for (Clone[] clones : CUtils.it(parallelProcessor)) {
            totalClonesCount += clones.length;
            for (Clone cl : clones) tmpOut.writeObject(cl);
        }
        assert report.getInitialCloneCount() == reader.numberOfClones();
    }
    assert report.getFinalCloneCount() == totalClonesCount;
    assert report.getFinalCloneCount() >= report.getInitialCloneCount();
    Clone[] clones = new Clone[totalClonesCount];
    try (PrimitivI tmpIn = new PrimitivI(new BufferedInputStream(new FileInputStream(parameters.getOutputFileName())))) {
        IOUtil.registerGeneReferences(tmpIn, genes, alignerParameters);
        int i = 0;
        for (Clone clone : CUtils.it(new PipeDataInputReader<>(Clone.class, tmpIn, totalClonesCount))) clones[i++] = clone;
    }
    Arrays.sort(clones, Comparator.comparingDouble(c -> -c.getCount()));
    for (int i = 0; i < clones.length; i++) clones[i] = clones[i].setId(i);
    CloneSet cloneSet = new CloneSet(Arrays.asList(clones), genes, alignerParameters.getFeaturesToAlignMap(), alignerParameters, cloneAssemblerParameters);
    try (CloneSetIO.CloneSetWriter writer = new CloneSetIO.CloneSetWriter(cloneSet, parameters.getOutputFileName())) {
        SmartProgressReporter.startProgressReport(writer);
        writer.write();
    }
    ReportWrapper reportWrapper = new ReportWrapper(command(), report);
    reportWrapper.setStartMillis(beginTimestamp);
    reportWrapper.setInputFiles(parameters.getInputFileName());
    reportWrapper.setOutputFiles(parameters.getOutputFileName());
    reportWrapper.setCommandLine(helper.getCommandLineArguments());
    reportWrapper.setFinishMillis(System.currentTimeMillis());
    // Writing report to stout
    System.out.println("============= Report ==============");
    Util.writeReportToStdout(report);
    if (parameters.report != null)
        Util.writeReport(parameters.report, reportWrapper);
    if (parameters.jsonReport != null)
        Util.writeJsonReport(parameters.jsonReport, reportWrapper);
}

Example 3 with ParallelProcessor

use of cc.redberry.pipe.blocks.ParallelProcessor in project mixcr by milaboratory.

the class RunMiXCR method align.

public static AlignResult align(RunMiXCRAnalysis parameters) throws Exception {
    VDJCAlignerParameters alignerParameters = parameters.alignerParameters;
    VDJCAligner aligner = VDJCAligner.createAligner(alignerParameters, parameters.isInputPaired(), alignerParameters.getMergerParameters() != null);
    List<VDJCGene> genes = new ArrayList<>();
    for (VDJCGene gene : VDJCLibraryRegistry.getDefault().getLibrary(parameters.library, parameters.species).getGenes(parameters.chains)) if (alignerParameters.containsRequiredFeature(gene) && (gene.isFunctional() || !parameters.isFunctionalOnly)) {
        genes.add(gene);
        aligner.addGene(gene);
    }
    AlignerReport report = new AlignerReport();
    aligner.setEventsListener(report);
    try (SequenceReaderCloseable<? extends SequenceRead> reader = parameters.getReader()) {
        // start progress reporting
        if (reader instanceof CanReportProgress)
            SmartProgressReporter.startProgressReport("align", (CanReportProgress) reader);
        OutputPort<Chunk<SequenceRead>> mainInputReads = CUtils.buffered((OutputPort) chunked(reader, 64), 16);
        OutputPort<VDJCAlignmentResult> alignments = unchunked(new ParallelProcessor(mainInputReads, chunked(aligner), parameters.threads));
        List<VDJCAlignments> als = new ArrayList<>();
        int ind = 0;
        for (VDJCAlignmentResult t : CUtils.it(new OrderedOutputPort<>(alignments, new Indexer<VDJCAlignmentResult>() {

            @Override
            public long getIndex(VDJCAlignmentResult r) {
                return r.read.getId();
            }
        }))) {
            if (t.alignment != null) {
                t.alignment.setAlignmentsIndex(ind++);
                als.add(t.alignment);
            }
        }
        return new AlignResult(parameters, reader.getNumberOfReads(), report, als, genes, aligner);
    }
}