Search in sources :

Example 1 with SequenceWriter

use of com.milaboratory.core.io.sequence.SequenceWriter 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);
}
Also used : VDJCAlignerParameters(com.milaboratory.mixcr.vdjaligners.VDJCAlignerParameters) VDJCAlignmentsWriter(com.milaboratory.mixcr.basictypes.VDJCAlignmentsWriter) VDJCAligner(com.milaboratory.mixcr.vdjaligners.VDJCAligner) ParallelProcessor(cc.redberry.pipe.blocks.ParallelProcessor) Target(com.milaboratory.core.Target) Indexer(cc.redberry.pipe.util.Indexer) CanReportProgress(com.milaboratory.util.CanReportProgress) SingleFastqWriter(com.milaboratory.core.io.sequence.fastq.SingleFastqWriter) VDJCAlignments(com.milaboratory.mixcr.basictypes.VDJCAlignments) PairedEndReadsLayout(com.milaboratory.core.PairedEndReadsLayout) VDJCAlignmentResult(com.milaboratory.mixcr.vdjaligners.VDJCAlignmentResult) PairedFastqWriter(com.milaboratory.core.io.sequence.fastq.PairedFastqWriter) Chunk(cc.redberry.pipe.util.Chunk) ProcessException(com.milaboratory.cli.ProcessException) CountLimitingOutputPort(cc.redberry.pipe.util.CountLimitingOutputPort) SequenceWriter(com.milaboratory.core.io.sequence.SequenceWriter) NucleotideSequence(com.milaboratory.core.sequence.NucleotideSequence) SequenceRead(com.milaboratory.core.io.sequence.SequenceRead)

Example 2 with SequenceWriter

use of com.milaboratory.core.io.sequence.SequenceWriter in project mixcr by milaboratory.

the class ActionExportCloneReads method go.

@Override
public void go(ActionHelper helper) throws Exception {
    try (ClnAReader clna = new ClnAReader(parameters.getInputFileName(), VDJCLibraryRegistry.createDefaultRegistry())) {
        VDJCAlignments firstAlignment = clna.readAllAlignments().take();
        if (firstAlignment == null)
            return;
        if (firstAlignment.getOriginalReads() == null)
            throw new ParameterException("Error: original reads were not saved in the .vdjca file: " + "re-run align with '-g' option.");
        int[] cid = parameters.getCloneIds();
        Supplier<IntStream> cloneIds;
        if (cid == null)
            cloneIds = () -> IntStream.range(0, clna.numberOfClones());
        else
            cloneIds = () -> IntStream.of(cid);
        long totalAlignments = cloneIds.get().mapToLong(clna::numberOfAlignmentsInClone).sum();
        AtomicLong alignmentsWritten = new AtomicLong();
        AtomicBoolean finished = new AtomicBoolean(false);
        SmartProgressReporter.startProgressReport("Writing reads", new CanReportProgress() {

            @Override
            public double getProgress() {
                return 1.0 * alignmentsWritten.get() / totalAlignments;
            }

            @Override
            public boolean isFinished() {
                return finished.get();
            }
        });
        boolean paired = firstAlignment.getOriginalReads().get(0).numberOfReads() == 2;
        boolean separate = parameters.doSeparate();
        SequenceWriter globalWriter = separate ? null : createWriter(paired, parameters.getOutputFileName());
        cloneIds.get().forEach(cloneId -> {
            try (SequenceWriter individualWriter = globalWriter == null ? createWriter(paired, cloneFile(parameters.getOutputFileName(), cloneId)) : null) {
                SequenceWriter actualWriter = globalWriter == null ? individualWriter : globalWriter;
                for (VDJCAlignments alignments : CUtils.it(clna.readAlignmentsOfClone(cloneId))) {
                    for (SequenceRead read : alignments.getOriginalReads()) actualWriter.write(read);
                    alignmentsWritten.incrementAndGet();
                }
            } catch (IOException e) {
                throw new RuntimeException(e);
            }
        });
    }
}
Also used : ClnAReader(com.milaboratory.mixcr.basictypes.ClnAReader) IOException(java.io.IOException) AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) AtomicLong(java.util.concurrent.atomic.AtomicLong) CanReportProgress(com.milaboratory.util.CanReportProgress) SequenceWriter(com.milaboratory.core.io.sequence.SequenceWriter) SequenceRead(com.milaboratory.core.io.sequence.SequenceRead) ParameterException(com.beust.jcommander.ParameterException) VDJCAlignments(com.milaboratory.mixcr.basictypes.VDJCAlignments) IntStream(java.util.stream.IntStream)

Aggregations

SequenceRead (com.milaboratory.core.io.sequence.SequenceRead)2 SequenceWriter (com.milaboratory.core.io.sequence.SequenceWriter)2 VDJCAlignments (com.milaboratory.mixcr.basictypes.VDJCAlignments)2 CanReportProgress (com.milaboratory.util.CanReportProgress)2 ParallelProcessor (cc.redberry.pipe.blocks.ParallelProcessor)1 Chunk (cc.redberry.pipe.util.Chunk)1 CountLimitingOutputPort (cc.redberry.pipe.util.CountLimitingOutputPort)1 Indexer (cc.redberry.pipe.util.Indexer)1 ParameterException (com.beust.jcommander.ParameterException)1 ProcessException (com.milaboratory.cli.ProcessException)1 PairedEndReadsLayout (com.milaboratory.core.PairedEndReadsLayout)1 Target (com.milaboratory.core.Target)1 PairedFastqWriter (com.milaboratory.core.io.sequence.fastq.PairedFastqWriter)1 SingleFastqWriter (com.milaboratory.core.io.sequence.fastq.SingleFastqWriter)1 NucleotideSequence (com.milaboratory.core.sequence.NucleotideSequence)1 ClnAReader (com.milaboratory.mixcr.basictypes.ClnAReader)1 VDJCAlignmentsWriter (com.milaboratory.mixcr.basictypes.VDJCAlignmentsWriter)1 VDJCAligner (com.milaboratory.mixcr.vdjaligners.VDJCAligner)1 VDJCAlignerParameters (com.milaboratory.mixcr.vdjaligners.VDJCAlignerParameters)1 VDJCAlignmentResult (com.milaboratory.mixcr.vdjaligners.VDJCAlignmentResult)1