Java Code Examples for htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder#setOption()

@Override
protected int doWork() {
    final ProgressLogger progress = new ProgressLogger(LOG, 10000);
    
    IOUtil.assertFileIsReadable(INPUT);
    IOUtil.assertFileIsWritable(OUTPUT);

    final VCFFileReader reader = new VCFFileReader(INPUT, REQUIRE_INDEX);
    final VCFHeader header = new VCFHeader(reader.getFileHeader());
    final SAMSequenceDictionary sequenceDictionary = header.getSequenceDictionary();
    if (CREATE_INDEX && sequenceDictionary == null) {
        throw new PicardException("A sequence dictionary must be available in the input file when creating indexed output.");
    }

    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder()
            .setOutputFile(OUTPUT)
            .setReferenceDictionary(sequenceDictionary);
    if (CREATE_INDEX)
        builder.setOption(Options.INDEX_ON_THE_FLY);
    else
        builder.unsetOption(Options.INDEX_ON_THE_FLY);
    final VariantContextWriter writer = builder.build();
    writer.writeHeader(header);
    final CloseableIterator<VariantContext> iterator = reader.iterator();

    while (iterator.hasNext()) {
        final VariantContext context = iterator.next();
        writer.add(context);
        progress.record(context.getContig(), context.getStart());
    }

    CloserUtil.close(iterator);
    CloserUtil.close(reader);
    writer.close();

    return 0;
}
 

@Override
protected int doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    IOUtil.assertFileIsReadable(SEQUENCE_DICTIONARY);
    IOUtil.assertFileIsWritable(OUTPUT);

    final SAMSequenceDictionary samSequenceDictionary = SAMSequenceDictionaryExtractor.extractDictionary(SEQUENCE_DICTIONARY.toPath());

    final VCFFileReader fileReader = new VCFFileReader(INPUT, false);
    final VCFHeader fileHeader = fileReader.getFileHeader();

    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder()
            .setReferenceDictionary(samSequenceDictionary)
            .clearOptions();
    if (CREATE_INDEX)
        builder.setOption(Options.INDEX_ON_THE_FLY);

    final VariantContextWriter vcfWriter = builder.setOutputFile(OUTPUT).build();
    fileHeader.setSequenceDictionary(samSequenceDictionary);
    vcfWriter.writeHeader(fileHeader);

    final ProgressLogger progress = new ProgressLogger(log, 10000);
    final CloseableIterator<VariantContext> iterator = fileReader.iterator();
    while (iterator.hasNext()) {
        final VariantContext context = iterator.next();
        vcfWriter.add(context);
        progress.record(context.getContig(), context.getStart());
    }

    CloserUtil.close(iterator);
    CloserUtil.close(fileReader);
    vcfWriter.close();

    return 0;
}
 

/**
 * Creates a VariantContextWriter whose outputFile type is based on the extension of the output file name.
 * The default options set by VariantContextWriter are cleared before applying ALLOW_MISSING_FIELDS_IN_HEADER (if
 * <code>lenientProcessing</code> is set), followed by the set of options specified by any <code>options</code> args.
 *
 * @param outPath output Path for this writer. May not be null.
 * @param referenceDictionary required if on the fly indexing is set, otherwise can be null
 * @param createMD5 true if an md5 file should be created
 * @param options variable length list of additional Options to be set for this writer
 * @returns VariantContextWriter must be closed by the caller
 */
public static VariantContextWriter createVCFWriter(
        final Path outPath,
        final SAMSequenceDictionary referenceDictionary,
        final boolean createMD5,
        final Options... options)
{
    Utils.nonNull(outPath);

    VariantContextWriterBuilder vcWriterBuilder =
            new VariantContextWriterBuilder().clearOptions().setOutputPath(outPath);

    if (VariantContextWriterBuilder.OutputType.UNSPECIFIED == VariantContextWriterBuilder.determineOutputTypeFromFile(outPath)) {
        // the only way the user has to specify an output type is by file extension, and htsjdk
        // throws if it can't map the file extension to a known vcf type, so fallback to a default
        // of VCF
        logger.warn(String.format(
                "Can't determine output variant file format from output file extension \"%s\". Defaulting to VCF.",
                FilenameUtils.getExtension(outPath.getFileName().toString())));
        vcWriterBuilder = vcWriterBuilder.setOutputFileType(VariantContextWriterBuilder.OutputType.VCF);
    }

    if (createMD5) {
        vcWriterBuilder.setCreateMD5();
    }

    if (null != referenceDictionary) {
        vcWriterBuilder = vcWriterBuilder.setReferenceDictionary(referenceDictionary);
    }

    for (Options opt : options) {
        vcWriterBuilder = vcWriterBuilder.setOption(opt);
    }

    return vcWriterBuilder.build();
}
 

private static VariantContextWriter getVariantContextWriter(final OutputStream outputStream,
                                                            final SAMSequenceDictionary referenceSequenceDictionary) {
    VariantContextWriterBuilder vcWriterBuilder = new VariantContextWriterBuilder()
            .clearOptions()
            .setOutputStream(outputStream);

    if (null != referenceSequenceDictionary) {
        vcWriterBuilder = vcWriterBuilder.setReferenceDictionary(referenceSequenceDictionary);
    }
    for (final Options opt : new Options[]{}) {
        vcWriterBuilder = vcWriterBuilder.setOption(opt);
    }

    return vcWriterBuilder.build();
}
 

@Override
protected int doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    IOUtil.assertFileIsWritable(OUTPUT);

    final VCFFileReader reader = new VCFFileReader(INPUT, false);
    final VCFHeader inputVcfHeader = new VCFHeader(reader.getFileHeader().getMetaDataInInputOrder());
    final SAMSequenceDictionary sequenceDictionary = inputVcfHeader.getSequenceDictionary();

    if (CREATE_INDEX && sequenceDictionary == null) {
        throw new PicardException("A sequence dictionary must be available (either through the input file or by setting it explicitly) when creating indexed output.");
    }

    final ProgressLogger progress = new ProgressLogger(Log.getInstance(MakeSitesOnlyVcf.class), 10000);

    // Setup the site-only file writer
    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder()
            .setOutputFile(OUTPUT)
            .setReferenceDictionary(sequenceDictionary);
    if (CREATE_INDEX)
        builder.setOption(Options.INDEX_ON_THE_FLY);
    else
        builder.unsetOption(Options.INDEX_ON_THE_FLY);
    final VariantContextWriter writer = builder.build();

    final VCFHeader header = new VCFHeader(inputVcfHeader.getMetaDataInInputOrder(), SAMPLE);
    writer.writeHeader(header);

    // Go through the input, strip the records and write them to the output
    final CloseableIterator<VariantContext> iterator = reader.iterator();
    while (iterator.hasNext()) {
        final VariantContext full = iterator.next();
        final VariantContext site = subsetToSamplesWithOriginalAnnotations(full, SAMPLE);
        writer.add(site);
        progress.record(site.getContig(), site.getStart());
    }

    CloserUtil.close(iterator);
    CloserUtil.close(reader);
    writer.close();

    return 0;
}
 

@Override
protected int doWork() {
    final ProgressLogger progress = new ProgressLogger(log, 10000);
    final List<String> sampleList = new ArrayList<String>();
    INPUT = IOUtil.unrollFiles(INPUT, IOUtil.VCF_EXTENSIONS);
    final Collection<CloseableIterator<VariantContext>> iteratorCollection = new ArrayList<CloseableIterator<VariantContext>>(INPUT.size());
    final Collection<VCFHeader> headers = new HashSet<VCFHeader>(INPUT.size());
    VariantContextComparator variantContextComparator = null;
    SAMSequenceDictionary sequenceDictionary = null;

    if (SEQUENCE_DICTIONARY != null) {
        sequenceDictionary = SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(SEQUENCE_DICTIONARY).getFileHeader().getSequenceDictionary();
    }

    for (final File file : INPUT) {
        IOUtil.assertFileIsReadable(file);
        final VCFFileReader fileReader = new VCFFileReader(file, false);
        final VCFHeader fileHeader = fileReader.getFileHeader();
        if (fileHeader.getContigLines().isEmpty()) {
            if (sequenceDictionary == null) {
                throw new IllegalArgumentException(SEQ_DICT_REQUIRED);
            } else {
                fileHeader.setSequenceDictionary(sequenceDictionary);
            }
        }

        if (variantContextComparator == null) {
            variantContextComparator = fileHeader.getVCFRecordComparator();
        } else {
            if (!variantContextComparator.isCompatible(fileHeader.getContigLines())) {
                throw new IllegalArgumentException(
                        "The contig entries in input file " + file.getAbsolutePath() + " are not compatible with the others.");
            }
        }

        if (sequenceDictionary == null) sequenceDictionary = fileHeader.getSequenceDictionary();

        if (sampleList.isEmpty()) {
            sampleList.addAll(fileHeader.getSampleNamesInOrder());
        } else {
            if (!sampleList.equals(fileHeader.getSampleNamesInOrder())) {
                throw new IllegalArgumentException("Input file " + file.getAbsolutePath() + " has sample entries that don't match the other files.");
            }
        }
        
        // add comments in the first header
        if (headers.isEmpty()) {
            COMMENT.stream().forEach(C -> fileHeader.addMetaDataLine(new VCFHeaderLine("MergeVcfs.comment", C)));
        }

        headers.add(fileHeader);
        iteratorCollection.add(fileReader.iterator());
    }

    if (CREATE_INDEX && sequenceDictionary == null) {
        throw new PicardException(String.format("Index creation failed. %s", SEQ_DICT_REQUIRED));
    }

    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder()
            .setOutputFile(OUTPUT)
            .setReferenceDictionary(sequenceDictionary);

    if (CREATE_INDEX) {
        builder.setOption(Options.INDEX_ON_THE_FLY);
    } else {
        builder.unsetOption(Options.INDEX_ON_THE_FLY);
    }
    final VariantContextWriter writer = builder.build();

    writer.writeHeader(new VCFHeader(VCFUtils.smartMergeHeaders(headers, false), sampleList));

    final MergingIterator<VariantContext> mergingIterator = new MergingIterator<VariantContext>(variantContextComparator, iteratorCollection);
    while (mergingIterator.hasNext()) {
        final VariantContext context = mergingIterator.next();
        writer.add(context);
        progress.record(context.getContig(), context.getStart());
    }

    CloserUtil.close(mergingIterator);
    writer.close();
    return 0;
}
 

@Override
protected int doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    final ProgressLogger progress = new ProgressLogger(log, 10000);

    final VCFFileReader fileReader = new VCFFileReader(INPUT, false);
    final VCFHeader fileHeader = fileReader.getFileHeader();

    final SAMSequenceDictionary sequenceDictionary =
            SEQUENCE_DICTIONARY != null
                    ? SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).getFileHeader(SEQUENCE_DICTIONARY).getSequenceDictionary()
                    : fileHeader.getSequenceDictionary();
    if (CREATE_INDEX && sequenceDictionary == null) {
        throw new PicardException("A sequence dictionary must be available (either through the input file or by setting it explicitly) when creating indexed output.");
    }

    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder()
            .setReferenceDictionary(sequenceDictionary)
            .clearOptions();
    if (CREATE_INDEX)
        builder.setOption(Options.INDEX_ON_THE_FLY);

    final VariantContextWriter snpWriter = builder.setOutputFile(SNP_OUTPUT).build();
    final VariantContextWriter indelWriter = builder.setOutputFile(INDEL_OUTPUT).build();
    snpWriter.writeHeader(fileHeader);
    indelWriter.writeHeader(fileHeader);

    int incorrectVariantCount = 0;

    final CloseableIterator<VariantContext> iterator = fileReader.iterator();
    while (iterator.hasNext()) {
        final VariantContext context = iterator.next();
        if (context.isIndel()) indelWriter.add(context);
        else if (context.isSNP()) snpWriter.add(context);
        else {
            if (STRICT) throw new IllegalStateException("Found a record with type " + context.getType().name());
            else incorrectVariantCount++;
        }

        progress.record(context.getContig(), context.getStart());
    }

    if (incorrectVariantCount > 0) {
        log.debug("Found " + incorrectVariantCount + " records that didn't match SNP or INDEL");
    }

    CloserUtil.close(iterator);
    CloserUtil.close(fileReader);
    snpWriter.close();
    indelWriter.close();

    return 0;
}