Examples with VariantContextWriterBuilder htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder used on opensource projects

Example 11 with VariantContextWriterBuilder

use of htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder in project gatk by broadinstitute.

the class LiftOverVcf method doWork.

@Override
protected Object doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    IOUtil.assertFileIsReadable(REFERENCE_SEQUENCE);
    IOUtil.assertFileIsReadable(CHAIN);
    IOUtil.assertFileIsWritable(OUTPUT);
    IOUtil.assertFileIsWritable(REJECT);
    ////////////////////////////////////////////////////////////////////////
    // Setup the inputs
    ////////////////////////////////////////////////////////////////////////
    final LiftOver liftOver = new LiftOver(CHAIN);
    final VCFFileReader in = new VCFFileReader(INPUT, false);
    logger.info("Loading up the target reference genome.");
    final ReferenceSequenceFileWalker walker = new ReferenceSequenceFileWalker(REFERENCE_SEQUENCE);
    final Map<String, byte[]> refSeqs = new HashMap<>();
    for (final SAMSequenceRecord rec : walker.getSequenceDictionary().getSequences()) {
        refSeqs.put(rec.getSequenceName(), walker.get(rec.getSequenceIndex()).getBases());
    }
    CloserUtil.close(walker);
    ////////////////////////////////////////////////////////////////////////
    // Setup the outputs
    ////////////////////////////////////////////////////////////////////////
    final VCFHeader inHeader = in.getFileHeader();
    final VCFHeader outHeader = new VCFHeader(inHeader);
    outHeader.setSequenceDictionary(walker.getSequenceDictionary());
    final VariantContextWriter out = new VariantContextWriterBuilder().setOption(Options.INDEX_ON_THE_FLY).setOutputFile(OUTPUT).setReferenceDictionary(walker.getSequenceDictionary()).build();
    out.writeHeader(outHeader);
    final VariantContextWriter rejects = new VariantContextWriterBuilder().setOutputFile(REJECT).unsetOption(Options.INDEX_ON_THE_FLY).build();
    final VCFHeader rejectHeader = new VCFHeader(in.getFileHeader());
    for (final VCFFilterHeaderLine line : FILTERS) rejectHeader.addMetaDataLine(line);
    rejects.writeHeader(rejectHeader);
    ////////////////////////////////////////////////////////////////////////
    // Read the input VCF, lift the records over and write to the sorting
    // collection.
    ////////////////////////////////////////////////////////////////////////
    long failedLiftover = 0, failedAlleleCheck = 0, total = 0;
    logger.info("Lifting variants over and sorting.");
    final SortingCollection<VariantContext> sorter = SortingCollection.newInstance(VariantContext.class, new VCFRecordCodec(outHeader), outHeader.getVCFRecordComparator(), MAX_RECORDS_IN_RAM, TMP_DIR);
    ProgressLogger progress = new ProgressLogger(logger, 1000000, "read");
    for (final VariantContext ctx : in) {
        ++total;
        final Interval source = new Interval(ctx.getContig(), ctx.getStart(), ctx.getEnd(), false, ctx.getContig() + ":" + ctx.getStart() + "-" + ctx.getEnd());
        final Interval target = liftOver.liftOver(source, 1.0);
        if (target == null) {
            rejects.add(new VariantContextBuilder(ctx).filter(FILTER_CANNOT_LIFTOVER).make());
            failedLiftover++;
        } else {
            // Fix the alleles if we went from positive to negative strand
            final List<Allele> alleles = new ArrayList<>();
            for (final Allele oldAllele : ctx.getAlleles()) {
                if (target.isPositiveStrand() || oldAllele.isSymbolic()) {
                    alleles.add(oldAllele);
                } else {
                    alleles.add(Allele.create(SequenceUtil.reverseComplement(oldAllele.getBaseString()), oldAllele.isReference()));
                }
            }
            // Build the new variant context
            final VariantContextBuilder builder = new VariantContextBuilder(ctx.getSource(), target.getContig(), target.getStart(), target.getEnd(), alleles);
            builder.id(ctx.getID());
            builder.attributes(ctx.getAttributes());
            builder.genotypes(ctx.getGenotypes());
            builder.filters(ctx.getFilters());
            builder.log10PError(ctx.getLog10PError());
            // Check that the reference allele still agrees with the reference sequence
            boolean mismatchesReference = false;
            for (final Allele allele : builder.getAlleles()) {
                if (allele.isReference()) {
                    final byte[] ref = refSeqs.get(target.getContig());
                    final String refString = StringUtil.bytesToString(ref, target.getStart() - 1, target.length());
                    if (!refString.equalsIgnoreCase(allele.getBaseString())) {
                        mismatchesReference = true;
                    }
                    break;
                }
            }
            if (mismatchesReference) {
                rejects.add(new VariantContextBuilder(ctx).filter(FILTER_MISMATCHING_REF_ALLELE).make());
                failedAlleleCheck++;
            } else {
                sorter.add(builder.make());
            }
        }
        progress.record(ctx.getContig(), ctx.getStart());
    }
    final NumberFormat pfmt = new DecimalFormat("0.0000%");
    final String pct = pfmt.format((failedLiftover + failedAlleleCheck) / (double) total);
    logger.info("Processed ", total, " variants.");
    logger.info(Long.toString(failedLiftover), " variants failed to liftover.");
    logger.info(Long.toString(failedAlleleCheck), " variants lifted over but had mismatching reference alleles after lift over.");
    logger.info(pct, " of variants were not successfully lifted over and written to the output.");
    rejects.close();
    in.close();
    ////////////////////////////////////////////////////////////////////////
    // Write the sorted outputs to the final output file
    ////////////////////////////////////////////////////////////////////////
    sorter.doneAdding();
    progress = new ProgressLogger(logger, 1000000, "written");
    logger.info("Writing out sorted records to final VCF.");
    for (final VariantContext ctx : sorter) {
        out.add(ctx);
        progress.record(ctx.getContig(), ctx.getStart());
    }
    out.close();
    sorter.cleanup();
    return null;
}

Example 12 with VariantContextWriterBuilder

use of htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder in project gatk by broadinstitute.

the class SplitVcfs method doWork.

@Override
protected Object doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    final ProgressLogger progress = new ProgressLogger(logger, 10000);
    final VCFFileReader fileReader = new VCFFileReader(INPUT);
    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 UserException("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);
    try (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) {
            logger.debug("Found " + incorrectVariantCount + " records that didn't match SNP or INDEL");
        }
        CloserUtil.close(iterator);
        CloserUtil.close(fileReader);
    }
    return null;
}

Example 13 with VariantContextWriterBuilder

use of htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder in project gatk by broadinstitute.

the class EvaluateCopyNumberTriStateCalls method openVCFWriter.

private VariantContextWriter openVCFWriter(final File outputFile, final Set<String> samples) {
    final VariantContextWriterBuilder builder = new VariantContextWriterBuilder();
    builder.setOutputFile(outputFile);
    builder.clearOptions();
    final VariantContextWriter result = builder.build();
    final VCFHeader header = new VCFHeader(Collections.emptySet(), samples);
    CopyNumberTriStateAllele.addHeaderLinesTo(header);
    EvaluationClass.addHeaderLinesTo(header);
    // Format annotations.
    header.addMetaDataLine(new VCFFormatHeaderLine(VCFConstants.GENOTYPE_KEY, 1, VCFHeaderLineType.Character, "Called genotype"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.CALL_QUALITY_KEY, 1, VCFHeaderLineType.Float, "Quality of the call"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.CALLED_SEGMENTS_COUNT_KEY, 1, VCFHeaderLineType.Integer, "Number of called segments that overlap with the truth"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.CALLED_ALLELE_COUNTS_KEY, VCFHeaderLineCount.G, VCFHeaderLineType.Integer, "Called allele count for mixed calls"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.TRUTH_COPY_FRACTION_KEY, 1, VCFHeaderLineType.Float, "Truth copy fraction estimated"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.TRUTH_QUALITY_KEY, 1, VCFHeaderLineType.Float, "Truth call quality"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.EVALUATION_CLASS_KEY, 1, VCFHeaderLineType.Character, "The evaluation class for the call or lack of call. It the values of the header key '" + EvaluationClass.VCF_HEADER_KEY + "'"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.TRUTH_GENOTYPE_KEY, 1, VCFHeaderLineType.Character, "The truth genotype"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.CALLED_TARGET_COUNT_KEY, 1, VCFHeaderLineType.Integer, "Number of targets covered by called segments"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VariantEvaluationContext.CALL_QUALITY_KEY, 1, VCFHeaderLineType.Float, "1 - The probability of th event in Phred scale (the maximum if ther are more than one segment"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VCFConstants.GENOTYPE_QUALITY_KEY, 1, VCFHeaderLineType.Integer, "The quality of the call (the maximum if there are more than one segment"));
    header.addMetaDataLine(new VCFFormatHeaderLine(VCFConstants.GENOTYPE_FILTER_KEY, VCFHeaderLineCount.UNBOUNDED, VCFHeaderLineType.Character, "Genotype filters"));
    // Info annotations.
    header.addMetaDataLine(new VCFInfoHeaderLine(VariantEvaluationContext.TRUTH_ALLELE_FREQUENCY_KEY, VCFHeaderLineCount.A, VCFHeaderLineType.Float, "The frequency of the alternative alleles in the truth callset"));
    header.addMetaDataLine(new VCFInfoHeaderLine(VariantEvaluationContext.TRUTH_ALLELE_NUMBER_KEY, 1, VCFHeaderLineType.Integer, "Total number of called alleles in the truth callset"));
    header.addMetaDataLine(new VCFInfoHeaderLine(VariantEvaluationContext.CALLS_ALLELE_FREQUENCY_KEY, VCFHeaderLineCount.A, VCFHeaderLineType.Float, "The frequency of the alternative alleles in the actual callset"));
    header.addMetaDataLine(new VCFInfoHeaderLine(VariantEvaluationContext.CALLS_ALLELE_NUMBER_KEY, 1, VCFHeaderLineType.Integer, "Total number of called alleles in the actual callset"));
    header.addMetaDataLine(new VCFInfoHeaderLine(VariantEvaluationContext.TRUTH_TARGET_COUNT_KEY, 1, VCFHeaderLineType.Integer, "Number of targets overlapped by this variant"));
    header.addMetaDataLine(new VCFInfoHeaderLine(VCFConstants.END_KEY, 1, VCFHeaderLineType.Integer, "Stop position for the variant"));
    // Filter annotations.
    for (final EvaluationFilter filter : EvaluationFilter.values()) {
        header.addMetaDataLine(new VCFFilterHeaderLine(filter.name(), filter.description));
        header.addMetaDataLine(new VCFFilterHeaderLine(filter.acronym, filter.description));
    }
    header.addMetaDataLine(new VCFFilterHeaderLine(EvaluationFilter.PASS, "Indicates that it passes all filters"));
    result.writeHeader(header);
    return result;
}

Example 14 with VariantContextWriterBuilder

use of htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder in project gatk by broadinstitute.

the class SVVCFWriter method getVariantContextWriter.

private static VariantContextWriter getVariantContextWriter(final OutputStream outputStream, final SAMSequenceDictionary referenceSequenceDictionary) {
    VariantContextWriterBuilder vcWriterBuilder = new VariantContextWriterBuilder().clearOptions().setOutputStream(outputStream);
    if (null != referenceSequenceDictionary) {
        vcWriterBuilder = vcWriterBuilder.setReferenceDictionary(referenceSequenceDictionary);
    }
    // todo: remove this when things are solid?
    vcWriterBuilder = vcWriterBuilder.setOption(Options.ALLOW_MISSING_FIELDS_IN_HEADER);
    for (final Options opt : new Options[] {}) {
        vcWriterBuilder = vcWriterBuilder.setOption(opt);
    }
    return vcWriterBuilder.build();
}