htsjdk.variant.vcf.VCFStandardHeaderLines Java Examples

The following examples show how to use htsjdk.variant.vcf.VCFStandardHeaderLines. You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example #1
Source File: StrelkaPostProcessApplication.java    From hmftools with GNU General Public License v3.0 6 votes vote down vote up 
@NotNull
public static VCFHeader generateOutputHeader(@NotNull final VCFHeader header, @NotNull final String sampleName) {
    final VCFHeader outputVCFHeader = new VCFHeader(header.getMetaDataInInputOrder(), Sets.newHashSet(sampleName));
    outputVCFHeader.addMetaDataLine(VCFStandardHeaderLines.getFormatLine("GT"));
    outputVCFHeader.addMetaDataLine(VCFStandardHeaderLines.getFormatLine("AD"));

    outputVCFHeader.addMetaDataLine(new VCFHeaderLine("StrelkaGATKCompatibility",
            "Added GT fields to strelka calls for gatk compatibility."));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine("MAPPABILITY", 1, VCFHeaderLineType.Float, "Mappability (percentage)"));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine("SOMATIC_PON_COUNT",
            1,
            VCFHeaderLineType.Integer,
            "Number of times the variant appears in the somatic PON"));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine("GERMLINE_PON_COUNT",
            1,
            VCFHeaderLineType.Integer,
            "Number of times the variant appears in the germline PON"));
    return outputVCFHeader;
}
Example #2
Source File: GVCFWriterUnitTest.java    From gatk with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
private static VCFHeader getMinimalVCFHeader() {
    final Set<VCFHeaderLine> headerlines = new LinkedHashSet<>();
    VCFStandardHeaderLines.addStandardFormatLines(headerlines, true,
            VCFConstants.GENOTYPE_KEY, VCFConstants.DEPTH_KEY,
            VCFConstants.GENOTYPE_QUALITY_KEY, VCFConstants.GENOTYPE_PL_KEY,
            VCFConstants.GENOTYPE_ALLELE_DEPTHS);

    VCFStandardHeaderLines.addStandardInfoLines(headerlines, true,
            VCFConstants.DEPTH_KEY,
            VCFConstants.RMS_MAPPING_QUALITY_KEY,
            VCFConstants.MAPPING_QUALITY_ZERO_KEY );

    Arrays.asList(GATKVCFConstants.BASE_QUAL_RANK_SUM_KEY,
           GATKVCFConstants.CLIPPING_RANK_SUM_KEY,
           GATKVCFConstants.MLE_ALLELE_COUNT_KEY,
           GATKVCFConstants.MLE_ALLELE_FREQUENCY_KEY,
           GATKVCFConstants.MAP_QUAL_RANK_SUM_KEY,
           GATKVCFConstants.READ_POS_RANK_SUM_KEY)
           .forEach( c -> headerlines.add(GATKVCFHeaderLines.getInfoLine(c)));

    headerlines.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.STRAND_BIAS_BY_SAMPLE_KEY));
    return new VCFHeader(headerlines, Collections.singleton(SAMPLE_NAME));
}
Example #3
Source File: FingerprintUtils.java    From picard with MIT License 6 votes vote down vote up 
private static VariantContextWriter getVariantContextWriter(final File outputFile,
                                                            final File referenceSequenceFileName,
                                                            final String sample,
                                                            final String source,
                                                            final ReferenceSequenceFile ref) {
    final VariantContextWriter variantContextWriter = new VariantContextWriterBuilder()
            .setReferenceDictionary(ref.getSequenceDictionary())
            .setOutputFile(outputFile).build();

    final Set<VCFHeaderLine> lines = new LinkedHashSet<>();
    lines.add(new VCFHeaderLine("reference", referenceSequenceFileName.getAbsolutePath()));
    lines.add(new VCFHeaderLine("source", source));
    lines.add(new VCFHeaderLine("fileDate", new Date().toString()));

    lines.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_PL_KEY));
    lines.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_ALLELE_DEPTHS));
    lines.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.DEPTH_KEY));

    final VCFHeader header = new VCFHeader(lines, Collections.singletonList(sample));
    header.setSequenceDictionary(ref.getSequenceDictionary());
    variantContextWriter.writeHeader(header);
    return variantContextWriter;
}
Example #4
Source File: StructuralVariantHeader.java    From hmftools with GNU General Public License v3.0 5 votes vote down vote up 
@NotNull
public static VCFHeader generateHeader(@NotNull final String purpleVersion, @NotNull final VCFHeader template) {
    final VCFHeader outputVCFHeader = new VCFHeader(template.getMetaDataInInputOrder(), template.getGenotypeSamples());
    outputVCFHeader.addMetaDataLine(new VCFHeaderLine("purpleVersion", purpleVersion));

    outputVCFHeader.addMetaDataLine(VCFStandardHeaderLines.getFormatLine("GT"));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(StructuralVariantFactory.RECOVERED,
            0,
            VCFHeaderLineType.Flag,
            RECOVERED_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(StructuralVariantFactory.INFERRED, 0, VCFHeaderLineType.Flag, INFERRED_DESC));
    outputVCFHeader.addMetaDataLine(new VCFFilterHeaderLine(INFERRED, INFERRED_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(StructuralVariantFactory.IMPRECISE,
            0,
            VCFHeaderLineType.Flag,
            IMPRECISE_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(CIPOS, 2, VCFHeaderLineType.Integer, CIPOS_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(SVTYPE, 1, VCFHeaderLineType.String, SVTYPE_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(PURPLE_AF_INFO, UNBOUNDED, VCFHeaderLineType.Float, PURPLE_AF_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(PURPLE_CN_INFO, UNBOUNDED, VCFHeaderLineType.Float, PURPLE_CN_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(RECOVERY_METHOD, 1, VCFHeaderLineType.String, RECOVERY_METHOD_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(RECOVERY_FILTER, UNBOUNDED, VCFHeaderLineType.String, RECOVERY_FILTER_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(PURPLE_JUNCTION_COPY_NUMBER_INFO, 1, VCFHeaderLineType.Float,
            PURPLE_JUNCTION_COPY_NUMBER_DESC));
    outputVCFHeader.addMetaDataLine(new VCFInfoHeaderLine(PURPLE_CN_CHANGE_INFO,
            UNBOUNDED,
            VCFHeaderLineType.Float,
            PURPLE_CN_CHANGE_DESC));
    return outputVCFHeader;
}
Example #5
Source File: GenomicsDBImportIntegrationTest.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
private static File createInputVCF(final String sampleName) {
    final String contig = "chr20";
    final SAMSequenceDictionary dict = new SAMSequenceDictionary(
            Collections.singletonList(new SAMSequenceRecord(contig, 64444167)));

    final VCFFormatHeaderLine formatField = new VCFFormatHeaderLine(SAMPLE_NAME_KEY, 1, VCFHeaderLineType.String,
                                                                    "the name of the sample this genotype came from");
    final Set<VCFHeaderLine> headerLines = new HashSet<>();
    headerLines.add(formatField);
    headerLines.add(new VCFFormatHeaderLine(ANOTHER_ATTRIBUTE_KEY, 1, VCFHeaderLineType.Integer, "Another value"));
    headerLines.add(VCFStandardHeaderLines.getFormatLine("GT"));

    final File out = createTempFile(sampleName +"_", ".vcf");
    try (final VariantContextWriter writer = GATKVariantContextUtils.createVCFWriter(out.toPath(), dict, false,
                                                                                     Options.INDEX_ON_THE_FLY)) {
        final VCFHeader vcfHeader = new VCFHeader(headerLines, Collections.singleton(sampleName));
        vcfHeader.setSequenceDictionary(dict);
        writer.writeHeader(vcfHeader);
        final Allele Aref = Allele.create("A", true);
        final Allele C = Allele.create("C");
        final List<Allele> alleles = Arrays.asList(Aref, C);
        final VariantContext variant = new VariantContextBuilder("invented", contig, INTERVAL.get(0).getStart(), INTERVAL.get(0).getStart(), alleles)
                .genotypes(new GenotypeBuilder(sampleName, alleles).attribute(SAMPLE_NAME_KEY, sampleName)
                                   .attribute(ANOTHER_ATTRIBUTE_KEY, 10).make())
                .make();
        writer.add(variant);
        return out;
    }
}
Example #6
Source File: VariantsSparkSinkUnitTest.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
private static VCFHeader getHeader() {
    final Set<VCFHeaderLine> headerlines = new LinkedHashSet<>();
    VCFStandardHeaderLines.addStandardFormatLines(headerlines, true,
                                                  VCFConstants.GENOTYPE_KEY,
                                                  VCFConstants.GENOTYPE_QUALITY_KEY,
                                                  VCFConstants.GENOTYPE_PL_KEY, VCFConstants.DEPTH_KEY);
    final SAMSequenceDictionary dict = new SAMSequenceDictionary(
            Collections.singletonList(new SAMSequenceRecord("1", 100)));
    final VCFHeader header = new VCFHeader(headerlines, Collections.singleton(SAMPLE));
    header.setSequenceDictionary(dict);
    return header;
}
Example #7
Source File: SVVCFWriter.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
@VisibleForTesting
static VCFHeader getVcfHeader(final SAMSequenceDictionary referenceSequenceDictionary) {
    final Set<VCFHeaderLine> headerLines = new HashSet<>(GATKSVVCFHeaderLines.getSymbAltAlleleLines());
    headerLines.addAll(GATKSVVCFHeaderLines.getInfoLines());
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.END_KEY));
    headerLines.addAll(GATKSVVCFHeaderLines.getFormatLines());
    headerLines.addAll(GATKSVVCFHeaderLines.getFilterLines());
    final VCFHeader header = new VCFHeader(new VCFHeader( headerLines ));
    header.setSequenceDictionary(referenceSequenceDictionary);
    return header;
}
Example #8
Source File: Mutect2Engine.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
public void writeHeader(final VariantContextWriter vcfWriter, final Set<VCFHeaderLine> defaultToolHeaderLines) {
    final Set<VCFHeaderLine> headerInfo = new HashSet<>();
    headerInfo.add(new VCFHeaderLine("MutectVersion", MUTECT_VERSION));
    headerInfo.add(new VCFHeaderLine(FilterMutectCalls.FILTERING_STATUS_VCF_KEY, "Warning: unfiltered Mutect 2 calls.  Please run " + FilterMutectCalls.class.getSimpleName() + " to remove false positives."));
    headerInfo.addAll(annotationEngine.getVCFAnnotationDescriptions(false));
    headerInfo.addAll(defaultToolHeaderLines);
    STANDARD_MUTECT_INFO_FIELDS.stream().map(GATKVCFHeaderLines::getInfoLine).forEach(headerInfo::add);

    VCFStandardHeaderLines.addStandardFormatLines(headerInfo, true,
            VCFConstants.GENOTYPE_KEY,
            VCFConstants.GENOTYPE_ALLELE_DEPTHS,
            VCFConstants.GENOTYPE_QUALITY_KEY,
            VCFConstants.DEPTH_KEY,
            VCFConstants.GENOTYPE_PL_KEY);
    headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.ALLELE_FRACTION_KEY));
    headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_ID_KEY));
    headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_GT_KEY));
    headerInfo.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.PHASE_SET_KEY));

    ReadUtils.getSamplesFromHeader(header).stream().filter(this::isTumorSample)
            .forEach(s -> headerInfo.add(new VCFHeaderLine(TUMOR_SAMPLE_KEY_IN_VCF_HEADER, s)));

    normalSamples.forEach(sample -> headerInfo.add(new VCFHeaderLine(NORMAL_SAMPLE_KEY_IN_VCF_HEADER, sample)));
    if (emitReferenceConfidence()) {
        headerInfo.addAll(referenceConfidenceModel.getVCFHeaderLines());
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.TUMOR_LOG_10_ODDS_KEY));
    }

    final VCFHeader vcfHeader = new VCFHeader(headerInfo, samplesList.asListOfSamples());
    vcfHeader.setSequenceDictionary(header.getSequenceDictionary());
    vcfWriter.writeHeader(vcfHeader);
}
Example #9
Source File: CombineGVCFs.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
private VariantContextWriter getVCFWriter() {
    final SortedSet<String> samples = getSamplesForVariants();

    final VCFHeader inputVCFHeader = new VCFHeader(getHeaderForVariants().getMetaDataInInputOrder(), samples);

    final Set<VCFHeaderLine> headerLines = new LinkedHashSet<>(inputVCFHeader.getMetaDataInInputOrder());
    headerLines.addAll(getDefaultToolVCFHeaderLines());

    headerLines.addAll(annotationEngine.getVCFAnnotationDescriptions());

    // add headers for annotations added by this tool
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.DEPTH_KEY));   // needed for gVCFs without DP tags
    if ( dbsnp.dbsnp != null  ) {
        VCFStandardHeaderLines.addStandardInfoLines(headerLines, true, VCFConstants.DBSNP_KEY);
    }

    if (somaticInput) {
        //single-sample M2 variant filter status will get moved to genotype filter
        headerLines.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_FILTER_KEY));

        if (!dropSomaticFilteringAnnotations) {
            //standard M2 INFO annotations for filtering will get moved to FORMAT field
            for (final String key : Mutect2FilteringEngine.STANDARD_MUTECT_INFO_FIELDS_FOR_FILTERING) {
                headerLines.add(GATKVCFHeaderLines.getEquivalentFormatHeaderLine(key));
            }
        }
    }

    VariantContextWriter writer = createVCFWriter(outputFile);

    final Set<String> sampleNameSet = new IndexedSampleList(samples).asSetOfSamples();
    final VCFHeader vcfHeader = new VCFHeader(headerLines, new TreeSet<>(sampleNameSet));
    writer.writeHeader(vcfHeader);

    return writer;
}
Example #10
Source File: VariantRecalibrationUtils.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
/**
 * Add the standard VCF header lines used with VQSR.
 * @param hInfo updated set of VCFHeaderLines
 */
protected static void addVQSRStandardHeaderLines(final Set<VCFHeaderLine> hInfo) {
    hInfo.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.END_KEY));
    hInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.VQS_LOD_KEY));
    hInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.CULPRIT_KEY));
    hInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.POSITIVE_LABEL_KEY));
    hInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.NEGATIVE_LABEL_KEY));
    hInfo.add(GATKVCFHeaderLines.getFilterLine(VCFConstants.PASSES_FILTERS_v4));
}
Example #11
Source File: GVCFBlockCombiner.java    From gatk with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
public void addRangesToHeader(VCFHeader header) {
    Utils.nonNull(header, "header cannot be null");

    header.addMetaDataLine(VCFStandardHeaderLines.getInfoLine(VCFConstants.END_KEY));
    header.addMetaDataLine(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.MIN_DP_FORMAT_KEY));

    for (final Range<Integer> partition : gqPartitions.asMapOfRanges().keySet()) {
        header.addMetaDataLine(rangeToVCFHeaderLine(partition));
    }
}
Example #12
Source File: HaplotypeCallerEngine.java    From gatk-protected with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
/**
 * Writes an appropriate VCF header, given our arguments, to the provided writer
 *
 * @param vcfWriter writer to which the header should be written
 */
public void writeHeader( final VariantContextWriter vcfWriter, final SAMSequenceDictionary sequenceDictionary,
                         final Set<VCFHeaderLine>  defaultToolHeaderLines) {
    Utils.nonNull(vcfWriter);

    final Set<VCFHeaderLine> headerInfo = new HashSet<>();
    headerInfo.addAll(defaultToolHeaderLines);

    headerInfo.addAll(genotypingEngine.getAppropriateVCFInfoHeaders());
    // all annotation fields from VariantAnnotatorEngine
    headerInfo.addAll(annotationEngine.getVCFAnnotationDescriptions());
    // all callers need to add these standard annotation header lines
    headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.DOWNSAMPLED_KEY));
    headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.MLE_ALLELE_COUNT_KEY));
    headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.MLE_ALLELE_FREQUENCY_KEY));
    // all callers need to add these standard FORMAT field header lines
    VCFStandardHeaderLines.addStandardFormatLines(headerInfo, true,
            VCFConstants.GENOTYPE_KEY,
            VCFConstants.GENOTYPE_QUALITY_KEY,
            VCFConstants.DEPTH_KEY,
            VCFConstants.GENOTYPE_PL_KEY);

    if ( ! hcArgs.doNotRunPhysicalPhasing ) {
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_ID_KEY));
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_GT_KEY));
    }

    // FILTER fields are added unconditionally as it's not always 100% certain the circumstances
    // where the filters are used.  For example, in emitting all sites the lowQual field is used
    headerInfo.add(GATKVCFHeaderLines.getFilterLine(GATKVCFConstants.LOW_QUAL_FILTER_NAME));

    if ( emitReferenceConfidence() ) {
        headerInfo.addAll(referenceConfidenceModel.getVCFHeaderLines());
    }

    final VCFHeader vcfHeader = new VCFHeader(headerInfo, sampleSet);
    vcfHeader.setSequenceDictionary(sequenceDictionary);
    vcfWriter.writeHeader(vcfHeader);
}
Example #13
Source File: RMSMappingQuality.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Override
public List<VCFInfoHeaderLine> getDescriptions() {
    return Arrays.asList(VCFStandardHeaderLines.getInfoLine(getKeyNames().get(0)));
}
Example #14
Source File: DepthPerAlleleBySample.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Override
public List<VCFFormatHeaderLine> getDescriptions() {
    return Collections.singletonList(VCFStandardHeaderLines.getFormatLine(getKeyNames().get(0)));
}
Example #15
Source File: DepthPerSampleHC.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Override
public List<VCFFormatHeaderLine> getDescriptions() {
    return Collections.singletonList(VCFStandardHeaderLines.getFormatLine(VCFConstants.DEPTH_KEY));
}
Example #16
Source File: Coverage.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Override
public List<VCFInfoHeaderLine> getDescriptions() {
    return Collections.singletonList(VCFStandardHeaderLines.getInfoLine(getKeyNames().get(0)));
}
Example #17
Source File: MappingQualityZero.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Override
public List<VCFInfoHeaderLine> getDescriptions() {
    return Collections.singletonList(VCFStandardHeaderLines.getInfoLine(getKeyNames().get(0)));
}
Example #18
Source File: HaplotypeCallerEngine.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
/**
 * Create a VCF header.
 *
 * @param sequenceDictionary sequence dictionary for the reads
 * @return a VCF header
 */
public VCFHeader makeVCFHeader( final SAMSequenceDictionary sequenceDictionary, final Set<VCFHeaderLine>  defaultToolHeaderLines ) {
    final Set<VCFHeaderLine> headerInfo = new HashSet<>();
    headerInfo.addAll(defaultToolHeaderLines);

    headerInfo.addAll(genotypingEngine.getAppropriateVCFInfoHeaders());
    // all annotation fields from VariantAnnotatorEngine
    headerInfo.addAll(annotationEngine.getVCFAnnotationDescriptions(emitReferenceConfidence()));
    // all callers need to add these standard annotation header lines
    headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.MLE_ALLELE_COUNT_KEY));
    headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.MLE_ALLELE_FREQUENCY_KEY));
    // all callers need to add these standard FORMAT field header lines
    VCFStandardHeaderLines.addStandardFormatLines(headerInfo, true,
            VCFConstants.GENOTYPE_KEY,
            VCFConstants.GENOTYPE_QUALITY_KEY,
            VCFConstants.DEPTH_KEY,
            VCFConstants.GENOTYPE_PL_KEY);

    if (hcArgs.standardArgs.genotypeArgs.supportVariants != null) {
        headerInfo.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_COUNT_KEY));
        headerInfo.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_FREQUENCY_KEY));
        headerInfo.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_NUMBER_KEY));
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.PHRED_SCALED_POSTERIORS_KEY));
        headerInfo.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.GENOTYPE_PRIOR_KEY));
    }

    if ( ! hcArgs.doNotRunPhysicalPhasing ) {
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_ID_KEY));
        headerInfo.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.HAPLOTYPE_CALLER_PHASING_GT_KEY));
        headerInfo.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.PHASE_SET_KEY));
    }

    // FILTER fields are added unconditionally as it's not always 100% certain the circumstances
    // where the filters are used.  For example, in emitting all sites the lowQual field is used
    headerInfo.add(GATKVCFHeaderLines.getFilterLine(GATKVCFConstants.LOW_QUAL_FILTER_NAME));

    if ( emitReferenceConfidence() ) {
        headerInfo.addAll(referenceConfidenceModel.getVCFHeaderLines());
    }

    final VCFHeader vcfHeader = new VCFHeader(headerInfo, sampleSet);
    vcfHeader.setSequenceDictionary(sequenceDictionary);
    return vcfHeader;
}
Example #19
Source File: GnarlyGenotyper.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
private void setupVCFWriter(VCFHeader inputVCFHeader, SampleList samples) {
    final Set<VCFHeaderLine> headerLines = new LinkedHashSet<>(inputVCFHeader.getMetaDataInInputOrder());
    headerLines.addAll(getDefaultToolVCFHeaderLines());

    // Remove GCVFBlocks
    headerLines.removeIf(vcfHeaderLine -> vcfHeaderLine.getKey().startsWith(GVCFWriter.GVCF_BLOCK));

    //add header for new filter
    headerLines.add(GATKVCFHeaderLines.getFilterLine(GATKVCFConstants.LOW_QUAL_FILTER_NAME));

    // add headers for annotations added by this tool
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_COUNT_KEY));
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_FREQUENCY_KEY));
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_NUMBER_KEY));
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.ALLELE_NUMBER_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AC_ADJUSTED_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.FISHER_STRAND_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.STRAND_ODDS_RATIO_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.SB_TABLE_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.EXCESS_HET_KEY));
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.QUAL_BY_DEPTH_KEY));
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.RMS_MAPPING_QUALITY_KEY));
    headerLines.add(VCFStandardHeaderLines.getInfoLine(VCFConstants.DEPTH_KEY));   // needed for gVCFs without DP tags
    if (inputVCFHeader.hasInfoLine(GATKVCFConstants.AS_QUAL_KEY) || inputVCFHeader.hasInfoLine(GATKVCFConstants.AS_RAW_QUAL_APPROX_KEY)) {  //use this as a proxy for all AS headers
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_ALT_ALLELE_DEPTH_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_BASE_QUAL_RANK_SUM_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_FISHER_STRAND_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_RMS_MAPPING_QUALITY_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_MAP_QUAL_RANK_SUM_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_QUAL_BY_DEPTH_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_READ_POS_RANK_SUM_KEY));
        headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.AS_STRAND_ODDS_RATIO_KEY));
    }

    if ( dbsnp.dbsnp != null  ) {
        VCFStandardHeaderLines.addStandardInfoLines(headerLines, true, VCFConstants.DBSNP_KEY);
    }

    vcfWriter = createVCFWriter(outputFile);
    if (outputDbName != null) {
        annotationDatabaseWriter = createVCFWriter(new File(outputDbName));
    }

    final Set<String> sampleNameSet = samples.asSetOfSamples();
    headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.RAW_GENOTYPE_COUNT_KEY));
    if (SUMMARIZE_PLs) {
        headerLines.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.REFERENCE_GENOTYPE_QUALITY));
        headerLines.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.GENOTYPE_QUALITY_BY_ALLELE_BALANCE));
        headerLines.add(GATKVCFHeaderLines.getFormatLine(GATKVCFConstants.GENOTYPE_QUALITY_BY_ALT_CONFIDENCE));
    }
    final VCFHeader vcfHeader = new VCFHeader(headerLines, new TreeSet<>(sampleNameSet));
    final VCFHeader dbHeader = new VCFHeader(headerLines);
    vcfWriter.writeHeader(vcfHeader);
    if (outputDbName != null) {
        annotationDatabaseWriter.writeHeader(dbHeader);
    }
}
Example #20
Source File: DepthPerAlleleBySampleUnitTest.java    From gatk with BSD 3-Clause "New" or "Revised" License 4 votes vote down vote up 
@Test
public void testDescription(){
    Assert.assertEquals(new DepthPerAlleleBySample().getKeyNames(), Collections.singletonList(VCFConstants.GENOTYPE_ALLELE_DEPTHS));
    Assert.assertEquals(new DepthPerAlleleBySample().getDescriptions(), Collections.singletonList(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_ALLELE_DEPTHS)));
}
Example #21
Source File: AnnotateStrelkaWithAllelicDepth.java    From hmftools with GNU General Public License v3.0 4 votes vote down vote up 
@NotNull
private VCFHeader generateOutputHeader(@NotNull final VCFHeader template) {
    final VCFHeader outputVCFHeader = new VCFHeader(template.getMetaDataInInputOrder(), template.getGenotypeSamples());
    outputVCFHeader.addMetaDataLine(VCFStandardHeaderLines.getFormatLine("AD"));
    return outputVCFHeader;
}