/**
* Helper method to subset a VC record, modifying some metadata stored in the INFO field (i.e. AN,
* AC, AF).
*
* @param vc the VariantContext record to subset
* @param samples the samples to extract
* @return the subsetted VariantContext
*/
private VariantContext subsetRecord(VariantContext vc, Set<String> samples) {
if (samples == null || samples.isEmpty()) return vc;
ArrayList<Genotype> genotypes = new ArrayList<Genotype>();
for (Map.Entry<String, Genotype> genotypePair : vc.getGenotypes().entrySet()) {
if (samples.contains(genotypePair.getKey())) genotypes.add(genotypePair.getValue());
}
VariantContext sub = vc.subContextFromGenotypes(genotypes, vc.getAlleles());
// if we have fewer alternate alleles in the selected VC than in the original VC, we need to
// strip out the GL/PLs (because they are no longer accurate)
if (vc.getAlleles().size() != sub.getAlleles().size())
sub = VariantContext.modifyGenotypes(sub, VariantContextUtils.stripPLs(vc.getGenotypes()));
HashMap<String, Object> attributes = new HashMap<String, Object>(sub.getAttributes());
int depth = 0;
for (String sample : sub.getSampleNames()) {
Genotype g = sub.getGenotype(sample);
if (g.isNotFiltered() && g.isCalled()) {
String dp = (String) g.getAttribute("DP");
if (dp != null
&& !dp.equals(VCFConstants.MISSING_DEPTH_v3)
&& !dp.equals(VCFConstants.MISSING_VALUE_v4)) {
depth += Integer.valueOf(dp);
}
}
}
if (KEEP_ORIGINAL_CHR_COUNTS) {
if (attributes.containsKey(VCFConstants.ALLELE_COUNT_KEY))
attributes.put("AC_Orig", attributes.get(VCFConstants.ALLELE_COUNT_KEY));
if (attributes.containsKey(VCFConstants.ALLELE_FREQUENCY_KEY))
attributes.put("AF_Orig", attributes.get(VCFConstants.ALLELE_FREQUENCY_KEY));
if (attributes.containsKey(VCFConstants.ALLELE_NUMBER_KEY))
attributes.put("AN_Orig", attributes.get(VCFConstants.ALLELE_NUMBER_KEY));
}
VariantContextUtils.calculateChromosomeCounts(sub, attributes, false);
attributes.put("DP", depth);
sub = VariantContext.modifyAttributes(sub, attributes);
return sub;
}
/**
* Subset VC record if necessary and emit the modified record (provided it satisfies criteria for
* printing)
*
* @param tracker the ROD tracker
* @param ref reference information
* @param context alignment info
* @return 1 if the record was printed to the output file, 0 if otherwise
*/
@Override
public Integer map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
if (tracker == null) return 0;
Collection<VariantContext> vcs =
tracker.getValues(variantCollection.variants, context.getLocation());
if (vcs == null || vcs.size() == 0) {
return 0;
}
for (VariantContext vc : vcs) {
if (MENDELIAN_VIOLATIONS) {
boolean foundMV = false;
for (MendelianViolation mv : mvSet) {
if (mv.isViolation(vc)) {
foundMV = true;
// System.out.println(vc.toString());
if (outMVFile != null)
outMVFileStream.format(
"MV@%s:%d. REF=%s, ALT=%s, AC=%d, momID=%s, dadID=%s, childID=%s, momG=%s, momGL=%s, dadG=%s, dadGL=%s, "
+ "childG=%s childGL=%s\n",
vc.getChr(),
vc.getStart(),
vc.getReference().getDisplayString(),
vc.getAlternateAllele(0).getDisplayString(),
vc.getChromosomeCount(vc.getAlternateAllele(0)),
mv.getSampleMom(),
mv.getSampleDad(),
mv.getSampleChild(),
vc.getGenotype(mv.getSampleMom()).toBriefString(),
vc.getGenotype(mv.getSampleMom()).getLikelihoods().getAsString(),
vc.getGenotype(mv.getSampleDad()).toBriefString(),
vc.getGenotype(mv.getSampleMom()).getLikelihoods().getAsString(),
vc.getGenotype(mv.getSampleChild()).toBriefString(),
vc.getGenotype(mv.getSampleChild()).getLikelihoods().getAsString());
}
}
if (!foundMV) break;
}
if (DISCORDANCE_ONLY) {
Collection<VariantContext> compVCs =
tracker.getValues(discordanceTrack, context.getLocation());
if (!isDiscordant(vc, compVCs)) return 0;
}
if (CONCORDANCE_ONLY) {
Collection<VariantContext> compVCs =
tracker.getValues(concordanceTrack, context.getLocation());
if (!isConcordant(vc, compVCs)) return 0;
}
if (alleleRestriction.equals(NumberAlleleRestriction.BIALLELIC) && !vc.isBiallelic())
continue;
if (alleleRestriction.equals(NumberAlleleRestriction.MULTIALLELIC) && vc.isBiallelic())
continue;
if (!selectedTypes.contains(vc.getType())) continue;
VariantContext sub = subsetRecord(vc, samples);
if ((sub.isPolymorphic() || !EXCLUDE_NON_VARIANTS)
&& (!sub.isFiltered() || !EXCLUDE_FILTERED)) {
for (VariantContextUtils.JexlVCMatchExp jexl : jexls) {
if (!VariantContextUtils.match(sub, jexl)) {
return 0;
}
}
if (SELECT_RANDOM_NUMBER) {
randomlyAddVariant(++variantNumber, sub, ref.getBase());
} else if (!SELECT_RANDOM_FRACTION
|| (GenomeAnalysisEngine.getRandomGenerator().nextDouble() < fractionRandom)) {
vcfWriter.add(sub);
}
}
}
return 1;
}
/** Set up the VCF writer, the sample expressions and regexs, and the JEXL matcher */
public void initialize() {
// Get list of samples to include in the output
List<String> rodNames = Arrays.asList(variantCollection.variants.getName());
Map<String, VCFHeader> vcfRods = VCFUtils.getVCFHeadersFromRods(getToolkit(), rodNames);
TreeSet<String> vcfSamples =
new TreeSet<String>(
SampleUtils.getSampleList(
vcfRods, VariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE));
Collection<String> samplesFromFile = SampleUtils.getSamplesFromFiles(sampleFiles);
Collection<String> samplesFromExpressions =
SampleUtils.matchSamplesExpressions(vcfSamples, sampleExpressions);
// first, add any requested samples
samples.addAll(samplesFromFile);
samples.addAll(samplesFromExpressions);
samples.addAll(sampleNames);
// if none were requested, we want all of them
if (samples.isEmpty()) {
samples.addAll(vcfSamples);
NO_SAMPLES_SPECIFIED = true;
}
// now, exclude any requested samples
Collection<String> XLsamplesFromFile = SampleUtils.getSamplesFromFiles(XLsampleFiles);
samples.removeAll(XLsamplesFromFile);
samples.removeAll(XLsampleNames);
if (samples.size() == 0 && !NO_SAMPLES_SPECIFIED)
throw new UserException(
"All samples requested to be included were also requested to be excluded.");
for (String sample : samples) logger.info("Including sample '" + sample + "'");
// if user specified types to include, add these, otherwise, add all possible variant context
// types to list of vc types to include
if (TYPES_TO_INCLUDE.isEmpty()) {
for (VariantContext.Type t : VariantContext.Type.values()) selectedTypes.add(t);
} else {
for (VariantContext.Type t : TYPES_TO_INCLUDE) selectedTypes.add(t);
}
// Initialize VCF header
Set<VCFHeaderLine> headerLines = VCFUtils.smartMergeHeaders(vcfRods.values(), logger);
headerLines.add(new VCFHeaderLine("source", "SelectVariants"));
if (KEEP_ORIGINAL_CHR_COUNTS) {
headerLines.add(
new VCFFormatHeaderLine("AC_Orig", 1, VCFHeaderLineType.Integer, "Original AC"));
headerLines.add(
new VCFFormatHeaderLine("AF_Orig", 1, VCFHeaderLineType.Float, "Original AF"));
headerLines.add(
new VCFFormatHeaderLine("AN_Orig", 1, VCFHeaderLineType.Integer, "Original AN"));
}
vcfWriter.writeHeader(new VCFHeader(headerLines, samples));
for (int i = 0; i < SELECT_EXPRESSIONS.size(); i++) {
// It's not necessary that the user supply select names for the JEXL expressions, since those
// expressions will only be needed for omitting records. Make up the select names here.
selectNames.add(String.format("select-%d", i));
}
jexls = VariantContextUtils.initializeMatchExps(selectNames, SELECT_EXPRESSIONS);
// Look at the parameters to decide which analysis to perform
DISCORDANCE_ONLY = discordanceTrack.isBound();
if (DISCORDANCE_ONLY)
logger.info(
"Selecting only variants discordant with the track: " + discordanceTrack.getName());
CONCORDANCE_ONLY = concordanceTrack.isBound();
if (CONCORDANCE_ONLY)
logger.info(
"Selecting only variants concordant with the track: " + concordanceTrack.getName());
if (MENDELIAN_VIOLATIONS) {
if (FAMILY_STRUCTURE_FILE != null) {
try {
for (final String line : new XReadLines(FAMILY_STRUCTURE_FILE)) {
MendelianViolation mv =
new MendelianViolation(line, MENDELIAN_VIOLATION_QUAL_THRESHOLD);
if (samples.contains(mv.getSampleChild())
&& samples.contains(mv.getSampleDad())
&& samples.contains(mv.getSampleMom())) mvSet.add(mv);
}
} catch (FileNotFoundException e) {
throw new UserException.CouldNotReadInputFile(FAMILY_STRUCTURE_FILE, e);
}
if (outMVFile != null)
try {
outMVFileStream = new PrintStream(outMVFile);
} catch (FileNotFoundException e) {
throw new UserException.CouldNotCreateOutputFile(
outMVFile, "Can't open output file", e);
}
} else
mvSet.add(new MendelianViolation(FAMILY_STRUCTURE, MENDELIAN_VIOLATION_QUAL_THRESHOLD));
} else if (!FAMILY_STRUCTURE.isEmpty()) {
mvSet.add(new MendelianViolation(FAMILY_STRUCTURE, MENDELIAN_VIOLATION_QUAL_THRESHOLD));
MENDELIAN_VIOLATIONS = true;
}
SELECT_RANDOM_NUMBER = numRandom > 0;
if (SELECT_RANDOM_NUMBER) {
logger.info("Selecting " + numRandom + " variants at random from the variant track");
variantArray = new RandomVariantStructure[numRandom];
}
SELECT_RANDOM_FRACTION = fractionRandom > 0;
if (SELECT_RANDOM_FRACTION)
logger.info(
"Selecting approximately "
+ 100.0 * fractionRandom
+ "% of the variants at random from the variant track");
}