private VariantContext getDbsnp(String rsID) {
if (dbsnpIterator == null) {
if (dbsnp == null)
throw new UserException.BadInput(
"No dbSNP rod was provided, but one is needed to decipher the correct indel alleles from the HapMap records");
RMDTrackBuilder builder =
new RMDTrackBuilder(
getToolkit().getReferenceDataSource().getReference().getSequenceDictionary(),
getToolkit().getGenomeLocParser(),
getToolkit().getArguments().unsafe,
getToolkit().getArguments().disableAutoIndexCreationAndLockingWhenReadingRods,
null);
dbsnpIterator =
builder
.createInstanceOfTrack(VCFCodec.class, new File(dbsnp.dbsnp.getSource()))
.getIterator();
// Note that we should really use some sort of seekable iterator here so that the search
// doesn't take forever
// (but it's complicated because the hapmap location doesn't match the dbsnp location, so we
// don't know where to seek to)
}
while (dbsnpIterator.hasNext()) {
GATKFeature feature = dbsnpIterator.next();
VariantContext vc = (VariantContext) feature.getUnderlyingObject();
if (vc.getID().equals(rsID)) return vc;
}
return null;
}
private Collection<VariantContext> getVariantContexts(
RefMetaDataTracker tracker, ReferenceContext ref) {
List<Feature> features = tracker.getValues(variants, ref.getLocus());
List<VariantContext> VCs = new ArrayList<VariantContext>(features.size());
for (Feature record : features) {
if (VariantContextAdaptors.canBeConvertedToVariantContext(record)) {
// we need to special case the HapMap format because indels aren't handled correctly
if (record instanceof RawHapMapFeature) {
// is it an indel?
RawHapMapFeature hapmap = (RawHapMapFeature) record;
if (hapmap.getAlleles()[0].equals(RawHapMapFeature.NULL_ALLELE_STRING)
|| hapmap.getAlleles()[1].equals(RawHapMapFeature.NULL_ALLELE_STRING)) {
// get the dbsnp object corresponding to this record (needed to help us distinguish
// between insertions and deletions)
VariantContext dbsnpVC = getDbsnp(hapmap.getName());
if (dbsnpVC == null || dbsnpVC.isMixed()) continue;
Map<String, Allele> alleleMap = new HashMap<String, Allele>(2);
alleleMap.put(
RawHapMapFeature.DELETION,
Allele.create(ref.getBase(), dbsnpVC.isSimpleInsertion()));
alleleMap.put(
RawHapMapFeature.INSERTION,
Allele.create(
(char) ref.getBase() + ((RawHapMapFeature) record).getAlleles()[1],
!dbsnpVC.isSimpleInsertion()));
hapmap.setActualAlleles(alleleMap);
// also, use the correct positioning for insertions
hapmap.updatePosition(dbsnpVC.getStart());
if (hapmap.getStart() < ref.getWindow().getStart()) {
logger.warn(
"Hapmap record at "
+ ref.getLocus()
+ " represents an indel too large to be converted; skipping...");
continue;
}
}
}
// ok, we might actually be able to turn this record in a variant context
VariantContext vc =
VariantContextAdaptors.toVariantContext(variants.getName(), record, ref);
if (vc != null) // sometimes the track has odd stuff in it that can't be converted
VCs.add(vc);
}
}
return VCs;
}
public Integer map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
if (tracker == null || !BaseUtils.isRegularBase(ref.getBase())) return 0;
Collection<VariantContext> contexts = getVariantContexts(tracker, ref);
for (VariantContext vc : contexts) {
VariantContextBuilder builder = new VariantContextBuilder(vc);
// set the appropriate sample name if necessary
if (sampleName != null && vc.hasGenotypes() && vc.hasGenotype(variants.getName())) {
Genotype g =
new GenotypeBuilder(vc.getGenotype(variants.getName())).name(sampleName).make();
builder.genotypes(g);
}
final VariantContext withID = variantOverlapAnnotator.annotateRsID(tracker, builder.make());
writeRecord(withID, tracker, ref.getLocus());
}
return 1;
}
private void writeRecord(VariantContext vc, RefMetaDataTracker tracker, GenomeLoc loc) {
if (!wroteHeader) {
wroteHeader = true;
// setup the header fields
Set<VCFHeaderLine> hInfo = new HashSet<VCFHeaderLine>();
hInfo.addAll(GATKVCFUtils.getHeaderFields(getToolkit(), Arrays.asList(variants.getName())));
hInfo.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_KEY));
allowedGenotypeFormatStrings.add(VCFConstants.GENOTYPE_KEY);
for (VCFHeaderLine field : hInfo) {
if (field instanceof VCFFormatHeaderLine) {
allowedGenotypeFormatStrings.add(((VCFFormatHeaderLine) field).getID());
}
}
samples = new LinkedHashSet<String>();
if (sampleName != null) {
samples.add(sampleName);
} else {
// try VCF first
samples =
SampleUtils.getSampleListWithVCFHeader(getToolkit(), Arrays.asList(variants.getName()));
if (samples.isEmpty()) {
List<Feature> features = tracker.getValues(variants, loc);
if (features.size() == 0)
throw new IllegalStateException(
"No rod data is present, but we just created a VariantContext");
Feature f = features.get(0);
if (f instanceof RawHapMapFeature)
samples.addAll(Arrays.asList(((RawHapMapFeature) f).getSampleIDs()));
else samples.addAll(vc.getSampleNames());
}
}
vcfwriter.writeHeader(new VCFHeader(hInfo, samples));
}
vc = GATKVariantContextUtils.purgeUnallowedGenotypeAttributes(vc, allowedGenotypeFormatStrings);
vcfwriter.add(vc);
}