private byte[] generateHaplotype(
final List<VariantContext> sourceVCs, final ReferenceContext refContext) {
final StringBuilder sb = new StringBuilder();
final int startPos = refContext.getWindow().getStart();
int currentPos = startPos;
final byte[] reference = refContext.getBases();
for (final VariantContext vc : sourceVCs) {
// add any missing reference context
int vcStart = vc.getStart();
final int refAlleleLength = vc.getReference().length();
if (refAlleleLength
== vc.getEnd()
- vc.getStart()) // this is a deletion (whereas for other events the padding base
// isn't part of the position)
vcStart++;
while (currentPos < vcStart) sb.append((char) reference[currentPos++ - startPos]);
// add the alt allele
sb.append(vc.getAlternateAllele(0).getBaseString());
// skip the reference allele
currentPos += refAlleleLength;
}
// add any missing reference context
final int stopPos = refContext.getWindow().getStop();
while (currentPos < stopPos) sb.append((char) reference[currentPos++ - startPos]);
return sb.toString().getBytes();
}
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;
}
private void resolveByHaplotype(final ReferenceContext refContext) {
final byte[] source1Haplotype = generateHaplotype(sourceVCs1, refContext);
final byte[] source2Haplotype = generateHaplotype(sourceVCs2, refContext);
final SWPairwiseAlignment swConsensus1 =
new SWPairwiseAlignment(
refContext.getBases(), source1Haplotype, SW_MATCH, SW_MISMATCH, SW_GAP, SW_GAP_EXTEND);
final SWPairwiseAlignment swConsensus2 =
new SWPairwiseAlignment(
refContext.getBases(), source2Haplotype, SW_MATCH, SW_MISMATCH, SW_GAP, SW_GAP_EXTEND);
// protect against SW failures
if (swConsensus1.getCigar().toString().contains("S")
|| swConsensus1.getCigar().getReferenceLength() < 20
|| swConsensus2.getCigar().toString().contains("S")
|| swConsensus2.getCigar().getReferenceLength() < 20) {
// TODO -- handle errors appropriately
logger.debug("Bad SW alignment; aborting at " + refContext.getLocus());
return;
}
// order results by start position
final TreeMap<Integer, VariantContext> source1Map =
new TreeMap<Integer, VariantContext>(
HaplotypeCallerGenotypingEngine.generateVCsFromAlignment(
new Haplotype(source1Haplotype, false, 0, swConsensus1.getCigar()),
refContext.getBases(),
refContext.getWindow(),
source1));
final TreeMap<Integer, VariantContext> source2Map =
new TreeMap<Integer, VariantContext>(
HaplotypeCallerGenotypingEngine.generateVCsFromAlignment(
new Haplotype(source2Haplotype, false, 0, swConsensus2.getCigar()),
refContext.getBases(),
refContext.getWindow(),
source2));
if (source1Map.size() == 0 || source2Map.size() == 0) {
// TODO -- handle errors appropriately
logger.debug("No source alleles; aborting at " + refContext.getLocus());
return;
}
// create lists and test for equality
final List<VariantContext> source1Alleles = new ArrayList<VariantContext>(source1Map.values());
final List<VariantContext> source2Alleles = new ArrayList<VariantContext>(source2Map.values());
writeAndPurgeAllEqualVariants(source1Alleles, source2Alleles, SAME_BY_HAPLOTYPE_STATUS);
if (source1Alleles.isEmpty()) {
writeAll(source2Alleles, source2, null);
} else if (source2Alleles.isEmpty()) {
writeAll(source1Alleles, source1, null);
} else {
writeDifferences(source1Alleles, source2Alleles);
}
}
@Override
public Long map(
final RefMetaDataTracker tracker,
final ReferenceContext ref,
final AlignmentContext context) {
GenomeLoc refLocus = ref.getLocus();
// process and remove any intervals in the map that are don't overlap the current locus anymore
// and add all new intervals that may overlap this reference locus
addNewOverlappingIntervals(refLocus);
outputFinishedIntervals(refLocus, ref.getBase());
// at this point, all intervals in intervalMap overlap with this locus, so update all of them
for (IntervalStratification intervalStratification : intervalMap.values())
intervalStratification.addLocus(context, ref);
return 1L;
}
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 static double computeGCContent(ReferenceContext ref) {
int gc = 0, at = 0;
for (byte base : ref.getBases()) {
int baseIndex = BaseUtils.simpleBaseToBaseIndex(base);
if (baseIndex == BaseUtils.Base.G.ordinal() || baseIndex == BaseUtils.Base.C.ordinal()) gc++;
else if (baseIndex == BaseUtils.Base.A.ordinal() || baseIndex == BaseUtils.Base.T.ordinal())
at++;
else ; // ignore
}
int sum = gc + at;
return (100.0 * gc) / (sum == 0 ? 1 : sum);
}
@Override
public CallableBaseState map(
RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
CalledState state;
if (BaseUtils.isNBase(ref.getBase())) {
state = CalledState.REF_N;
} else {
// count up the depths of all and QC+ bases
int rawDepth = 0, QCDepth = 0, lowMAPQDepth = 0;
for (PileupElement e : context.getBasePileup()) {
rawDepth++;
if (e.getMappingQual() <= maxLowMAPQ) lowMAPQDepth++;
if (e.getMappingQual() >= minMappingQuality
&& (e.getQual() >= minBaseQuality || e.isDeletion())) {
QCDepth++;
}
}
// System.out.printf("%s rawdepth = %d QCDepth = %d lowMAPQ = %d%n", context.getLocation(),
// rawDepth, QCDepth, lowMAPQDepth);
if (rawDepth == 0) {
state = CalledState.NO_COVERAGE;
} else if (rawDepth >= minDepthLowMAPQ
&& MathUtils.ratio(lowMAPQDepth, rawDepth) >= maxLowMAPQFraction) {
state = CalledState.POOR_MAPPING_QUALITY;
} else if (QCDepth < minDepth) {
state = CalledState.LOW_COVERAGE;
} else if (rawDepth >= maxDepth && maxDepth != -1) {
state = CalledState.EXCESSIVE_COVERAGE;
} else {
state = CalledState.CALLABLE;
}
}
return new CallableBaseState(getToolkit().getGenomeLocParser(), context.getLocation(), state);
}
/**
* For each site of interest, annotate based on the requested annotation types
*
* @param tracker the meta-data tracker
* @param ref the reference base
* @param context the context for the given locus
* @return 1 if the locus was successfully processed, 0 if otherwise
*/
public Integer map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
if (tracker == null) return 0;
// get the variant contexts for all the variants at the location
Collection<VariantContext> VCs =
tracker.getValues(variantCollection.variants, context.getLocation());
if (VCs.isEmpty()) return 0;
Collection<VariantContext> annotatedVCs = VCs;
// if the reference base is not ambiguous, we can annotate
Map<String, AlignmentContext> stratifiedContexts;
if (BaseUtils.simpleBaseToBaseIndex(ref.getBase()) != -1) {
stratifiedContexts = AlignmentContextUtils.splitContextBySampleName(context.getBasePileup());
annotatedVCs = new ArrayList<>(VCs.size());
for (VariantContext vc : VCs)
annotatedVCs.add(engine.annotateContext(tracker, ref, stratifiedContexts, vc));
}
for (VariantContext annotatedVC : annotatedVCs) vcfWriter.add(annotatedVC);
return 1;
}
@Override
public Map<String, Object> annotate(
final RefMetaDataTracker tracker,
final AnnotatorCompatible walker,
final ReferenceContext ref,
final Map<String, AlignmentContext> stratifiedContexts,
final VariantContext vc,
final Map<String, PerReadAlleleLikelihoodMap> stratifiedPerReadAlleleLikelihoodMap) {
// Can not annotate if failed initialization conditions
if (!canAnnotate) return null;
RodBinding<VariantContext> snpEffRodBinding = walker.getSnpEffRodBinding();
// Get only SnpEff records that start at this locus, not merely span it:
List<VariantContext> snpEffRecords = tracker.getValues(snpEffRodBinding, ref.getLocus());
// Within this set, look for a SnpEff record whose ref/alt alleles match the record to annotate.
// If there is more than one such record, we only need to pick the first one, since the
// biological
// effects will be the same across all such records:
VariantContext matchingRecord = getMatchingSnpEffRecord(snpEffRecords, vc);
if (matchingRecord == null) {
return null;
}
// Parse the SnpEff INFO field annotation from the matching record into individual effect
// objects:
List<SnpEffEffect> effects = parseSnpEffRecord(matchingRecord);
if (effects.isEmpty()) {
return null;
}
// Add only annotations for one of the most biologically-significant effects from this set:
SnpEffEffect mostSignificantEffect = getMostSignificantEffect(effects);
return mostSignificantEffect.getAnnotations();
}
public Event map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
boolean hasIndel = false;
boolean hasInsertion = false;
boolean hasPointEvent = false;
int furthestStopPos = -1;
// look at the rods for indels or SNPs
if (tracker != null) {
for (VariantContext vc : tracker.getValues(known)) {
switch (vc.getType()) {
case INDEL:
hasIndel = true;
if (vc.isSimpleInsertion()) hasInsertion = true;
break;
case SNP:
hasPointEvent = true;
break;
case MIXED:
hasPointEvent = true;
hasIndel = true;
if (vc.isSimpleInsertion()) hasInsertion = true;
break;
default:
break;
}
if (hasIndel) furthestStopPos = vc.getEnd();
}
}
// look at the normal context to get deletions and positions with high entropy
final ReadBackedPileup pileup = context.getBasePileup();
int mismatchQualities = 0, totalQualities = 0;
final byte refBase = ref.getBase();
for (PileupElement p : pileup) {
// check the ends of the reads to see how far they extend
furthestStopPos = Math.max(furthestStopPos, p.getRead().getAlignmentEnd());
// is it a deletion or insertion?
if (p.isDeletion() || p.isBeforeInsertion()) {
hasIndel = true;
if (p.isBeforeInsertion()) hasInsertion = true;
}
// look for mismatches
else if (lookForMismatchEntropy) {
if (p.getBase() != refBase) mismatchQualities += p.getQual();
totalQualities += p.getQual();
}
}
// make sure we're supposed to look for high entropy
if (lookForMismatchEntropy
&& pileup.getNumberOfElements() >= minReadsAtLocus
&& (double) mismatchQualities / (double) totalQualities >= mismatchThreshold)
hasPointEvent = true;
// return null if no event occurred
if (!hasIndel && !hasPointEvent) return null;
// return null if we didn't find any usable reads/rods associated with the event
if (furthestStopPos == -1) return null;
GenomeLoc eventLoc = context.getLocation();
if (hasInsertion)
eventLoc =
getToolkit()
.getGenomeLocParser()
.createGenomeLoc(eventLoc.getContig(), eventLoc.getStart(), eventLoc.getStart() + 1);
EVENT_TYPE eventType =
(hasIndel
? (hasPointEvent ? EVENT_TYPE.BOTH : EVENT_TYPE.INDEL_EVENT)
: EVENT_TYPE.POINT_EVENT);
return new Event(eventLoc, furthestStopPos, eventType);
}