/** @{inheritDoc} */
@Override
public SAMRecordIterator query(
final String sequence, final int start, final int end, final boolean contained) {
GenomeLoc region = genomeLocParser.createGenomeLoc(sequence, start, end);
List<SAMRecord> coveredSubset = new ArrayList<SAMRecord>();
for (SAMRecord read : reads) {
GenomeLoc readPosition = genomeLocParser.createGenomeLoc(read);
if (contained && region.containsP(readPosition)) coveredSubset.add(read);
else if (!contained && readPosition.overlapsP(region)) coveredSubset.add(read);
}
final Iterator<SAMRecord> iterator = coveredSubset.iterator();
return new SAMRecordIterator() {
public boolean hasNext() {
return iterator.hasNext();
}
public SAMRecord next() {
return iterator.next();
}
public void close() {}
public void remove() {
iterator.remove();
}
public SAMRecordIterator assertSorted(SAMFileHeader.SortOrder sortOrder) {
return this;
}
};
}
/**
* Adds all intervals that overlap the current reference locus to the intervalMap
*
* @param refLocus the current reference locus
*/
private void addNewOverlappingIntervals(final GenomeLoc refLocus) {
GenomeLoc interval = intervalListIterator.peek();
while (interval != null && !interval.isPast(refLocus)) {
intervalMap.put(interval, createIntervalStatistic(interval));
intervalListIterator.next();
interval = intervalListIterator.peek();
}
}
public boolean isReportableEvent() {
return getToolkit()
.getGenomeLocParser()
.isValidGenomeLoc(loc.getContig(), eventStartPos, eventStopPos, true)
&& eventStopPos >= 0
&& eventStopPos - eventStartPos < maxIntervalSize;
}
/**
* Loop over all of the reads in this likelihood map and realign them to its most likely haplotype
*
* @param haplotypes the collection of haplotypes
* @param paddedReferenceLoc the active region
*/
public void realignReadsToMostLikelyHaplotype(
final Collection<Haplotype> haplotypes, final GenomeLoc paddedReferenceLoc) {
// we need to remap the Alleles back to the Haplotypes; inefficient but unfortunately this is a
// requirement currently
final Map<Allele, Haplotype> alleleToHaplotypeMap = new HashMap<>(haplotypes.size());
Haplotype refHaplotype = null;
for (final Haplotype haplotype : haplotypes) {
alleleToHaplotypeMap.put(Allele.create(haplotype.getBases()), haplotype);
if (refHaplotype == null && haplotype.isReference()) refHaplotype = haplotype;
}
final Map<GATKSAMRecord, Map<Allele, Double>> newLikelihoodReadMap =
new LinkedHashMap<>(likelihoodReadMap.size());
for (final Map.Entry<GATKSAMRecord, Map<Allele, Double>> entry : likelihoodReadMap.entrySet()) {
final MostLikelyAllele bestAllele =
PerReadAlleleLikelihoodMap.getMostLikelyAllele(entry.getValue());
final GATKSAMRecord alignedToRef =
AlignmentUtils.createReadAlignedToRef(
entry.getKey(),
alleleToHaplotypeMap.get(bestAllele.getMostLikelyAllele()),
refHaplotype,
paddedReferenceLoc.getStart(),
bestAllele.isInformative());
newLikelihoodReadMap.put(alignedToRef, entry.getValue());
}
likelihoodReadMap.clear();
likelihoodReadMap.putAll(newLikelihoodReadMap);
}
/**
* Outputs all intervals that are behind the current reference locus
*
* @param refLocus the current reference locus
* @param refBase the reference allele
*/
private void outputFinishedIntervals(final GenomeLoc refLocus, final byte refBase) {
// output any intervals that were finished
final List<GenomeLoc> toRemove = new LinkedList<>();
for (GenomeLoc key : intervalMap.keySet()) {
if (key.isBefore(refLocus)) {
final IntervalStratification intervalStats = intervalMap.get(key);
outputStatsToVCF(intervalStats, Allele.create(refBase, true));
if (hasMissingLoci(intervalStats)) {
outputMissingInterval(intervalStats);
}
toRemove.add(key);
}
}
for (GenomeLoc key : toRemove) {
intervalMap.remove(key);
}
}
public Event(GenomeLoc loc, int furthestStopPos, EVENT_TYPE type) {
this.loc = loc;
this.furthestStopPos = furthestStopPos;
this.type = type;
if (type == EVENT_TYPE.INDEL_EVENT || type == EVENT_TYPE.BOTH) {
eventStartPos = loc.getStart();
eventStopPos = loc.getStop();
} else {
eventStartPos = -1;
eventStopPos = -1;
}
if (type == EVENT_TYPE.POINT_EVENT || type == EVENT_TYPE.BOTH) {
pointEvents.add(loc.getStart());
}
}
private void outputMissingInterval(final IntervalStratification stats) {
final GenomeLoc interval = stats.getInterval();
final boolean missing[] = new boolean[interval.size()];
Arrays.fill(missing, true);
for (AbstractStratification sample : stats.getElements()) {
if (hasMissingStatuses(sample)) {
int pos = 0;
for (AbstractStratification locus : sample.getElements()) {
if (locus.callableStatuses().isEmpty()) {
missing[pos] = false;
}
pos++;
}
}
}
int start = -1;
boolean insideMissing = false;
for (int i = 0; i < missing.length; i++) {
if (missing[i] && !insideMissing) {
start = interval.getStart() + i;
insideMissing = true;
} else if (!missing[i] && insideMissing) {
final int stop = interval.getStart() + i - 1;
outputMissingInterval(interval.getContig(), start, stop);
insideMissing = false;
}
}
if (insideMissing) {
outputMissingInterval(interval.getContig(), start, interval.getStop());
}
}
public int compareTo(Element other) {
if (nextLoc == null) {
if (other.nextLoc != null)
return 1; // null means no more data available, so its after any non-null position
return 0;
}
if (other.nextLoc == null) return -1; // we can get to this point only if this.nextLoc != null
return nextLoc.compareTo(other.nextLoc);
}
private void writeDifferences(
final List<VariantContext> source1Alleles, final List<VariantContext> source2Alleles) {
int currentIndex1 = 0, currentIndex2 = 0;
final int size1 = source1Alleles.size(), size2 = source2Alleles.size();
VariantContext current1 = source1Alleles.get(0);
VariantContext current2 = source2Alleles.get(0);
while (currentIndex1 < size1 || currentIndex2 < size2) {
if (current1 == null) {
writeOne(current2, source2, null);
currentIndex2++;
current2 = (currentIndex2 < size2 ? source2Alleles.get(currentIndex2) : null);
} else if (current2 == null) {
writeOne(current1, source1, null);
currentIndex1++;
current1 = (currentIndex1 < size1 ? source1Alleles.get(currentIndex1) : null);
} else {
final GenomeLoc loc1 = getToolkit().getGenomeLocParser().createGenomeLoc(current1);
final GenomeLoc loc2 = getToolkit().getGenomeLocParser().createGenomeLoc(current2);
if (loc1.getStart() == loc2.getStart() || loc1.overlapsP(loc2)) {
String status;
if (loc1.getStart() == loc2.getStart()) {
final String allele1 = current1.getAlternateAllele(0).getBaseString();
final String allele2 = current2.getAlternateAllele(0).getBaseString();
if (allele1.indexOf(allele2) != -1 || allele2.indexOf(allele1) != -1)
status = ONE_ALLELE_SUBSET_OF_OTHER_STATUS;
else status = SAME_START_DIFFERENT_ALLELES_STATUS;
} else {
status = OVERLAPPING_EVENTS_STATUS;
}
writeOne(current1, INTERSECTION_SET, status);
currentIndex1++;
currentIndex2++;
current1 = (currentIndex1 < size1 ? source1Alleles.get(currentIndex1) : null);
current2 = (currentIndex2 < size2 ? source2Alleles.get(currentIndex2) : null);
} else if (loc1.isBefore(loc2)) {
writeOne(current1, source1, null);
currentIndex1++;
current1 = (currentIndex1 < size1 ? source1Alleles.get(currentIndex1) : null);
} else {
writeOne(current2, source2, null);
currentIndex2++;
current2 = (currentIndex2 < size2 ? source2Alleles.get(currentIndex2) : null);
}
}
}
}
/**
* Takes the interval, finds it in the stash, prints it to the VCF
*
* @param stats The statistics of the interval
* @param refAllele the reference allele
*/
private void outputStatsToVCF(final IntervalStratification stats, final Allele refAllele) {
GenomeLoc interval = stats.getInterval();
final List<Allele> alleles = new ArrayList<>();
final Map<String, Object> attributes = new HashMap<>();
final ArrayList<Genotype> genotypes = new ArrayList<>();
for (String sample : samples) {
final GenotypeBuilder gb = new GenotypeBuilder(sample);
SampleStratification sampleStat = stats.getSampleStatistics(sample);
gb.attribute(
GATKVCFConstants.AVG_INTERVAL_DP_BY_SAMPLE_KEY,
sampleStat.averageCoverage(interval.size()));
gb.attribute(GATKVCFConstants.LOW_COVERAGE_LOCI, sampleStat.getNLowCoveredLoci());
gb.attribute(GATKVCFConstants.ZERO_COVERAGE_LOCI, sampleStat.getNUncoveredLoci());
gb.filters(statusToStrings(stats.getSampleStatistics(sample).callableStatuses(), false));
genotypes.add(gb.make());
}
alleles.add(refAllele);
alleles.add(SYMBOLIC_ALLELE);
VariantContextBuilder vcb =
new VariantContextBuilder(
"DiagnoseTargets",
interval.getContig(),
interval.getStart(),
interval.getStop(),
alleles);
vcb = vcb.log10PError(VariantContext.NO_LOG10_PERROR);
vcb.filters(new LinkedHashSet<>(statusToStrings(stats.callableStatuses(), true)));
attributes.put(VCFConstants.END_KEY, interval.getStop());
attributes.put(GATKVCFConstants.AVG_INTERVAL_DP_KEY, stats.averageCoverage(interval.size()));
attributes.put(GATKVCFConstants.INTERVAL_GC_CONTENT_KEY, stats.gcContent());
vcb = vcb.attributes(attributes);
vcb = vcb.genotypes(genotypes);
vcfWriter.add(vcb.make());
}
private void writeAndPurgeAllEqualVariants(
final List<VariantContext> sourceVCs1,
final List<VariantContext> sourceVCs2,
final String status) {
int currentIndex1 = 0, currentIndex2 = 0;
int size1 = sourceVCs1.size(), size2 = sourceVCs2.size();
VariantContext current1 = (currentIndex1 < size1 ? sourceVCs1.get(currentIndex1) : null);
VariantContext current2 = (currentIndex2 < size2 ? sourceVCs2.get(currentIndex2) : null);
while (current1 != null && current2 != null) {
final GenomeLoc loc1 = getToolkit().getGenomeLocParser().createGenomeLoc(current1);
final GenomeLoc loc2 = getToolkit().getGenomeLocParser().createGenomeLoc(current2);
if (loc1.equals(loc2)
|| (loc1.getStart() == loc2.getStart()
&& (current1.getAlternateAlleles().size() > 1
|| current2.getAlternateAlleles().size() > 1))) {
// test the alleles
if (determineAndWriteOverlap(current1, current2, status)) {
sourceVCs1.remove(currentIndex1);
sourceVCs2.remove(currentIndex2);
size1--;
size2--;
} else {
currentIndex1++;
currentIndex2++;
}
current1 = (currentIndex1 < size1 ? sourceVCs1.get(currentIndex1) : null);
current2 = (currentIndex2 < size2 ? sourceVCs2.get(currentIndex2) : null);
} else if (loc1.isBefore(loc2)) {
currentIndex1++;
current1 = (currentIndex1 < size1 ? sourceVCs1.get(currentIndex1) : null);
} else {
currentIndex2++;
current2 = (currentIndex2 < size2 ? sourceVCs2.get(currentIndex2) : null);
}
}
}
public void onTraversalDone(EventPair sum) {
if (sum.left != null && sum.left.isReportableEvent()) sum.intervals.add(sum.left.getLoc());
if (sum.right != null && sum.right.isReportableEvent()) sum.intervals.add(sum.right.getLoc());
if (FilenameUtils.getExtension(out.getName()).equals("interval_list")) {
final SAMFileHeader masterSequenceDictionaryHeader = new SAMFileHeader();
masterSequenceDictionaryHeader.setSequenceDictionary(
getToolkit().getMasterSequenceDictionary());
final IntervalList intervalList = new IntervalList(masterSequenceDictionaryHeader);
for (GenomeLoc loc : sum.intervals) {
intervalList.add(new Interval(loc.getContig(), loc.getStart(), loc.getStop()));
}
intervalList.write(out);
} else {
try (BufferedWriter bufferedWriter = IOUtil.openFileForBufferedWriting(out)) {
for (GenomeLoc loc : sum.intervals) {
bufferedWriter.write(loc.toString());
bufferedWriter.newLine();
}
} catch (final IOException e) {
throw new GATKException("Error writing out intervals to file: " + out.getAbsolutePath(), e);
}
}
}
public String toString() {
return String.format(
"%s\t%d\t%d\t%s", loc.getContig(), loc.getStart() - 1, loc.getStop(), state);
}
/**
* Updating the location of this CalledBaseState by the new stop location
*
* @param newStop
*/
public void update(GenomeLoc newStop) {
loc = genomeLocParser.createGenomeLoc(loc.getContig(), loc.getStart(), newStop.getStop());
}
public GenomeLoc getLoc() {
return getToolkit()
.getGenomeLocParser()
.createGenomeLoc(loc.getContig(), eventStartPos, eventStopPos);
}
/**
* can we flash back to the specified location?
*
* @param location the location to try and flash back to
* @return true if we can, false otherwise
*/
public boolean canFlashBackTo(GenomeLoc location) {
GenomeLoc farthestBack =
(pastQueue.size() > 0) ? pastQueue.getFirst().getLocation() : iterator.peekNextLocation();
return (!farthestBack.isPast(location));
}
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);
}