HistogramGenerator addRead(final GATKRead read) {
final byte[] quals =
(useOriginalQualities
? ReadUtils.getOriginalBaseQualities(read)
: read.getBaseQualities());
if (quals == null) {
return this;
}
final int length = quals.length;
final boolean isReverseStrand = read.isReverseStrand();
ensureArraysBigEnough(length + 1);
for (int i = 0; i < length; ++i) {
final int cycle = isReverseStrand ? length - i : i + 1;
if (read.isPaired() && read.isSecondOfPair()) {
secondReadTotalsByCycle[cycle] += quals[i];
secondReadCountsByCycle[cycle] += 1;
} else {
firstReadTotalsByCycle[cycle] += quals[i];
firstReadCountsByCycle[cycle] += 1;
}
}
return this;
}
public void setRead(final GATKRead read) {
if (!read.isEmpty()) {
this.read = read;
if (!read.isUnmapped())
loc =
genomeLocParser.createGenomeLoc(
read.getContig(), ReadUtils.getSoftStart(read), ReadUtils.getSoftEnd(read));
}
}
@Override
public void processElement(ProcessContext c) throws Exception {
String dest = c.element();
logger.info("Saving to " + dest);
Iterable<GATKRead> reads = c.sideInput(iterableView);
OutputStream outputStream = BucketUtils.createFile(dest, c.getPipelineOptions());
try (SAMFileWriter writer =
new SAMFileWriterFactory().makeBAMWriter(header, false, outputStream)) {
for (GATKRead r : reads) {
final SAMRecord sr = r.convertToSAMRecord(header);
writer.addAlignment(sr);
}
}
}
@Override
public void recordValues(
final GATKRead read, final SAMFileHeader header, final ReadCovariates values) {
final SAMReadGroupRecord rg = ReadUtils.getSAMReadGroupRecord(read, header);
final String readGroupId = getID(rg);
final int key = keyForReadGroup(readGroupId);
final int readLength = read.getLength();
for (int i = 0; i < readLength; i++) {
values.addCovariate(key, key, key, i);
}
}
/**
* Add a read to the manager
*
* @param read the read to add
*/
public void addRead(final GATKRead read) {
if (read == null)
throw new IllegalArgumentException("read added to manager is null, which is not allowed");
// if the new read is on a different contig or we have too many reads, then we need to flush the
// queue and clear the map
final boolean tooManyReads = getNReadsInQueue() >= MAX_RECORDS_IN_MEMORY;
final boolean encounteredNewContig =
getNReadsInQueue() > 0 && !waitingReads.peek().read.getContig().equals(read.getContig());
if (tooManyReads || encounteredNewContig) {
if (DEBUG)
logger.warn(
"Flushing queue on "
+ (tooManyReads
? "too many reads"
: ("move to new contig: "
+ read.getContig()
+ " from "
+ waitingReads.peek().read.getContig()))
+ " at "
+ read.getStart());
final int targetQueueSize = encounteredNewContig ? 0 : MAX_RECORDS_IN_MEMORY / 2;
// write the required number of waiting reads to disk
while (getNReadsInQueue() > targetQueueSize) writer.addRead(waitingReads.poll().read);
}
final SplitRead splitRead = new SplitRead(read);
// fix overhangs, as needed
for (final Splice splice : splices) fixSplit(splitRead, splice);
// add the new read to the queue
waitingReads.add(splitRead);
}
@Test
public void testPerReadAlleleLikelihoodMap() {
final PerReadAlleleLikelihoodMap map = new PerReadAlleleLikelihoodMap();
final Allele alleleA = Allele.create("A");
final double lik = -1.0; // ignored
final int[] MQs = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, QualityUtils.MAPPING_QUALITY_UNAVAILABLE};
final List<Integer> MQsList = Arrays.asList(ArrayUtils.toObject(MQs));
// MQ 255 are excluded from the calculations, we test it here.
final List<Integer> MQsListOK = new ArrayList<>(MQsList);
// NOTE: if we just call remove(i), Java thinks i is an index.
// A workaround for this overloading bogosity to to call removeAll and pass a collection
// (casting i to (Object) would work too but it's more error prone)
MQsListOK.removeAll(Collections.singleton(QualityUtils.MAPPING_QUALITY_UNAVAILABLE));
final int n1A = MQs.length;
for (int i = 0; i < n1A; i++) {
final GATKRead read = ArtificialReadUtils.createArtificialRead(TextCigarCodec.decode("10M"));
read.setMappingQuality(MQs[i]);
map.add(read, alleleA, lik);
}
final Map<String, PerReadAlleleLikelihoodMap> perReadAlleleLikelihoodMap =
Collections.singletonMap("sample1", map);
final VariantContext vc = makeVC();
final ReferenceContext referenceContext = null;
final Map<String, Object> annotate =
new RMSMappingQuality().annotate(referenceContext, vc, perReadAlleleLikelihoodMap);
Assert.assertEquals(annotate.size(), 1, "size");
Assert.assertEquals(
annotate.keySet(), Collections.singleton(VCFConstants.RMS_MAPPING_QUALITY_KEY), "annots");
final double rms = MathUtils.rms(MQsListOK); // only those are MQ0
Assert.assertEquals(
annotate.get(VCFConstants.RMS_MAPPING_QUALITY_KEY), String.format("%.2f", rms));
}
