/**
* Update the recalibration statistics using the information in recalInfo
*
* @param recalInfo data structure holding information about the recalibration values for a single
* read
*/
@Requires("recalInfo != null")
public void updateDataForRead(final ReadRecalibrationInfo recalInfo) {
final GATKSAMRecord read = recalInfo.getRead();
final ReadCovariates readCovariates = recalInfo.getCovariatesValues();
final RecalibrationTables tables = getUpdatableRecalibrationTables();
final NestedIntegerArray<RecalDatum> qualityScoreTable = tables.getQualityScoreTable();
for (int offset = 0; offset < read.getReadBases().length; offset++) {
if (!recalInfo.skip(offset)) {
for (final EventType eventType : EventType.values()) {
final int[] keys = readCovariates.getKeySet(offset, eventType);
final int eventIndex = eventType.ordinal();
final byte qual = recalInfo.getQual(eventType, offset);
final double isError = recalInfo.getErrorFraction(eventType, offset);
RecalUtils.incrementDatumOrPutIfNecessary(
qualityScoreTable, qual, isError, keys[0], keys[1], eventIndex);
for (int i = 2; i < covariates.length; i++) {
if (keys[i] < 0) continue;
RecalUtils.incrementDatumOrPutIfNecessary(
tables.getTable(i), qual, isError, keys[0], keys[1], keys[i], eventIndex);
}
}
}
}
}
@Test(expectedExceptions = IllegalStateException.class)
public void testStrandlessReadsFailSetStrand() {
final byte[] bases = {'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A'};
final byte[] quals = {20, 20, 20, 20, 20, 20, 20, 20};
GATKSAMRecord read = ArtificialSAMUtils.createArtificialRead(bases, quals, "6M");
read.setIsStrandless(true);
read.setReadNegativeStrandFlag(true);
}
@BeforeClass
public void init() {
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
read = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, BASES.length());
read.setReadUnmappedFlag(true);
read.setReadBases(new String(BASES).getBytes());
read.setBaseQualityString(new String(QUALS));
}
@Test
public void realignAtContigBorderTest() {
final int contigEnd = header.getSequence(0).getSequenceLength();
final GATKSAMRecord read =
ArtificialSAMUtils.createArtificialRead(header, "goodRead", 0, contigEnd - 1, 2);
read.setCigarString("2M");
Assert.assertEquals(IndelRealigner.realignmentProducesBadAlignment(read, contigEnd), false);
read.setCigarString("1M1D1M");
Assert.assertEquals(IndelRealigner.realignmentProducesBadAlignment(read, contigEnd), true);
}
private void verifySortednessOfReads(final List<GATKSAMRecord> reads) {
int lastStart = -1;
for (GATKSAMRecord read : reads) {
Assert.assertTrue(
lastStart <= read.getAlignmentStart(),
"Reads should be sorted but weren't. Found read with start "
+ read.getAlignmentStart()
+ " while last was "
+ lastStart);
lastStart = read.getAlignmentStart();
}
}
@Override
protected Double getElementForRead(final GATKSAMRecord read, final int refLoc) {
final int offset =
ReadUtils.getReadCoordinateForReferenceCoordinate(
read.getSoftStart(), read.getCigar(), refLoc, ReadUtils.ClippingTail.RIGHT_TAIL, true);
if (offset == ReadUtils.CLIPPING_GOAL_NOT_REACHED) return null;
int readPos = AlignmentUtils.calcAlignmentByteArrayOffset(read.getCigar(), offset, false, 0, 0);
final int numAlignedBases = AlignmentUtils.getNumAlignedBasesCountingSoftClips(read);
if (readPos > numAlignedBases / 2) readPos = numAlignedBases - (readPos + 1);
return (double) readPos;
}
/**
* Shallow copy of everything, except for the attribute list and the temporary attributes. A new
* list of the attributes is created for both, but the attributes themselves are copied by
* reference. This should be safe because callers should never modify a mutable value returned by
* any of the get() methods anyway.
*
* @return a shallow copy of the GATKSAMRecord
*/
@Override
public Object clone() {
try {
final GATKSAMRecord clone = (GATKSAMRecord) super.clone();
if (temporaryAttributes != null) {
clone.temporaryAttributes = new HashMap<>();
for (Object attribute : temporaryAttributes.keySet())
clone.setTemporaryAttribute(attribute, temporaryAttributes.get(attribute));
}
return clone;
} catch (final CloneNotSupportedException e) {
throw new RuntimeException(e);
}
}
@Override
protected Double getElementForRead(final GATKSAMRecord read, final int refLoc) {
return (double)
read.getBaseQualities()[
ReadUtils.getReadCoordinateForReferenceCoordinateUpToEndOfRead(
read, refLoc, ReadUtils.ClippingTail.RIGHT_TAIL)];
}
private GATKSAMRecord revertSoftClippedBases(GATKSAMRecord read) {
GATKSAMRecord unclipped = (GATKSAMRecord) read.clone();
Cigar unclippedCigar = new Cigar();
int matchesCount = 0;
for (CigarElement element : read.getCigar().getCigarElements()) {
if (element.getOperator() == CigarOperator.SOFT_CLIP
|| element.getOperator() == CigarOperator.MATCH_OR_MISMATCH)
matchesCount += element.getLength();
else if (matchesCount > 0) {
unclippedCigar.add(new CigarElement(matchesCount, CigarOperator.MATCH_OR_MISMATCH));
matchesCount = 0;
unclippedCigar.add(element);
} else unclippedCigar.add(element);
}
if (matchesCount > 0)
unclippedCigar.add(new CigarElement(matchesCount, CigarOperator.MATCH_OR_MISMATCH));
unclipped.setCigar(unclippedCigar);
final int newStart =
read.getAlignmentStart() + calculateAlignmentStartShift(read.getCigar(), unclippedCigar);
unclipped.setAlignmentStart(newStart);
if (newStart <= 0) {
// if the start of the unclipped read occurs before the contig,
// we must hard clip away the bases since we cannot represent reads with
// negative or 0 alignment start values in the SAMRecord (e.g., 0 means unaligned)
return hardClip(unclipped, 0, -newStart);
} else {
return unclipped;
}
}
@Override
public Integer map(
ReferenceContext referenceContext,
GATKSAMRecord read,
RefMetaDataTracker RefMetaDataTracker) {
final String rgID = read.getReadGroup().getId();
final PerReadGroupInfo info = readGroupInfo.get(rgID);
if (info.needsMoreData()) {
info.readLength.add(read.getReadLength());
info.nReadsSeen++;
if (read.getReadPairedFlag()) {
info.nReadsPaired++;
if (read.getInferredInsertSize() != 0) {
info.insertSize.add(Math.abs(read.getInferredInsertSize()));
}
}
}
return null;
}
/** Ensure that splitting read groups still works when dealing with null read groups. */
@Test
public void testSplitByNullReadGroups() {
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
GATKSAMRecord read3 = ArtificialSAMUtils.createArtificialRead(header, "read3", 0, 1, 10);
ReadBackedPileup pileup =
new ReadBackedPileupImpl(null, Arrays.asList(read1, read2, read3), Arrays.asList(1, 1, 1));
ReadBackedPileup nullRgPileup = pileup.getPileupForReadGroup(null);
List<GATKSAMRecord> nullRgReads = nullRgPileup.getReads();
Assert.assertEquals(
nullRgPileup.getNumberOfElements(), 3, "Wrong number of reads in null read group");
Assert.assertEquals(
nullRgReads.get(0),
read1,
"Read " + read1.getReadName() + " should be in null rg but isn't");
Assert.assertEquals(
nullRgReads.get(1),
read2,
"Read " + read2.getReadName() + " should be in null rg but isn't");
Assert.assertEquals(
nullRgReads.get(2),
read3,
"Read " + read3.getReadName() + " should be in null rg but isn't");
ReadBackedPileup rg1Pileup = pileup.getPileupForReadGroup("rg1");
Assert.assertNull(rg1Pileup, "Pileup for non-existent read group should return null");
}
@Test
public void testGetPileupForSample() {
String sample1 = "sample1";
String sample2 = "sample2";
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
readGroupOne.setSample(sample1);
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
readGroupTwo.setSample(sample2);
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
Map<String, ReadBackedPileupImpl> sampleToPileupMap =
new HashMap<String, ReadBackedPileupImpl>();
sampleToPileupMap.put(
sample1, new ReadBackedPileupImpl(null, Collections.singletonList(read1), 0));
sampleToPileupMap.put(
sample2, new ReadBackedPileupImpl(null, Collections.singletonList(read2), 0));
ReadBackedPileup pileup = new ReadBackedPileupImpl(null, sampleToPileupMap);
ReadBackedPileup sample2Pileup = pileup.getPileupForSample(sample2);
Assert.assertEquals(
sample2Pileup.getNumberOfElements(), 1, "Sample 2 pileup has wrong number of elements");
Assert.assertEquals(
sample2Pileup.getReads().get(0), read2, "Sample 2 pileup has incorrect read");
ReadBackedPileup missingSamplePileup = pileup.getPileupForSample("missing");
Assert.assertNull(missingSamplePileup, "Pileup for sample 'missing' should be null but isn't");
missingSamplePileup = pileup.getPileupForSample("not here");
Assert.assertNull(missingSamplePileup, "Pileup for sample 'not here' should be null but isn't");
}
/**
* Creates a new GATKSAMRecord with the source read's header, read group and mate information, but
* with the following fields set to user-supplied values: - Read Bases - Base Qualities - Base
* Insertion Qualities - Base Deletion Qualities
*
* <p>Cigar string is empty (not-null)
*
* <p>Use this method if you want to create a new GATKSAMRecord based on another GATKSAMRecord,
* but with modified bases and qualities
*
* @param read a read to copy the header from
* @param readBases an array containing the new bases you wish use in place of the originals
* @param baseQualities an array containing the new base qualities you wish use in place of the
* originals
* @param baseInsertionQualities an array containing the new base insertion qaulities
* @param baseDeletionQualities an array containing the new base deletion qualities
* @return a read with modified bases and qualities, safe for the GATK
*/
public static GATKSAMRecord createQualityModifiedRead(
final GATKSAMRecord read,
final byte[] readBases,
final byte[] baseQualities,
final byte[] baseInsertionQualities,
final byte[] baseDeletionQualities) {
if (baseQualities.length != readBases.length
|| baseInsertionQualities.length != readBases.length
|| baseDeletionQualities.length != readBases.length)
throw new IllegalArgumentException(
"Read bases and read quality arrays aren't the same size: Bases:"
+ readBases.length
+ " vs Base Q's:"
+ baseQualities.length
+ " vs Insert Q's:"
+ baseInsertionQualities.length
+ " vs Delete Q's:"
+ baseDeletionQualities.length);
final GATKSAMRecord processedRead = GATKSAMRecord.emptyRead(read);
processedRead.setReadBases(readBases);
processedRead.setBaseQualities(baseQualities, EventType.BASE_SUBSTITUTION);
processedRead.setBaseQualities(baseInsertionQualities, EventType.BASE_INSERTION);
processedRead.setBaseQualities(baseDeletionQualities, EventType.BASE_DELETION);
return processedRead;
}
public double[] computeReadHaplotypeLikelihoods(
ReadBackedPileup pileup, HashMap<Allele, Haplotype> haplotypesInVC) {
double[][] haplotypeLikehoodMatrix = new double[haplotypesInVC.size()][haplotypesInVC.size()];
double readLikelihoods[][] = new double[pileup.getReads().size()][haplotypesInVC.size()];
int i = 0;
for (GATKSAMRecord read : pileup.getReads()) {
if (ReadUtils.is454Read(read)) {
continue;
}
// for each read/haplotype combination, compute likelihoods, ie -10*log10(Pr(R | Hi))
// = sum_j(-10*log10(Pr(R_j | Hi) since reads are assumed to be independent
int j = 0;
for (Map.Entry<Allele, Haplotype> a : haplotypesInVC.entrySet()) {
readLikelihoods[i][j] = computeReadLikelihoodGivenHaplotype(a.getValue(), read);
if (DEBUG) {
System.out.print(read.getReadName() + " ");
System.out.format(
"%d %d S:%d US:%d E:%d UE:%d C:%s %3.4f\n",
i,
j,
read.getAlignmentStart(),
read.getUnclippedStart(),
read.getAlignmentEnd(),
read.getUnclippedEnd(),
read.getCigarString(),
readLikelihoods[i][j]);
}
j++;
}
i++;
}
for (i = 0; i < haplotypesInVC.size(); i++) {
for (int j = i; j < haplotypesInVC.size(); j++) {
// combine likelihoods of haplotypeLikelihoods[i], haplotypeLikelihoods[j]
// L(Hi, Hj) = sum_reads ( Pr(R|Hi)/2 + Pr(R|Hj)/2)
// readLikelihoods[k][j] has log10(Pr(R_k) | H[j] )
double[] readLikelihood = new double[2]; // diploid sample
for (int readIdx = 0; readIdx < pileup.getReads().size(); readIdx++) {
readLikelihood[0] = -readLikelihoods[readIdx][i] / 10;
readLikelihood[1] = -readLikelihoods[readIdx][j] / 10;
// Compute log10(10^x1/2 + 10^x2/2) = log10(10^x1+x0^x2)-log10(2)
// First term is approximated by Jacobian log with table lookup.
// Second term is a constant added to both likelihoods so will be ignored
haplotypeLikehoodMatrix[i][j] +=
MathUtils.approximateLog10SumLog10(readLikelihood[0], readLikelihood[1]);
}
}
}
return getHaplotypeLikelihoods(haplotypeLikehoodMatrix);
}
/**
* Is this read poorly modelled by all of the alleles in this map?
*
* <p>A read is poorly modeled when it's likelihood is below what would be expected for a read
* originating from one of the alleles given the maxErrorRatePerBase of the reads in general.
*
* <p>This function makes a number of key assumptions. First, that the likelihoods reflect the
* total likelihood of the read. In other words, that the read would be fully explained by one of
* the alleles. This means that the allele should be something like the full haplotype from which
* the read might originate.
*
* <p>It further assumes that each error in the read occurs with likelihood of -3 (Q30 confidence
* per base). So a read with a 10% error rate with Q30 bases that's 100 bp long we'd expect to see
* 10 real Q30 errors even against the true haplotype. So for this read to be well modelled by at
* least one allele we'd expect a likelihood to be >= 10 * -3.
*
* @param read the read we want to evaluate
* @param log10Likelihoods a list of the log10 likelihoods of the read against a set of
* haplotypes.
* @param maxErrorRatePerBase the maximum error rate we'd expect for this read per base, in real
* space. So 0.01 means a 1% error rate
* @return true if none of the log10 likelihoods imply that the read truly originated from one of
* the haplotypes
*/
protected boolean readIsPoorlyModelled(
final GATKSAMRecord read,
final Collection<Double> log10Likelihoods,
final double maxErrorRatePerBase) {
final double maxErrorsForRead =
Math.min(2.0, Math.ceil(read.getReadLength() * maxErrorRatePerBase));
final double log10QualPerBase = -4.0;
final double log10MaxLikelihoodForTrueAllele = maxErrorsForRead * log10QualPerBase;
for (final double log10Likelihood : log10Likelihoods)
if (log10Likelihood >= log10MaxLikelihoodForTrueAllele) return false;
return true;
}
/**
* @param read a read containing the variant
* @return number of hard clipped and low qual bases at the read end (where end is right end
* w.r.t. the reference)
*/
public static int getNumClippedBasesAtEnd(final GATKSAMRecord read) {
// check for hard clips (never consider these bases):
final Cigar c = read.getCigar();
CigarElement last = c.getCigarElement(c.numCigarElements() - 1);
int numEndClippedBases = 0;
if (last.getOperator() == CigarOperator.H) {
numEndClippedBases = last.getLength();
}
final byte[] unclippedReadBases = read.getReadBases();
final byte[] unclippedReadQuals = read.getBaseQualities();
// Do a stricter base clipping than provided by CIGAR string, since this one may be too
// conservative,
// and may leave a string of Q2 bases still hanging off the reads.
// TODO: this code may not even get used because HaplotypeCaller already hard clips low quality
// tails
for (int i = unclippedReadBases.length - numEndClippedBases - 1; i >= 0; i--) {
if (unclippedReadQuals[i] < PairHMMIndelErrorModel.BASE_QUAL_THRESHOLD) numEndClippedBases++;
else break;
}
return numEndClippedBases;
}
private List<PileupElement> makeReads(final int n, final int mapq, final String op) {
final int readLength = 3;
final List<PileupElement> elts = new LinkedList<PileupElement>();
for (int i = 0; i < n; i++) {
GATKSAMRecord read =
ArtificialSAMUtils.createArtificialRead(header, "read", 0, 1, readLength);
read.setReadBases(Utils.dupBytes((byte) 'A', readLength));
read.setBaseQualities(Utils.dupBytes((byte) 30, readLength));
read.setCigarString("1M1" + op + "1M");
read.setMappingQuality(mapq);
final int baseOffset = op.equals("M") ? 1 : 0;
final CigarElement cigarElement = read.getCigar().getCigarElement(1);
elts.add(new PileupElement(read, baseOffset, cigarElement, 1, 0));
}
return elts;
}
@Test
public void testRBPMappingQuals() {
// create a read with high MQ
final GATKSAMRecord read = ArtificialSAMUtils.createArtificialRead(header, "read", 0, 1, 10);
read.setReadBases(Utils.dupBytes((byte) 'A', 10));
read.setBaseQualities(Utils.dupBytes((byte) 30, 10));
read.setCigarString("10M");
read.setMappingQuality(200); // set a MQ higher than max signed byte
// now create the RBP
final List<PileupElement> elts = new LinkedList<>();
elts.add(new PileupElement(read, 0, read.getCigar().getCigarElement(0), 0, 0));
final Map<String, ReadBackedPileupImpl> pileupsBySample = new HashMap<>();
pileupsBySample.put("foo", new ReadBackedPileupImpl(loc, elts));
final ReadBackedPileup pileup = new ReadBackedPileupImpl(loc, pileupsBySample);
Assert.assertEquals(pileup.getMappingQuals()[0], 200);
}
/**
* Creates an empty GATKSAMRecord with the read's header, read group and mate information, but
* empty (not-null) fields: - Cigar String - Read Bases - Base Qualities
*
* <p>Use this method if you want to create a new empty GATKSAMRecord based on another
* GATKSAMRecord
*
* @param read a read to copy the header from
* @return a read with no bases but safe for the GATK
*/
public static GATKSAMRecord emptyRead(GATKSAMRecord read) {
final GATKSAMRecord emptyRead = new GATKSAMRecord(read.getHeader());
emptyRead.setReferenceIndex(read.getReferenceIndex());
emptyRead.setAlignmentStart(0);
emptyRead.setMappingQuality(0);
// setting read indexing bin last
emptyRead.setFlags(read.getFlags());
emptyRead.setMateReferenceIndex(read.getMateReferenceIndex());
emptyRead.setMateAlignmentStart(read.getMateAlignmentStart());
emptyRead.setInferredInsertSize(read.getInferredInsertSize());
emptyRead.setCigarString("");
emptyRead.setReadBases(new byte[0]);
emptyRead.setBaseQualities(new byte[0]);
SAMReadGroupRecord samRG = read.getReadGroup();
emptyRead.clearAttributes();
if (samRG != null) {
GATKSAMReadGroupRecord rg = new GATKSAMReadGroupRecord(samRG);
emptyRead.setReadGroup(rg);
}
GATKBin.setReadIndexingBin(emptyRead, 0);
return emptyRead;
}
/**
* Clips the bases in read according to this operation's start and stop. Uses the clipping
* representation used is the one provided by algorithm argument.
*
* @param algorithm clipping algorithm to use
* @param originalRead the read to be clipped
*/
public GATKSAMRecord apply(ClippingRepresentation algorithm, GATKSAMRecord originalRead) {
GATKSAMRecord read = (GATKSAMRecord) originalRead.clone();
byte[] quals = read.getBaseQualities();
byte[] bases = read.getReadBases();
byte[] newBases = new byte[bases.length];
byte[] newQuals = new byte[quals.length];
switch (algorithm) {
// important note:
// it's not safe to call read.getReadBases()[i] = 'N' or read.getBaseQualities()[i] = 0
// because you're not guaranteed to get a pointer to the actual array of bytes in the
// GATKSAMRecord
case WRITE_NS:
for (int i = 0; i < bases.length; i++) {
if (i >= start && i <= stop) {
newBases[i] = 'N';
} else {
newBases[i] = bases[i];
}
}
read.setReadBases(newBases);
break;
case WRITE_Q0S:
for (int i = 0; i < quals.length; i++) {
if (i >= start && i <= stop) {
newQuals[i] = 0;
} else {
newQuals[i] = quals[i];
}
}
read.setBaseQualities(newQuals);
break;
case WRITE_NS_Q0S:
for (int i = 0; i < bases.length; i++) {
if (i >= start && i <= stop) {
newQuals[i] = 0;
newBases[i] = 'N';
} else {
newQuals[i] = quals[i];
newBases[i] = bases[i];
}
}
read.setBaseQualities(newBases);
read.setReadBases(newBases);
break;
case HARDCLIP_BASES:
read = hardClip(read, start, stop);
break;
case SOFTCLIP_BASES:
if (read.getReadUnmappedFlag()) {
// we can't process unmapped reads
throw new UserException("Read Clipper cannot soft clip unmapped reads");
}
// System.out.printf("%d %d %d%n", stop, start, read.getReadLength());
int myStop = stop;
if ((stop + 1 - start) == read.getReadLength()) {
// BAM representation issue -- we can't SOFTCLIP away all bases in a read, just leave it
// alone
// Walker.logger.info(String.format("Warning, read %s has all bases clip but this can't be
// represented with SOFTCLIP_BASES, just leaving it alone", read.getReadName()));
// break;
myStop--; // just decrement stop
}
if (start > 0 && myStop != read.getReadLength() - 1)
throw new RuntimeException(
String.format(
"Cannot apply soft clipping operator to the middle of a read: %s to be clipped at %d-%d",
read.getReadName(), start, myStop));
Cigar oldCigar = read.getCigar();
int scLeft = 0, scRight = read.getReadLength();
if (start == 0) scLeft = myStop + 1;
else scRight = start;
Cigar newCigar = softClip(oldCigar, scLeft, scRight);
read.setCigar(newCigar);
int newClippedStart = getNewAlignmentStartOffset(newCigar, oldCigar);
int newStart = read.getAlignmentStart() + newClippedStart;
read.setAlignmentStart(newStart);
break;
case REVERT_SOFTCLIPPED_BASES:
read = revertSoftClippedBases(read);
break;
default:
throw new IllegalStateException("Unexpected Clipping operator type " + algorithm);
}
return read;
}
/**
* Hard clip bases from read, from start to stop in base coordinates
*
* <p>If start == 0, then we will clip from the front of the read, otherwise we clip from the
* right. If start == 0 and stop == 10, this would clip out the first 10 bases of the read.
*
* <p>Note that this function works with reads with negative alignment starts, in order to allow
* us to hardClip reads that have had their soft clips reverted and so might have negative
* alignment starts
*
* <p>Works properly with reduced reads and insertion/deletion base qualities
*
* @param read a non-null read
* @param start a start >= 0 and < read.length
* @param stop a stop >= 0 and < read.length.
* @return a cloned version of read that has been properly trimmed down
*/
private GATKSAMRecord hardClip(GATKSAMRecord read, int start, int stop) {
// If the read is unmapped there is no Cigar string and neither should we create a new cigar
// string
final CigarShift cigarShift =
(read.getReadUnmappedFlag())
? new CigarShift(new Cigar(), 0, 0)
: hardClipCigar(read.getCigar(), start, stop);
// the cigar may force a shift left or right (or both) in case we are left with insertions
// starting or ending the read after applying the hard clip on start/stop.
final int newLength =
read.getReadLength()
- (stop - start + 1)
- cigarShift.shiftFromStart
- cigarShift.shiftFromEnd;
final byte[] newBases = new byte[newLength];
final byte[] newQuals = new byte[newLength];
final int copyStart =
(start == 0) ? stop + 1 + cigarShift.shiftFromStart : cigarShift.shiftFromStart;
System.arraycopy(read.getReadBases(), copyStart, newBases, 0, newLength);
System.arraycopy(read.getBaseQualities(), copyStart, newQuals, 0, newLength);
final GATKSAMRecord hardClippedRead = (GATKSAMRecord) read.clone();
hardClippedRead
.resetSoftStartAndEnd(); // reset the cached soft start and end because they may have
// changed now that the read was hard clipped. No need to calculate
// them now. They'll be lazily calculated on the next call to
// getSoftStart()/End()
hardClippedRead.setBaseQualities(newQuals);
hardClippedRead.setReadBases(newBases);
hardClippedRead.setCigar(cigarShift.cigar);
if (start == 0)
hardClippedRead.setAlignmentStart(
read.getAlignmentStart()
+ calculateAlignmentStartShift(read.getCigar(), cigarShift.cigar));
if (read.hasBaseIndelQualities()) {
final byte[] newBaseInsertionQuals = new byte[newLength];
final byte[] newBaseDeletionQuals = new byte[newLength];
System.arraycopy(
read.getBaseInsertionQualities(), copyStart, newBaseInsertionQuals, 0, newLength);
System.arraycopy(
read.getBaseDeletionQualities(), copyStart, newBaseDeletionQuals, 0, newLength);
hardClippedRead.setBaseQualities(newBaseInsertionQuals, EventType.BASE_INSERTION);
hardClippedRead.setBaseQualities(newBaseDeletionQuals, EventType.BASE_DELETION);
}
return hardClippedRead;
}
// TODO: this is bizarre -- this code counts hard clips, but then subtracts them from the read
// length, which already doesn't count hard clips
public static int getNumAlignedBases(final GATKSAMRecord read) {
return read.getReadLength() - getNumClippedBasesAtStart(read) - getNumClippedBasesAtEnd(read);
}
/** Ensure that basic read group splitting works. */
@Test
public void testSplitByReadGroup() {
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read3 = ArtificialSAMUtils.createArtificialRead(header, "read3", 0, 1, 10);
read3.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read4 = ArtificialSAMUtils.createArtificialRead(header, "read4", 0, 1, 10);
read4.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read5 = ArtificialSAMUtils.createArtificialRead(header, "read5", 0, 1, 10);
read5.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read6 = ArtificialSAMUtils.createArtificialRead(header, "read6", 0, 1, 10);
read6.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read7 = ArtificialSAMUtils.createArtificialRead(header, "read7", 0, 1, 10);
read7.setAttribute("RG", readGroupOne.getId());
ReadBackedPileup pileup =
new ReadBackedPileupImpl(
null,
Arrays.asList(read1, read2, read3, read4, read5, read6, read7),
Arrays.asList(1, 1, 1, 1, 1, 1, 1));
ReadBackedPileup rg1Pileup = pileup.getPileupForReadGroup("rg1");
List<GATKSAMRecord> rg1Reads = rg1Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 4, "Wrong number of reads in read group rg1");
Assert.assertEquals(
rg1Reads.get(0), read1, "Read " + read1.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(1), read3, "Read " + read3.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(2), read6, "Read " + read6.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(3), read7, "Read " + read7.getReadName() + " should be in rg1 but isn't");
ReadBackedPileup rg2Pileup = pileup.getPileupForReadGroup("rg2");
List<GATKSAMRecord> rg2Reads = rg2Pileup.getReads();
Assert.assertEquals(rg2Reads.size(), 3, "Wrong number of reads in read group rg2");
Assert.assertEquals(
rg2Reads.get(0), read2, "Read " + read2.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(1), read4, "Read " + read4.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(2), read5, "Read " + read5.getReadName() + " should be in rg2 but isn't");
}
@Test
public void testStrandlessReads() {
final byte[] bases = {'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A'};
final byte[] quals = {20, 20, 20, 20, 20, 20, 20, 20};
GATKSAMRecord read = ArtificialSAMUtils.createArtificialRead(bases, quals, "6M");
Assert.assertEquals(read.isStrandless(), false);
read.setReadNegativeStrandFlag(false);
Assert.assertEquals(read.isStrandless(), false);
Assert.assertEquals(read.getReadNegativeStrandFlag(), false);
read.setReadNegativeStrandFlag(true);
Assert.assertEquals(read.isStrandless(), false);
Assert.assertEquals(read.getReadNegativeStrandFlag(), true);
read.setReadNegativeStrandFlag(true);
read.setIsStrandless(true);
Assert.assertEquals(read.isStrandless(), true);
Assert.assertEquals(
read.getReadNegativeStrandFlag(),
false,
"negative strand flag should return false even through its set for a strandless read");
}
/** Ensure that splitting read groups still works when dealing with a sample-split pileup. */
@Test
public void testSplitBySample() {
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
readGroupOne.setSample("sample1");
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
readGroupTwo.setSample("sample2");
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read3 = ArtificialSAMUtils.createArtificialRead(header, "read3", 0, 1, 10);
read3.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read4 = ArtificialSAMUtils.createArtificialRead(header, "read4", 0, 1, 10);
read4.setAttribute("RG", readGroupTwo.getId());
ReadBackedPileupImpl sample1Pileup =
new ReadBackedPileupImpl(null, Arrays.asList(read1, read3), Arrays.asList(1, 1));
ReadBackedPileupImpl sample2Pileup =
new ReadBackedPileupImpl(null, Arrays.asList(read2, read4), Arrays.asList(1, 1));
Map<String, ReadBackedPileupImpl> sampleToPileupMap =
new HashMap<String, ReadBackedPileupImpl>();
sampleToPileupMap.put(readGroupOne.getSample(), sample1Pileup);
sampleToPileupMap.put(readGroupTwo.getSample(), sample2Pileup);
ReadBackedPileup compositePileup = new ReadBackedPileupImpl(null, sampleToPileupMap);
ReadBackedPileup rg1Pileup = compositePileup.getPileupForReadGroup("rg1");
List<GATKSAMRecord> rg1Reads = rg1Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 2, "Wrong number of reads in read group rg1");
Assert.assertEquals(
rg1Reads.get(0), read1, "Read " + read1.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(1), read3, "Read " + read3.getReadName() + " should be in rg1 but isn't");
ReadBackedPileup rg2Pileup = compositePileup.getPileupForReadGroup("rg2");
List<GATKSAMRecord> rg2Reads = rg2Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 2, "Wrong number of reads in read group rg2");
Assert.assertEquals(
rg2Reads.get(0), read2, "Read " + read2.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(1), read4, "Read " + read4.getReadName() + " should be in rg2 but isn't");
}
@Override
protected boolean isUsableRead(final GATKSAMRecord read, final int refLoc) {
return super.isUsableRead(read, refLoc)
&& read.getSoftStart() + read.getCigar().getReadLength() > refLoc;
}
@Override
public boolean filter(ReferenceContext ref, GATKSAMRecord read) {
return !(read.getReadFailsVendorQualityCheckFlag() || read.getReadUnmappedFlag());
}
