/**
* Checks if a read contains adaptor sequences. If it does, hard clips them out.
*
* <p>Note: To see how a read is checked for adaptor sequence see ReadUtils.getAdaptorBoundary()
*
* @return a new read without adaptor sequence
*/
private GATKSAMRecord hardClipAdaptorSequence() {
final int adaptorBoundary = ReadUtils.getAdaptorBoundary(read);
if (adaptorBoundary == ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY
|| !ReadUtils.isInsideRead(read, adaptorBoundary)) return read;
return read.getReadNegativeStrandFlag()
? hardClipByReferenceCoordinatesLeftTail(adaptorBoundary)
: hardClipByReferenceCoordinatesRightTail(adaptorBoundary);
}
/**
* Generic functionality to hard clip a read, used internally by
* hardClipByReferenceCoordinatesLeftTail and hardClipByReferenceCoordinatesRightTail. Should not
* be used directly.
*
* <p>Note, it REQUIRES you to give the directionality of your hard clip (i.e. whether you're
* clipping the left of right tail) by specifying either refStart < 0 or refStop < 0.
*
* @param refStart first base to clip (inclusive)
* @param refStop last base to clip (inclusive)
* @return a new read, without the clipped bases
*/
@Requires({
"!read.getReadUnmappedFlag()",
"refStart < 0 || refStop < 0"
}) // can't handle unmapped reads, as we're using reference coordinates to clip
protected GATKSAMRecord hardClipByReferenceCoordinates(int refStart, int refStop) {
if (read.isEmpty()) return read;
int start;
int stop;
// Determine the read coordinate to start and stop hard clipping
if (refStart < 0) {
if (refStop < 0)
throw new ReviewedStingException(
"Only one of refStart or refStop must be < 0, not both ("
+ refStart
+ ", "
+ refStop
+ ")");
start = 0;
stop =
ReadUtils.getReadCoordinateForReferenceCoordinate(
read, refStop, ReadUtils.ClippingTail.LEFT_TAIL);
} else {
if (refStop >= 0)
throw new ReviewedStingException(
"Either refStart or refStop must be < 0 (" + refStart + ", " + refStop + ")");
start =
ReadUtils.getReadCoordinateForReferenceCoordinate(
read, refStart, ReadUtils.ClippingTail.RIGHT_TAIL);
stop = read.getReadLength() - 1;
}
if (start < 0 || stop > read.getReadLength() - 1)
throw new ReviewedStingException(
"Trying to clip before the start or after the end of a read");
if (start > stop)
throw new ReviewedStingException(
String.format(
"START (%d) > (%d) STOP -- this should never happen, please check read: %s (CIGAR: %s)",
start, stop, read, read.getCigarString()));
if (start > 0 && stop < read.getReadLength() - 1)
throw new ReviewedStingException(
String.format(
"Trying to clip the middle of the read: start %d, stop %d, cigar: %s",
start, stop, read.getCigarString()));
this.addOp(new ClippingOp(start, stop));
GATKSAMRecord clippedRead = clipRead(ClippingRepresentation.HARDCLIP_BASES);
this.ops = null;
return clippedRead;
}
@Test(enabled = true)
public void testGetMaxReadLength() {
for (final int minLength : Arrays.asList(5, 30, 50)) {
for (final int maxLength : Arrays.asList(50, 75, 100)) {
final List<GATKSAMRecord> reads = new ArrayList<GATKSAMRecord>();
for (int readLength = minLength; readLength <= maxLength; readLength++) {
reads.add(ReadUtils.createRandomRead(readLength));
}
Assert.assertEquals(
ReadUtils.getMaxReadLength(reads), maxLength, "max length does not match");
}
}
final List<GATKSAMRecord> reads = new LinkedList<GATKSAMRecord>();
Assert.assertEquals(
ReadUtils.getMaxReadLength(reads), 0, "Empty list should have max length of zero");
}
private static void runTest(
final GATKSAMReadGroupRecord rg, final String expected, final ReadGroupCovariate covariate) {
GATKSAMRecord read = ReadUtils.createRandomRead(10);
read.setReadGroup(rg);
ReadCovariates readCovariates = new ReadCovariates(read.getReadLength(), 1);
covariate.recordValues(read, readCovariates);
verifyCovariateArray(readCovariates.getMismatchesKeySet(), expected, covariate);
}
@Test(enabled = true)
public void testGetBasesReverseComplement() {
int iterations = 1000;
Random random = GenomeAnalysisEngine.getRandomGenerator();
while (iterations-- > 0) {
final int l = random.nextInt(1000);
GATKSAMRecord read = GATKSAMRecord.createRandomRead(l);
byte[] original = read.getReadBases();
byte[] reconverted = new byte[l];
String revComp = ReadUtils.getBasesReverseComplement(read);
for (int i = 0; i < l; i++) {
reconverted[l - 1 - i] = BaseUtils.getComplement((byte) revComp.charAt(i));
}
Assert.assertEquals(reconverted, original);
}
}
@Test(enabled = true)
public void testReadWithNsRefAfterDeletion() throws FileNotFoundException {
final IndexedFastaSequenceFile seq =
new CachingIndexedFastaSequenceFile(new File(b37KGReference));
final SAMFileHeader header =
ArtificialSAMUtils.createArtificialSamHeader(seq.getSequenceDictionary());
final int readLength = 76;
final GATKSAMRecord read =
ArtificialSAMUtils.createArtificialRead(header, "myRead", 0, 8975, readLength);
read.setReadBases(Utils.dupBytes((byte) 'A', readLength));
read.setBaseQualities(Utils.dupBytes((byte) 30, readLength));
read.setCigarString("3M414N1D73M");
final int result =
ReadUtils.getReadCoordinateForReferenceCoordinateUpToEndOfRead(
read, 9393, ReadUtils.ClippingTail.LEFT_TAIL);
Assert.assertEquals(result, 3);
}
private ArrayList<Allele> computeConsensusAlleles(
ReferenceContext ref,
Map<String, AlignmentContext> contexts,
AlignmentContextUtils.ReadOrientation contextType) {
Allele refAllele = null, altAllele = null;
GenomeLoc loc = ref.getLocus();
ArrayList<Allele> aList = new ArrayList<Allele>();
HashMap<String, Integer> consensusIndelStrings = new HashMap<String, Integer>();
int insCount = 0, delCount = 0;
// quick check of total number of indels in pileup
for (Map.Entry<String, AlignmentContext> sample : contexts.entrySet()) {
AlignmentContext context = AlignmentContextUtils.stratify(sample.getValue(), contextType);
final ReadBackedExtendedEventPileup indelPileup = context.getExtendedEventPileup();
insCount += indelPileup.getNumberOfInsertions();
delCount += indelPileup.getNumberOfDeletions();
}
if (insCount < minIndelCountForGenotyping && delCount < minIndelCountForGenotyping)
return aList;
for (Map.Entry<String, AlignmentContext> sample : contexts.entrySet()) {
// todo -- warning, can be duplicating expensive partition here
AlignmentContext context = AlignmentContextUtils.stratify(sample.getValue(), contextType);
final ReadBackedExtendedEventPileup indelPileup = context.getExtendedEventPileup();
for (ExtendedEventPileupElement p : indelPileup.toExtendedIterable()) {
// SAMRecord read = p.getRead();
GATKSAMRecord read = ReadUtils.hardClipAdaptorSequence(p.getRead());
if (read == null) continue;
if (ReadUtils.is454Read(read)) {
continue;
}
/* if (DEBUG && p.isIndel()) {
System.out.format("Read: %s, cigar: %s, aln start: %d, aln end: %d, p.len:%d, Type:%s, EventBases:%s\n",
read.getReadName(),read.getCigar().toString(),read.getAlignmentStart(),read.getAlignmentEnd(),
p.getEventLength(),p.getType().toString(), p.getEventBases());
}
*/
String indelString = p.getEventBases();
if (p.isInsertion()) {
boolean foundKey = false;
if (read.getAlignmentEnd() == loc.getStart()) {
// first corner condition: a read has an insertion at the end, and we're right at the
// insertion.
// In this case, the read could have any of the inserted bases and we need to build a
// consensus
for (String s : consensusIndelStrings.keySet()) {
int cnt = consensusIndelStrings.get(s);
if (s.startsWith(indelString)) {
// case 1: current insertion is prefix of indel in hash map
consensusIndelStrings.put(s, cnt + 1);
foundKey = true;
break;
} else if (indelString.startsWith(s)) {
// case 2: indel stored in hash table is prefix of current insertion
// In this case, new bases are new key.
consensusIndelStrings.remove(s);
consensusIndelStrings.put(indelString, cnt + 1);
foundKey = true;
break;
}
}
if (!foundKey)
// none of the above: event bases not supported by previous table, so add new key
consensusIndelStrings.put(indelString, 1);
} else if (read.getAlignmentStart() == loc.getStart() + 1) {
// opposite corner condition: read will start at current locus with an insertion
for (String s : consensusIndelStrings.keySet()) {
int cnt = consensusIndelStrings.get(s);
if (s.endsWith(indelString)) {
// case 1: current insertion is suffix of indel in hash map
consensusIndelStrings.put(s, cnt + 1);
foundKey = true;
break;
} else if (indelString.endsWith(s)) {
// case 2: indel stored in hash table is suffix of current insertion
// In this case, new bases are new key.
consensusIndelStrings.remove(s);
consensusIndelStrings.put(indelString, cnt + 1);
foundKey = true;
break;
}
}
if (!foundKey)
// none of the above: event bases not supported by previous table, so add new key
consensusIndelStrings.put(indelString, 1);
} else {
// normal case: insertion somewhere in the middle of a read: add count to hash map
int cnt =
consensusIndelStrings.containsKey(indelString)
? consensusIndelStrings.get(indelString)
: 0;
consensusIndelStrings.put(indelString, cnt + 1);
}
} else if (p.isDeletion()) {
indelString = String.format("D%d", p.getEventLength());
int cnt =
consensusIndelStrings.containsKey(indelString)
? consensusIndelStrings.get(indelString)
: 0;
consensusIndelStrings.put(indelString, cnt + 1);
}
}
/* if (DEBUG) {
int icount = indelPileup.getNumberOfInsertions();
int dcount = indelPileup.getNumberOfDeletions();
if (icount + dcount > 0)
{
List<Pair<String,Integer>> eventStrings = indelPileup.getEventStringsWithCounts(ref.getBases());
System.out.format("#ins: %d, #del:%d\n", insCount, delCount);
for (int i=0 ; i < eventStrings.size() ; i++ ) {
System.out.format("%s:%d,",eventStrings.get(i).first,eventStrings.get(i).second);
// int k=0;
}
System.out.println();
}
} */
}
int maxAlleleCnt = 0;
String bestAltAllele = "";
for (String s : consensusIndelStrings.keySet()) {
int curCnt = consensusIndelStrings.get(s);
if (curCnt > maxAlleleCnt) {
maxAlleleCnt = curCnt;
bestAltAllele = s;
}
// if (DEBUG)
// System.out.format("Key:%s, number: %d\n",s,consensusIndelStrings.get(s) );
} // gdebug-
if (maxAlleleCnt < minIndelCountForGenotyping) return aList;
if (bestAltAllele.startsWith("D")) {
// get deletion length
int dLen = Integer.valueOf(bestAltAllele.substring(1));
// get ref bases of accurate deletion
int startIdxInReference = (int) (1 + loc.getStart() - ref.getWindow().getStart());
// System.out.println(new String(ref.getBases()));
byte[] refBases =
Arrays.copyOfRange(ref.getBases(), startIdxInReference, startIdxInReference + dLen);
if (Allele.acceptableAlleleBases(refBases)) {
refAllele = Allele.create(refBases, true);
altAllele = Allele.create(Allele.NULL_ALLELE_STRING, false);
}
} else {
// insertion case
if (Allele.acceptableAlleleBases(bestAltAllele)) {
refAllele = Allele.create(Allele.NULL_ALLELE_STRING, true);
altAllele = Allele.create(bestAltAllele, false);
}
}
if (refAllele != null && altAllele != null) {
aList.add(0, refAllele);
aList.add(1, altAllele);
}
return aList;
}
@Test(dataProvider = "HasWellDefinedFragmentSizeData")
private void testHasWellDefinedFragmentSize(
final String name, final GATKSAMRecord read, final boolean expected) {
Assert.assertEquals(ReadUtils.hasWellDefinedFragmentSize(read), expected);
}
@Test(enabled = false)
public void testCovariateGeneration() {
final String RGID = "id";
final int length = 10;
final RecalibrationArgumentCollection RAC = new RecalibrationArgumentCollection();
GATKSAMRecord read = ReadUtils.createRandomRead(length, false);
GATKSAMReadGroupRecord rg = new GATKSAMReadGroupRecord(RGID);
rg.setPlatform("illumina");
read.setReadGroup(rg);
final byte[] mQuals = read.getBaseQualities(EventType.BASE_SUBSTITUTION);
final byte[] iQuals = read.getBaseQualities(EventType.BASE_INSERTION);
final byte[] dQuals = read.getBaseQualities(EventType.BASE_DELETION);
ReadGroupCovariate rgCov = new ReadGroupCovariate();
QualityScoreCovariate qsCov = new QualityScoreCovariate();
ContextCovariate coCov = new ContextCovariate();
CycleCovariate cyCov = new CycleCovariate();
rgCov.initialize(RAC);
qsCov.initialize(RAC);
coCov.initialize(RAC);
cyCov.initialize(RAC);
Covariate[] requestedCovariates = new Covariate[4];
requestedCovariates[0] = rgCov;
requestedCovariates[1] = qsCov;
requestedCovariates[2] = coCov;
requestedCovariates[3] = cyCov;
ReadCovariates rc = RecalDataManager.computeCovariates(read, requestedCovariates);
// check that the length is correct
Assert.assertEquals(rc.getMismatchesKeySet().length, length);
Assert.assertEquals(rc.getInsertionsKeySet().length, length);
Assert.assertEquals(rc.getDeletionsKeySet().length, length);
for (int i = 0; i < length; i++) {
// check that read group is always the same
Assert.assertEquals(rgCov.formatKey(rc.getMismatchesKeySet(i)[0]), RGID);
Assert.assertEquals(rgCov.formatKey(rc.getInsertionsKeySet(i)[0]), RGID);
Assert.assertEquals(rgCov.formatKey(rc.getDeletionsKeySet(i)[0]), RGID);
// check quality score
Assert.assertEquals(qsCov.formatKey(rc.getMismatchesKeySet(i)[1]), "" + mQuals[i]);
Assert.assertEquals(qsCov.formatKey(rc.getInsertionsKeySet(i)[1]), "" + iQuals[i]);
Assert.assertEquals(qsCov.formatKey(rc.getDeletionsKeySet(i)[1]), "" + dQuals[i]);
// check context
Assert.assertEquals(
coCov.formatKey(rc.getMismatchesKeySet(i)[2]),
ContextCovariateUnitTest.expectedContext(read, i, RAC.MISMATCHES_CONTEXT_SIZE));
Assert.assertEquals(
coCov.formatKey(rc.getInsertionsKeySet(i)[2]),
ContextCovariateUnitTest.expectedContext(read, i, RAC.INDELS_CONTEXT_SIZE));
Assert.assertEquals(
coCov.formatKey(rc.getDeletionsKeySet(i)[2]),
ContextCovariateUnitTest.expectedContext(read, i, RAC.INDELS_CONTEXT_SIZE));
// check cycle
Assert.assertEquals(cyCov.formatKey(rc.getMismatchesKeySet(i)[3]), "" + (i + 1));
Assert.assertEquals(cyCov.formatKey(rc.getInsertionsKeySet(i)[3]), "" + (i + 1));
Assert.assertEquals(cyCov.formatKey(rc.getDeletionsKeySet(i)[3]), "" + (i + 1));
}
}
public static GATKSAMRecord createRandomRead(int length, boolean allowNs) {
byte[] quals = ReadUtils.createRandomReadQuals(length);
byte[] bbases = ReadUtils.createRandomReadBases(length, allowNs);
return ArtificialSAMUtils.createArtificialRead(bbases, quals, bbases.length + "M");
}