private boolean isAssemblyTriggerCandidate(SAMRecord read, CompareToReference2 c2r) {
boolean isCandidate = false;
// Increment candidate count for indels
if (read.getCigarString().contains("I") || read.getCigarString().contains("D")) {
isCandidate = true;
}
// Increment candidate count for substantial high quality soft clipping
// TODO: Check for chimera directly?
if (read.getCigarString().contains("S")) {
if (c2r.numHighQualityMismatches(read, MIN_CANDIDATE_BASE_QUALITY) > (readLength / 10)) {
isCandidate = true;
}
}
// Increment candidate count if read contains at least 3 high quality mismatches
if (SAMRecordUtils.getIntAttribute(read, "NM") >= 3) {
if (c2r.numHighQualityMismatches(read, MIN_CANDIDATE_BASE_QUALITY) > 3) {
isCandidate = true;
}
}
return isCandidate;
}
/** Note: this is the only getKey function that handles unmapped reads specially! */
public static long getKey(final SAMRecord rec) {
final int refIdx = rec.getReferenceIndex();
final int start = rec.getAlignmentStart();
if (!(rec.getReadUnmappedFlag() || refIdx < 0 || start < 0)) return getKey(refIdx, start);
// Put unmapped reads at the end, but don't give them all the exact same
// key so that they can be distributed to different reducers.
//
// A random number would probably be best, but to ensure that the same
// record always gets the same key we use a fast hash instead.
//
// We avoid using hashCode(), because it's not guaranteed to have the
// same value across different processes.
int hash = 0;
byte[] var;
if ((var = rec.getVariableBinaryRepresentation()) != null) {
// Undecoded BAM record: just hash its raw data.
hash = (int) MurmurHash3.murmurhash3(var, hash);
} else {
// Decoded BAM record or any SAM record: hash a few representative
// fields together.
hash = (int) MurmurHash3.murmurhash3(rec.getReadName(), hash);
hash = (int) MurmurHash3.murmurhash3(rec.getReadBases(), hash);
hash = (int) MurmurHash3.murmurhash3(rec.getBaseQualities(), hash);
hash = (int) MurmurHash3.murmurhash3(rec.getCigarString(), hash);
}
return getKey0(Integer.MAX_VALUE, hash);
}
/**
* HACK TO CREATE GATKSAMRECORD BASED ONLY A SAMRECORD FOR TESTING PURPOSES ONLY
*
* @param read
*/
public GATKSAMRecord(final SAMRecord read) {
super(read.getHeader());
super.setReferenceIndex(read.getReferenceIndex());
super.setAlignmentStart(read.getAlignmentStart());
super.setReadName(read.getReadName());
super.setMappingQuality(read.getMappingQuality());
// indexing bin done below
super.setCigar(read.getCigar());
super.setFlags(read.getFlags());
super.setMateReferenceIndex(read.getMateReferenceIndex());
super.setMateAlignmentStart(read.getMateAlignmentStart());
super.setInferredInsertSize(read.getInferredInsertSize());
SAMReadGroupRecord samRG = read.getReadGroup();
SAMBinaryTagAndValue samAttr = GATKBin.getReadBinaryAttributes(read);
if (samAttr == null) {
clearAttributes();
} else {
setAttributes(samAttr);
}
if (samRG != null) {
GATKSAMReadGroupRecord rg = new GATKSAMReadGroupRecord(samRG);
setReadGroup(rg);
}
super.setFileSource(read.getFileSource());
super.setReadName(read.getReadName());
super.setCigarString(read.getCigarString());
super.setReadBases(read.getReadBases());
super.setBaseQualities(read.getBaseQualities());
// From SAMRecord constructor: Do this after the above because setCigarString will clear it.
GATKBin.setReadIndexingBin(this, GATKBin.getReadIndexingBin(read));
}
private boolean isHardClipped(SAMRecord read) {
return read.getCigarString().contains("H");
}