static
void
getSingleReadSVCandidates(
const ReadScannerOptions& opt,
const ReadScannerDerivOptions& dopt,
const bam_record& localRead,
const SimpleAlignment& localAlign,
const chromMap_t& chromToIndex,
const reference_contig_segment& refSeq,
std::vector<SVObservation>& candidates)
{
using namespace illumina::common;
const bool isRead2(localRead.is_paired() && (localRead.read_no() == 2));
const FRAGSOURCE::index_t fragSource(isRead2 ? FRAGSOURCE::READ2 : FRAGSOURCE::READ1);
// - process any large indels in the localRead:
getSVCandidatesFromReadIndels(opt, dopt, localAlign, fragSource, candidates);
#ifdef DEBUG_SCANNER
log_os << __FUNCTION__ << ": post-indels candidate_size: " << candidates.size() << "\n";
#endif
// a read can provide SA split evidence or semi-aligned/soft-clip, but not both.
// this prevents split reads from triggering spurious local assembles. It is
// possible for a read to genuinely contain evidence of both, but this should
// be very rare.
if (localRead.isSASplit())
{
getSACandidatesFromRead(opt, dopt, localRead, localAlign, fragSource, chromToIndex,
candidates);
#ifdef DEBUG_SCANNER
log_os << __FUNCTION__ << ": post-split read candidate_size: " << candidates.size() << "\n";
#endif
}
else
{
if (dopt.isSmallCandidates)
{
getSVCandidatesFromSemiAligned(opt, dopt, localRead, localAlign, fragSource, refSeq,
candidates);
}
#ifdef DEBUG_SCANNER
log_os << __FUNCTION__ << ": post-semialigned candidate_size: " << candidates.size() << "\n";
#endif
}
}
bool
SVLocusScanner::
isSVEvidence(
const bam_record& bamRead,
const unsigned defaultReadGroupIndex,
const reference_contig_segment& refSeq,
SVLocusEvidenceCount* incountsPtr) const
{
// exclude innie read pairs which are anomalously short:
const bool isAnom(isNonCompressedAnomalous(bamRead,defaultReadGroupIndex));
const bool isSplit(bamRead.isSASplit());
getAlignment(bamRead,_bamAlign);
const bool isIndel(isLocalIndelEvidence(_bamAlign));
const bool isAssm((_dopt.isSmallCandidates) && ((!isSplit) && isSemiAlignedEvidence(bamRead, _bamAlign, refSeq)));
const bool isEvidence(isAnom || isSplit || isIndel || isAssm);
if (nullptr != incountsPtr)
{
SVLocusEvidenceCount& incounts(*incountsPtr);
incounts.total++;
if (isAnom) incounts.anom++;
if (isSplit) incounts.split++;
if (isIndel) incounts.indel++;
if (isAssm) incounts.assm++;
if (! isEvidence) incounts.ignored++;
if (isAnom)
{
if (isMateInsertionEvidenceCandidate(bamRead, getMinMapQ()))
{
// these counts are used to generate background noise rates in later candidate generation stages:
incounts.remoteRecoveryCandidates++;
}
}
}
return isEvidence;
}