unsigned UngappedAligner::alignUngapped(
FragmentMetadata &fragmentMetadata,
Cigar &cigarBuffer,
const flowcell::ReadMetadata &readMetadata,
const matchSelector::FragmentSequencingAdapterClipper &adapterClipper,
const reference::ContigList &contigList,
const isaac::reference::ContigAnnotations &contigAnnotations) const
{
const unsigned cigarOffset = cigarBuffer.size();
// Don't reset alignment to preserve the seed-based anchors.
// fragmentMetadata.resetAlignment();
ISAAC_ASSERT_MSG(!fragmentMetadata.isAligned(), "alignUngapped is expected to be performend on a clean fragment");
fragmentMetadata.resetClipping();
const reference::Contig &contig = contigList[fragmentMetadata.contigId];
const Read &read = fragmentMetadata.getRead();
const bool reverse = fragmentMetadata.reverse;
const std::vector<char> &sequence = read.getStrandSequence(reverse);
const reference::Contig &reference = contig;
std::vector<char>::const_iterator sequenceBegin = sequence.begin();
std::vector<char>::const_iterator sequenceEnd = sequence.end();
adapterClipper.clip(contig, fragmentMetadata, sequenceBegin, sequenceEnd);
clipReadMasking(read, fragmentMetadata, sequenceBegin, sequenceEnd);
clipReference(reference.size(), fragmentMetadata, sequenceBegin, sequenceEnd);
const unsigned firstMappedBaseOffset = std::distance(sequence.begin(), sequenceBegin);
if (firstMappedBaseOffset)
{
cigarBuffer.addOperation(firstMappedBaseOffset, Cigar::SOFT_CLIP);
}
const unsigned mappedBases = std::distance(sequenceBegin, sequenceEnd);
if (mappedBases)
{
const Cigar::OpCode opCode = Cigar::ALIGN;
cigarBuffer.addOperation(mappedBases, opCode);
}
const unsigned clipEndBases = std::distance(sequenceEnd, sequence.end());
if (clipEndBases)
{
cigarBuffer.addOperation(clipEndBases, Cigar::SOFT_CLIP);
}
const unsigned ret = updateFragmentCigar(
readMetadata, contigList, contigAnnotations, fragmentMetadata,
fragmentMetadata.reverse, fragmentMetadata.contigId, fragmentMetadata.position, cigarBuffer, cigarOffset);
if (!ret)
{
fragmentMetadata.setUnaligned();
}
return ret;
}
/**
* \brief Sets the sequence iterators according to the masking information stored in the read.
* Adjusts fragment.position to point at the first non-clipped base.
*
*/
void AlignerBase::clipReadMasking(
const alignment::Read &read,
FragmentMetadata &fragment,
std::vector<char>::const_iterator &sequenceBegin,
std::vector<char>::const_iterator &sequenceEnd)
{
std::vector<char>::const_iterator maskedBegin;
std::vector<char>::const_iterator maskedEnd;
if (fragment.reverse)
{
maskedBegin = read.getReverseSequence().begin() + read.getEndCyclesMasked();
maskedEnd = read.getReverseSequence().end() - read.getBeginCyclesMasked();
}
else
{
maskedBegin = read.getForwardSequence().begin() + read.getBeginCyclesMasked();
maskedEnd = read.getForwardSequence().end() - read.getEndCyclesMasked();
}
if (maskedBegin > sequenceBegin)
{
fragment.incrementClipLeft(std::distance(sequenceBegin, maskedBegin));
sequenceBegin = maskedBegin;
}
if (maskedEnd < sequenceEnd)
{
fragment.incrementClipRight(std::distance(maskedEnd, sequenceEnd));
sequenceEnd = maskedEnd;
}
}
bool alignsCorrectly(const std::size_t readNumber, const FragmentMetadata &fragment)
{
const reference::ReferencePosition oriPos = getAlignmentPositionFromName(readNumber, fragment);
if (oriPos.isTooManyMatch())
{
return true;
}
// ISAAC_THREAD_CERR << "oriPos:" << oriPos << " name " << common::makeFastIoString(fragment.getCluster().nameBegin(), fragment.getCluster().nameEnd()) << std::endl;
return fragment.isReverse() == oriPos.reverse() &&
fragment.getContigId() == oriPos.getContigId() &&
uint64_t(fragment.getPosition()) == oriPos.getPosition();
}
unsigned AlignerBase::updateFragmentCigar(
const flowcell::ReadMetadataList &readMetadataList,
const reference::ContigList &contigList,
const isaac::reference::ContigAnnotations &contigAnnotations,
FragmentMetadata &fragmentMetadata,
unsigned contigId,
const long strandPosition,
const Cigar &cigarBuffer,
const unsigned cigarOffset) const
{
return fragmentMetadata.updateAlignment(
alignmentCfg_,
readMetadataList,
contigList, contigAnnotations,
contigId, strandPosition,
cigarBuffer, cigarOffset);
}
/**
* \return true, if the clipping changed the alignment position
*/
bool SemialignedEndsClipper::clip(
const std::vector<reference::Contig> &contigList,
FragmentMetadata &fragmentMetadata)
{
if (!fragmentMetadata.isAligned() || fragmentMetadata.gapCount)
{
return false;
}
bool ret = clipLeftSide(contigList, fragmentMetadata);
if (clipRightSide(contigList, fragmentMetadata))
{
ret = true;
}
if (ret)
{
ISAAC_THREAD_CERR_DEV_TRACE(" SemialignedEndsClipper::clip: " << fragmentMetadata);
}
return ret;
}
reference::ReferencePosition getAlignmentPositionFromName(const std::size_t readNumber, const FragmentMetadata &fragment)
{
// numbers are 1-based
const auto name = getReadName(readNumber - 1, fragment);
if (name.second == name.first)
{
return reference::ReferencePosition(reference::ReferencePosition::TooManyMatch);
}
if ('u' == *name.first)
{
ISAAC_ASSERT_MSG(false, common::makeFastIoString(fragment.getCluster().nameBegin(), fragment.getCluster().nameEnd()) << " " << fragment);
return reference::ReferencePosition(reference::ReferencePosition::NoMatch);
}
return reference::ReferencePosition(
std::atol(&*name.first + 2),
std::atol(&*std::find(name.first + 2, name.second, ':') + 1),
false,
'r' == *name.first);
}
std::pair<BclClusters::const_iterator, BclClusters::const_iterator> getReadName(const std::size_t readIndex, const FragmentMetadata &fragment)
{
std::pair<BclClusters::const_iterator, BclClusters::const_iterator> ret = {fragment.getCluster().nameBegin(), fragment.getCluster().nameEnd()};
if (ret.second == ret.first)
{
return ret;
}
if (!readIndex)
{
ret.first = fragment.getCluster().nameBegin();
ret.second = std::find(fragment.getCluster().nameBegin(), fragment.getCluster().nameEnd(), '-');
}
else
{
ret.first = std::find(fragment.getCluster().nameBegin(), fragment.getCluster().nameEnd(), '-') + 1;
ret.second = fragment.getCluster().nameEnd();
}
return ret;
}
bool SemialignedEndsClipper::clipRightSide(
const std::vector<reference::Contig> &contigList,
FragmentMetadata &fragmentMetadata)
{
const Read &read = fragmentMetadata.getRead();
std::reverse_iterator<std::vector<char>::const_iterator> sequenceRBegin(
read.getStrandSequence(fragmentMetadata.reverse).end());
unsigned oldCigarOffset = fragmentMetadata.cigarOffset;
unsigned oldCigarLength = fragmentMetadata.cigarLength;
Cigar::Component operation = Cigar::decode(
fragmentMetadata.cigarBuffer->at(oldCigarOffset + oldCigarLength - 1));
unsigned softClippedEndBases = 0;
if (Cigar::SOFT_CLIP == operation.second)
{
if (2 > fragmentMetadata.cigarLength)
{
//when the adapter sequence
// happens to be at the extremities of the read, the whole read gets clipped away.
return false;
}
--oldCigarLength;
softClippedEndBases = operation.first;
sequenceRBegin += operation.first;
operation = Cigar::decode(fragmentMetadata.cigarBuffer->at(oldCigarOffset + oldCigarLength - 1));
}
if (Cigar::ALIGN == operation.second)
{
unsigned mappedEndBases = operation.first;
std::reverse_iterator<std::vector<char>::const_iterator> sequenceREnd = sequenceRBegin + mappedEndBases;
const std::vector<char> &reference = contigList.at(fragmentMetadata.contigId).forward_;
std::reverse_iterator<std::vector<char>::const_iterator> referenceRBegin(reference.begin() + fragmentMetadata.position +
fragmentMetadata.getObservedLength());
std::reverse_iterator<std::vector<char>::const_iterator> referenceREnd(reference.begin());
std::pair<unsigned, unsigned> clipped = clipMismatches<CONSECUTIVE_MATCHES_MIN>(sequenceRBegin, sequenceREnd,
referenceRBegin, referenceREnd,
[](char c){return c;});
if (clipped.first)
{
fragmentMetadata.cigarOffset = cigarBuffer_.size();
fragmentMetadata.observedLength -= clipped.first;
softClippedEndBases += clipped.first;
fragmentMetadata.editDistance -= clipped.second;
mappedEndBases -= clipped.first;
cigarBuffer_.insert(cigarBuffer_.end(),
fragmentMetadata.cigarBuffer->begin() + oldCigarOffset,
fragmentMetadata.cigarBuffer->begin() + oldCigarOffset + oldCigarLength - 1);
cigarBuffer_.addOperation(mappedEndBases, Cigar::ALIGN);
cigarBuffer_.addOperation(softClippedEndBases, Cigar::SOFT_CLIP);
fragmentMetadata.cigarBuffer = &cigarBuffer_;
fragmentMetadata.cigarLength = cigarBuffer_.size() - fragmentMetadata.cigarOffset;
return true;
}
}
return false;
}