// Soft Clip from the beginning of the read to the specified reference position.
int32_t CigarHelper::softClipBeginByRefPos(SamRecord& record,
int32_t refPosition0Based,
CigarRoller& newCigar,
int32_t &new0BasedPosition)
{
newCigar.clear();
Cigar* cigar = record.getCigarInfo();
if(cigar == NULL)
{
// Failed to get the cigar.
ErrorHandler::handleError("Soft clipping, but failed to read the cigar");
return(NO_CLIP);
}
// No cigar or position in the record, so return no clip.
if((cigar->size() == 0) || (record.get0BasedPosition() == -1))
{
return(NO_CLIP);
}
// Check to see if the reference position occurs before the record starts,
// if it does, do no clipping.
if(refPosition0Based < record.get0BasedPosition())
{
// Not within this read, so nothing to clip.
newCigar.Set(record.getCigar());
return(NO_CLIP);
}
// The position falls after the read starts, so loop through until the
// position or the end of the read is found.
int32_t readClipPosition = 0;
bool clipWritten = false;
new0BasedPosition = record.get0BasedPosition();
for(int i = 0; i < cigar->size(); i++)
{
const Cigar::CigarOperator* op = &(cigar->getOperator(i));
if(clipWritten)
{
// Clip point has been found, so just add everything.
newCigar += *op;
// Go to the next operation.
continue;
}
// The clip point has not yet been found, so check to see if we found
// it now.
// Not a clip, check to see if the operation is found in the
// reference.
if(Cigar::foundInReference(*op))
{
// match, mismatch, deletion, skip
// increment the current reference position to just past this
// operation.
new0BasedPosition += op->count;
// Check to see if this is also in the query, because otherwise
// the operation is still being consumed.
if(Cigar::foundInQuery(*op))
{
// Also in the query, determine if the entire thing should
// be clipped or just part of it.
uint32_t numKeep = 0;
// Check to see if we have hit our clip position.
if(refPosition0Based < new0BasedPosition)
{
// The specified clip position is in this cigar operation.
numKeep = new0BasedPosition - refPosition0Based - 1;
if(numKeep > op->count)
{
// Keep the entire read. This happens because
// we keep reading until the first match/mismatch
// after the clip.
numKeep = op->count;
}
}
// Add the part of this operation that is being clipped
// to the clip count.
readClipPosition += (op->count - numKeep);
// Only write the clip if we found a match/mismatch
// to write. Otherwise we will keep accumulating clips
// for the case of insertions.
if(numKeep > 0)
{
new0BasedPosition -= numKeep;
newCigar.Add(Cigar::softClip, readClipPosition);
// Add the clipped part of this cigar to the clip
// position.
newCigar.Add(op->operation, numKeep);
// Found a match after the clip point, so stop
// consuming cigar operations.
clipWritten = true;
continue;
}
}
}
else
{
// Only add hard clips. The softclips will be added in
// when the total number is found.
if(op->operation == Cigar::hardClip)
{
// Check if this is the first operation, if so, just write it.
if(i == 0)
{
newCigar += *op;
}
// Check if it is the last operation (otherwise skip it).
else if(i == (cigar->size() - 1))
{
// Check whether or not the clip was ever written, and if
// not, write it.
if(clipWritten == false)
{
newCigar.Add(Cigar::softClip, readClipPosition);
// Since no match/mismatch was ever found, set
// the new ref position to the original one.
new0BasedPosition = record.get0BasedPosition();
clipWritten = true;
}
// Add the hard clip.
newCigar += *op;
}
}
// Not yet to the clip position, so do not add this operation.
if(Cigar::foundInQuery(*op))
{
// Found in the query, so update the read clip position.
readClipPosition += op->count;
}
}
} // End loop through cigar.
// Check whether or not the clip was ever written, and if
// not, write it.
if(clipWritten == false)
{
newCigar.Add(Cigar::softClip, readClipPosition);
// Since no match/mismatch was ever found, set
// the new ref position to the original one.
new0BasedPosition = record.get0BasedPosition();
}
// Subtract 1 since readClipPosition atually contains the first 0based
// position that is not clipped.
return(readClipPosition - 1);
}
// Soft clip the cigar from the front and/or the back, writing the value
// into the new cigar.
SamFilter::FilterStatus SamFilter::softClip(Cigar& oldCigar,
int32_t numFrontClips,
int32_t numBackClips,
int32_t& startPos,
CigarRoller& updatedCigar)
{
int32_t readLength = oldCigar.getExpectedQueryBaseCount();
int32_t endClipPos = readLength - numBackClips;
FilterStatus status = NONE;
if((numFrontClips != 0) || (numBackClips != 0))
{
// Clipping from front and/or from the back.
// Check to see if the entire read was clipped.
int32_t totalClips = numFrontClips + numBackClips;
if(totalClips >= readLength)
{
/////////////////////////////
// The entire read is clipped, so rather than clipping it,
// filter it out.
return(FILTERED);
}
// Part of the read was clipped.
status = CLIPPED;
// Loop through, creating an updated cigar.
int origCigarOpIndex = 0;
// Track how many read positions are covered up to this
// point by the cigar to determine up to up to what
// point in the cigar is affected by this clipping.
int32_t numPositions = 0;
// Track if any non-clips are in the new cigar.
bool onlyClips = true;
const Cigar::CigarOperator* op = NULL;
//////////////////
// Clip from front
while((origCigarOpIndex < oldCigar.size()) &&
(numPositions < numFrontClips))
{
op = &(oldCigar.getOperator(origCigarOpIndex));
switch(op->operation)
{
case Cigar::hardClip:
// Keep this operation as the new clips do not
// affect other clips.
updatedCigar += *op;
break;
case Cigar::del:
case Cigar::skip:
// Skip and delete are going to be dropped, and
// are not in the read, so the read index doesn't
// need to be updated
break;
case Cigar::insert:
case Cigar::match:
case Cigar::mismatch:
case Cigar::softClip:
// Update the read index as these types
// are found in the read.
numPositions += op->count;
break;
case Cigar::none:
default:
// Nothing to do for none.
break;
};
++origCigarOpIndex;
}
// If bases were clipped from the front, add the clip and
// any partial cigar operation as necessary.
if(numFrontClips != 0)
{
// Add the softclip to the front of the read.
updatedCigar.Add(Cigar::softClip, numFrontClips);
// Add the rest of the last Cigar operation if
// it is not entirely clipped.
int32_t newCount = numPositions - numFrontClips;
if(newCount > 0)
{
// Before adding it, check to see if the same
// operation is clipped from the end.
// numPositions greater than the endClipPos
// means that it is equal or past that position,
// so shorten the number of positions.
if(numPositions > endClipPos)
{
newCount -= (numPositions - endClipPos);
}
if(newCount > 0)
{
updatedCigar.Add(op->operation, newCount);
if(!Cigar::isClip(op->operation))
{
onlyClips = false;
}
}
}
}
// Add operations until the point of the end clip is reached.
// For example...
// 2M1D3M = MMDMMM readLength = 5
// readIndex: 01 234
// at cigarOpIndex 0 (2M), numPositions = 2.
// at cigarOpIndex 1 (1D), numPositions = 2.
// at cigarOpIndex 2 (3M), numPositions = 5.
// if endClipPos = 2, we still want to consume the 1D, so
// need to keep looping until numPositions > endClipPos
while((origCigarOpIndex < oldCigar.size()) &&
(numPositions <= endClipPos))
{
op = &(oldCigar.getOperator(origCigarOpIndex));
// Update the numPositions count if the operations indicates
// bases within the read.
if(!Cigar::foundInQuery(op->operation))
{
// This operation is not in the query read sequence,
// so it is not yet to the endClipPos, just add the
// operation do not increment the number of positions.
updatedCigar += *op;
if(!Cigar::isClip(op->operation))
{
onlyClips = false;
}
}
else
{
// This operation appears in the query sequence, so
// check to see if the clip occurs in this operation.
// endClipPos is 0 based & numPositions is a count.
// If endClipPos is 4, then it is the 5th position.
// If 4 positions are covered so far (numPositions = 4),
// then we are right at endCLipPos: 4-4 = 0, none of
// this operation should be kept.
// If only 3 positions were covered, then we are at offset
// 3, so offset 3 should be added: 4-3 = 1.
uint32_t numPosTilClip = endClipPos - numPositions;
if(numPosTilClip < op->count)
{
// this operation is partially clipped, write the part
// that was not clipped if it is not all clipped.
if(numPosTilClip != 0)
{
updatedCigar.Add(op->operation,
numPosTilClip);
if(!Cigar::isClip(op->operation))
{
onlyClips = false;
}
}
}
else
{
// This operation is not clipped, so add it
updatedCigar += *op;
if(!Cigar::isClip(op->operation))
{
onlyClips = false;
}
}
// This operation occurs in the query sequence, so
// increment the number of positions covered.
numPositions += op->count;
}
// Move to the next cigar position.
++origCigarOpIndex;
}
//////////////////
// Add the softclip to the back.
if(numBackClips != 0)
{
// Add the softclip to the end
updatedCigar.Add(Cigar::softClip, numBackClips);
}
//////////////////
// Add any hardclips remaining in the original cigar to the back.
while(origCigarOpIndex < oldCigar.size())
{
op = &(oldCigar.getOperator(origCigarOpIndex));
if(op->operation == Cigar::hardClip)
{
// Keep this operation as the new clips do not
// affect other clips.
updatedCigar += *op;
}
++origCigarOpIndex;
}
// Check to see if the new cigar is only clips.
if(onlyClips)
{
// Only clips in the new cigar, so mark the read as filtered
// instead of updating the cigar.
/////////////////////////////
// The entire read was clipped.
status = FILTERED;
}
else
{
// Part of the read was clipped.
// Update the starting position if a clip was added to
// the front.
if(numFrontClips > 0)
{
// Convert from query index to reference position (from the
// old cigar)
// Get the position for the last front clipped position by
// getting the position associated with the clipped base on
// the reference. Then add one to get to the first
// non-clipped position.
int32_t lastFrontClipPos = numFrontClips - 1;
int32_t newStartPos = oldCigar.getRefPosition(lastFrontClipPos,
startPos);
if(newStartPos != Cigar::INDEX_NA)
{
// Add one to get first non-clipped position.
startPos = newStartPos + 1;
}
}
}
}
return(status);
}
// Soft Clip from the end of the read at the specified reference position.
int32_t CigarHelper::softClipEndByRefPos(SamRecord& record,
int32_t refPosition0Based,
CigarRoller& newCigar)
{
newCigar.clear();
Cigar* cigar = record.getCigarInfo();
if(cigar == NULL)
{
// Failed to get the cigar.
ErrorHandler::handleError("Soft clipping, but failed to read the cigar");
return(NO_CLIP);
}
// No cigar or position in the record, so return no clip.
if((cigar->size() == 0) || (record.get0BasedPosition() == -1))
{
return(NO_CLIP);
}
// Check to see if the reference position occurs after the record ends,
// if so, do no clipping.
if(refPosition0Based > record.get0BasedAlignmentEnd())
{
// Not within this read, so nothing to clip.
newCigar.Set(record.getCigar());
return(NO_CLIP);
}
// The position falls before the read ends, so loop through until the
// position is found.
int32_t currentRefPosition = record.get0BasedPosition();
int32_t readClipPosition = 0;
for(int i = 0; i < cigar->size(); i++)
{
const Cigar::CigarOperator* op = &(cigar->getOperator(i));
// If the operation is found in the reference, increase the
// reference position.
if(Cigar::foundInReference(*op))
{
// match, mismatch, deletion, skip
// increment the current reference position to just past
// this operation.
currentRefPosition += op->count;
}
// Check to see if we have hit our clip position.
if(refPosition0Based < currentRefPosition)
{
// If this read is also in the query (match/mismatch),
// write the partial op to the new cigar.
int32_t numKeep = 0;
if(Cigar::foundInQuery(*op))
{
numKeep = op->count - (currentRefPosition - refPosition0Based);
if(numKeep > 0)
{
newCigar.Add(op->operation, numKeep);
readClipPosition += numKeep;
}
}
else if(Cigar::isClip(*op))
{
// This is a hard clip, so write it.
newCigar.Add(op->operation, op->count);
}
else
{
// Not found in the query (skip/deletion),
// so don't write any of the operation.
}
// Found the clip point, so break.
break;
}
else if(refPosition0Based == currentRefPosition)
{
newCigar += *op;
if(Cigar::foundInQuery(*op))
{
readClipPosition += op->count;
}
}
else
{
// Not yet to the clip position, so add this operation/size to
// the new cigar.
newCigar += *op;
if(Cigar::foundInQuery(*op))
{
// Found in the query, so update the read clip position.
readClipPosition += op->count;
}
}
} // End loop through cigar.
// Before adding the softclip, read from the end of the cigar checking to
// see if the operations are in the query, removing operations that are
// not (pad/delete/skip) until a hardclip or an operation in the query is
// found. We do not want a pad/delete/skip right before a softclip.
for(int j = newCigar.size() - 1; j >= 0; j--)
{
const Cigar::CigarOperator* op = &(newCigar.getOperator(j));
if(!Cigar::foundInQuery(*op) && !Cigar::isClip(*op))
{
// pad/delete/skip
newCigar.Remove(j);
}
else if(Cigar::foundInQuery(*op) & Cigar::isClip(*op))
{
// Soft clip, so increment the clip position for the return value.
// Remove the softclip since the readClipPosition is used to
// calculate teh size of the soft clip added.
readClipPosition -= op->count;
newCigar.Remove(j);
}
else
{
// Found a cigar operation that should not be deleted, so stop deleting.
break;
}
}
// Determine the number of soft clips.
int32_t numSoftClips = record.getReadLength() - readClipPosition;
// NOTE that if the previous operation is a softclip, the CigarRoller logic
// will merge this with that one.
newCigar.Add(Cigar::softClip, numSoftClips);
// Check if an ending hard clip needs to be added.
if(cigar->size() != 0)
{
const Cigar::CigarOperator* lastOp =
&(cigar->getOperator(cigar->size() - 1));
if(lastOp->operation == Cigar::hardClip)
{
newCigar += *lastOp;
}
}
return(readClipPosition);
}