bool SamFile::checkRecordInSection(SamRecord& record)
{
bool recordFound = true;
if(myRefID == BamIndex::REF_ID_ALL)
{
return(true);
}
// Check to see if it is in the correct reference/position.
if(record.getReferenceID() != myRefID)
{
// Incorrect reference ID, return no more records.
myStatus = SamStatus::NO_MORE_RECS;
return(false);
}
// Found a record.
recordFound = true;
// If start/end position are set, verify that the alignment falls
// within those.
// If the alignment start is greater than the end of the region,
// return NO_MORE_RECS.
// Since myEndPos is Exclusive 0-based, anything >= myEndPos is outside
// of the region.
if((myEndPos != -1) && (record.get0BasedPosition() >= myEndPos))
{
myStatus = SamStatus::NO_MORE_RECS;
return(false);
}
// We know the start is less than the end position, so the alignment
// overlaps the region if the alignment end position is greater than the
// start of the region.
if((myStartPos != -1) && (record.get0BasedAlignmentEnd() < myStartPos))
{
// If it does not overlap the region, so go to the next
// record...set recordFound back to false.
recordFound = false;
}
if(!myOverlapSection)
{
// Needs to be fully contained. Not fully contained if
// 1) the record start position is < the region start position.
// or
// 2) the end position is specified and the record end position
// is greater than or equal to the region end position.
// (equal to since the region is exclusive.
if((record.get0BasedPosition() < myStartPos) ||
((myEndPos != -1) &&
(record.get0BasedAlignmentEnd() >= myEndPos)))
{
// This record is not fully contained, so move on to the next
// record.
recordFound = false;
}
}
return(recordFound);
}
void validateRead1ModQuality(SamRecord& samRecord)
{
//////////////////////////////////////////
// Validate Record 1
// Create record structure for validating.
int expectedBlockSize = 89;
const char* expectedReferenceName = "1";
const char* expectedMateReferenceName = "1";
const char* expectedMateReferenceNameOrEqual = "=";
bamRecordStruct* expectedRecordPtr =
(bamRecordStruct *) malloc(expectedBlockSize + sizeof(int));
char tag[3];
char type;
void* value;
bamRecordStruct* bufferPtr;
unsigned char* varPtr;
expectedRecordPtr->myBlockSize = expectedBlockSize;
expectedRecordPtr->myReferenceID = 0;
expectedRecordPtr->myPosition = 1010;
expectedRecordPtr->myReadNameLength = 23;
expectedRecordPtr->myMapQuality = 0;
expectedRecordPtr->myBin = 4681;
expectedRecordPtr->myCigarLength = 2;
expectedRecordPtr->myFlag = 73;
expectedRecordPtr->myReadLength = 5;
expectedRecordPtr->myMateReferenceID = 0;
expectedRecordPtr->myMatePosition = 1010;
expectedRecordPtr->myInsertSize = 0;
// Check the alignment end
assert(samRecord.get0BasedAlignmentEnd() == 1016);
assert(samRecord.get1BasedAlignmentEnd() == 1017);
assert(samRecord.getAlignmentLength() == 7);
assert(samRecord.get0BasedUnclippedStart() == 1010);
assert(samRecord.get1BasedUnclippedStart() == 1011);
assert(samRecord.get0BasedUnclippedEnd() == 1016);
assert(samRecord.get1BasedUnclippedEnd() == 1017);
// Check the accessors.
assert(samRecord.getBlockSize() == expectedRecordPtr->myBlockSize);
assert(samRecord.getReferenceID() == expectedRecordPtr->myReferenceID);
assert(strcmp(samRecord.getReferenceName(), expectedReferenceName) == 0);
assert(samRecord.get1BasedPosition() == expectedRecordPtr->myPosition + 1);
assert(samRecord.get0BasedPosition() == expectedRecordPtr->myPosition);
assert(samRecord.getReadNameLength() ==
expectedRecordPtr->myReadNameLength);
assert(samRecord.getMapQuality() == expectedRecordPtr->myMapQuality);
assert(samRecord.getBin() == expectedRecordPtr->myBin);
assert(samRecord.getCigarLength() == expectedRecordPtr->myCigarLength);
assert(samRecord.getFlag() == expectedRecordPtr->myFlag);
assert(samRecord.getReadLength() == expectedRecordPtr->myReadLength);
assert(samRecord.getMateReferenceID() ==
expectedRecordPtr->myMateReferenceID);
assert(strcmp(samRecord.getMateReferenceName(),
expectedMateReferenceName) == 0);
assert(strcmp(samRecord.getMateReferenceNameOrEqual(),
expectedMateReferenceNameOrEqual) == 0);
assert(samRecord.get1BasedMatePosition() ==
expectedRecordPtr->myMatePosition + 1);
assert(samRecord.get0BasedMatePosition() ==
expectedRecordPtr->myMatePosition);
assert(samRecord.getInsertSize() == expectedRecordPtr->myInsertSize);
assert(strcmp(samRecord.getReadName(), "1:1011:F:255+17M15D20M") == 0);
assert(strcmp(samRecord.getCigar(), "5M2D") == 0);
assert(strcmp(samRecord.getSequence(), "CCGAA") == 0);
assert(strcmp(samRecord.getQuality(), "ABCDE") == 0);
assert(samRecord.getNumOverlaps(1010, 1017) == 5);
assert(samRecord.getNumOverlaps(1010, 1016) == 5);
assert(samRecord.getNumOverlaps(1012, 1017) == 3);
assert(samRecord.getNumOverlaps(1015, 1017) == 0);
assert(samRecord.getNumOverlaps(1017, 1010) == 0);
assert(samRecord.getNumOverlaps(1013, 1011) == 0);
assert(samRecord.getNumOverlaps(-1, 1017) == 5);
// Reset the tag iter, since the tags have already been read.
samRecord.resetTagIter();
// Check the tags.
assert(samRecord.getNextSamTag(tag, type, &value) == true);
assert(tag[0] == 'A');
assert(tag[1] == 'M');
assert(type == 'i');
assert(*(char*)value == 0);
assert(samRecord.getNextSamTag(tag, type, &value) == true);
assert(tag[0] == 'M');
assert(tag[1] == 'D');
assert(type == 'Z');
assert(*(String*)value == "37");
assert(samRecord.getNextSamTag(tag, type, &value) == true);
assert(tag[0] == 'N');
assert(tag[1] == 'M');
assert(type == 'i');
assert(*(char*)value == 0);
assert(samRecord.getNextSamTag(tag, type, &value) == true);
assert(tag[0] == 'X');
assert(tag[1] == 'T');
assert(type == 'A');
assert(*(char*)value == 'R');
// No more tags, should return false.
assert(samRecord.getNextSamTag(tag, type, &value) == false);
assert(samRecord.getNextSamTag(tag, type, &value) == false);
// Get the record ptr.
bufferPtr = (bamRecordStruct*)samRecord.getRecordBuffer();
// Validate the buffers match.
assert(bufferPtr->myBlockSize == expectedRecordPtr->myBlockSize);
assert(bufferPtr->myReferenceID == expectedRecordPtr->myReferenceID);
assert(bufferPtr->myPosition == expectedRecordPtr->myPosition);
assert(bufferPtr->myReadNameLength == expectedRecordPtr->myReadNameLength);
assert(bufferPtr->myMapQuality == expectedRecordPtr->myMapQuality);
assert(bufferPtr->myBin == expectedRecordPtr->myBin);
assert(bufferPtr->myCigarLength == expectedRecordPtr->myCigarLength);
assert(bufferPtr->myFlag == expectedRecordPtr->myFlag);
assert(bufferPtr->myReadLength == expectedRecordPtr->myReadLength);
assert(bufferPtr->myMateReferenceID ==
expectedRecordPtr->myMateReferenceID);
assert(bufferPtr->myMatePosition == expectedRecordPtr->myMatePosition);
assert(bufferPtr->myInsertSize == expectedRecordPtr->myInsertSize);
// Validate the variable length fields in the buffer.
// Set the pointer to the start of the variable fields.
varPtr = (unsigned char*)(&(bufferPtr->myData[0]));
// Validate the readname.
for(int i = 0; i < expectedRecordPtr->myReadNameLength; i++)
{
assert(*varPtr == samRecord.getReadName()[i]);
varPtr++;
}
// Validate the cigar.
// The First cigar is 5M which is 5 << 4 | 0 = 80
assert(*(unsigned int*)varPtr == 80);
// Increment the varptr the size of an int.
varPtr += 4;
// The 2nd cigar is 2D which is 2 << 4 | 2 = 34
assert(*(unsigned int*)varPtr == 34);
// Increment the varptr the size of an int.
varPtr += 4;
// Validate the sequence.
// CC = 0x22
assert(*varPtr == 0x22);
varPtr++;
// GA = 0x41
assert(*varPtr == 0x41);
varPtr++;
// A = 0x10
assert(*varPtr == 0x10);
varPtr++;
// Validate the Quality
for(int i = 0; i < expectedRecordPtr->myReadLength; i++)
{
assert(*varPtr == samRecord.getQuality()[i] - 33);
varPtr++;
}
// Validate the tags.
assert(*varPtr == 'A');
varPtr++;
assert(*varPtr == 'M');
varPtr++;
assert(*varPtr == 'C');
varPtr++;
assert(*varPtr == 0);
varPtr++;
assert(*varPtr == 'M');
varPtr++;
assert(*varPtr == 'D');
varPtr++;
assert(*varPtr == 'Z');
varPtr++;
assert(*varPtr == '3');
varPtr++;
assert(*varPtr == '7');
varPtr++;
assert(*varPtr == 0);
varPtr++;
assert(*varPtr == 'N');
varPtr++;
assert(*varPtr == 'M');
varPtr++;
assert(*varPtr == 'C');
varPtr++;
assert(*varPtr == 0);
varPtr++;
assert(*varPtr == 'X');
varPtr++;
assert(*varPtr == 'T');
varPtr++;
assert(*varPtr == 'A');
varPtr++;
assert(*varPtr == 'R');
varPtr++;
}
int GapInfo::processFile(const char* inputFileName, const char* outputFileName,
const char* refFile, bool detailed,
bool checkFirst, bool checkStrand)
{
// Open the file for reading.
SamFile samIn;
samIn.OpenForRead(inputFileName);
// Read the sam header.
SamFileHeader samHeader;
samIn.ReadHeader(samHeader);
SamRecord samRecord;
GenomeSequence* refPtr = NULL;
if(strcmp(refFile, "") != 0)
{
refPtr = new GenomeSequence(refFile);
}
IFILE outFile = ifopen(outputFileName, "w");
// Map for summary.
std::map<int, int> gapInfoMap;
// Keep reading records until ReadRecord returns false.
while(samIn.ReadRecord(samHeader, samRecord))
{
uint16_t samFlags = samRecord.getFlag();
if((!SamFlag::isMapped(samFlags)) ||
(!SamFlag::isMateMapped(samFlags)) ||
(!SamFlag::isPaired(samFlags)) ||
(samFlags & SamFlag::SECONDARY_ALIGNMENT) ||
(SamFlag::isDuplicate(samFlags)) ||
(SamFlag::isQCFailure(samFlags)))
{
// unmapped, mate unmapped, not paired,
// not the primary alignment,
// duplicate, fails vendor quality check
continue;
}
// No gap info if the chromosome names are different or
// are unknown.
int32_t refID = samRecord.getReferenceID();
if((refID != samRecord.getMateReferenceID()) || (refID == -1))
{
continue;
}
int32_t readStart = samRecord.get0BasedPosition();
int32_t mateStart = samRecord.get0BasedMatePosition();
// If the mate starts first, then the pair was processed by
// the mate.
if(mateStart < readStart)
{
continue;
}
if((mateStart == readStart) && (SamFlag::isReverse(samFlags)))
{
// read and mate start at the same position, so
// only process the forward strand.
continue;
}
// Process this read pair.
int32_t readEnd = samRecord.get0BasedAlignmentEnd();
int32_t gapSize = mateStart - readEnd - 1;
if(detailed)
{
// Output the gap info.
ifprintf(outFile, "%s\t%d\t%d",
samRecord.getReferenceName(), readEnd+1, gapSize);
// Check if it is not the first or if it is not the forward strand.
if(checkFirst && !SamFlag::isFirstFragment(samFlags))
{
ifprintf(outFile, "\tNotFirst");
}
if(checkStrand && SamFlag::isReverse(samFlags))
{
ifprintf(outFile, "\tReverse");
}
ifprintf(outFile, "\n");
}
else
{
// Summary.
// Skip reads that are not the forward strand.
if(SamFlag::isReverse(samFlags))
{
// continue
continue;
}
// Forward.
// Check the reference for 'N's.
if(refPtr != NULL)
{
genomeIndex_t chromStartIndex =
refPtr->getGenomePosition(samRecord.getReferenceName());
if(chromStartIndex == INVALID_GENOME_INDEX)
{
// Invalid position, so continue to the next one.
continue;
}
bool skipRead = false;
for(int i = readEnd + 1; i < mateStart; i++)
{
if((*refPtr)[i] == 'N')
{
// 'N' in the reference, so continue to the next read.
skipRead = true;
break;
}
}
if(skipRead)
{
continue;
}
}
// Update the gapInfo.
gapInfoMap[gapSize]++;
}
}
if(!detailed)
{
// Output the summary.
ifprintf(outFile, "GapSize\tNumPairs\n");
for(std::map<int,int>::iterator iter = gapInfoMap.begin();
iter != gapInfoMap.end(); iter++)
{
ifprintf(outFile, "%d\t%d\n", (*iter).first, (*iter).second);
}
}
SamStatus::Status returnStatus = samIn.GetStatus();
if(returnStatus == SamStatus::NO_MORE_RECS)
{
return(SamStatus::SUCCESS);
}
return(returnStatus);
}
// 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);
}
int WriteRegion::execute(int argc, char **argv)
{
// Extract command line arguments.
String inFile = "";
String outFile = "";
String indexFile = "";
String readName = "";
String bed = "";
myStart = UNSPECIFIED_INT;
myEnd = UNSPECIFIED_INT;
myPrevStart = UNSPECIFIED_INT;
myPrevEnd = UNSPECIFIED_INT;
myRefID = UNSET_REF;
myRefName.Clear();
myPrevRefName.Clear();
myBedRefID = SamReferenceInfo::NO_REF_ID;
bool lshift = false;
bool noeof = false;
bool params = false;
myWithinReg = false;
myWroteReg = false;
ParameterList inputParameters;
BEGIN_LONG_PARAMETERS(longParameterList)
LONG_PARAMETER_GROUP("Required Parameters")
LONG_STRINGPARAMETER("in", &inFile)
LONG_STRINGPARAMETER("out", &outFile)
LONG_PARAMETER_GROUP("Optional Region Parameters")
LONG_STRINGPARAMETER("bamIndex", &indexFile)
LONG_STRINGPARAMETER("refName", &myRefName)
LONG_INTPARAMETER("refID", &myRefID)
LONG_INTPARAMETER("start", &myStart)
LONG_INTPARAMETER("end", &myEnd)
LONG_STRINGPARAMETER("bed", &bed)
LONG_PARAMETER("withinReg", &myWithinReg)
LONG_STRINGPARAMETER("readName", &readName)
LONG_PARAMETER_GROUP("Optional Other Parameters")
LONG_PARAMETER("lshift", &lshift)
LONG_PARAMETER("noeof", &noeof)
LONG_PARAMETER("params", ¶ms)
END_LONG_PARAMETERS();
inputParameters.Add(new LongParameters ("Input Parameters",
longParameterList));
inputParameters.Read(argc-1, &(argv[1]));
// If no eof block is required for a bgzf file, set the bgzf file type to
// not look for it.
if(noeof)
{
// Set that the eof block is not required.
BgzfFileType::setRequireEofBlock(false);
}
// Check to see if the in file was specified, if not, report an error.
if(inFile == "")
{
usage();
// mandatory argument was not specified.
inputParameters.Status();
std::cerr << "Missing mandatory argument: --in" << std::endl;
return(-1);
}
if(outFile == "")
{
usage();
// mandatory argument was not specified.
inputParameters.Status();
std::cerr << "Missing mandatory argument: --out" << std::endl;
return(-1);
}
if(indexFile == "")
{
// In file was not specified, so set it to the in file
// + ".bai"
indexFile = inFile + ".bai";
}
if(myRefID != UNSET_REF && myRefName.Length() != 0)
{
std::cerr << "Can't specify both refID and refName" << std::endl;
inputParameters.Status();
return(-1);
}
if(myRefID != UNSET_REF && bed.Length() != 0)
{
std::cerr << "Can't specify both refID and bed" << std::endl;
inputParameters.Status();
return(-1);
}
if(myRefName.Length() != 0 && bed.Length() != 0)
{
std::cerr << "Can't specify both refName and bed" << std::endl;
inputParameters.Status();
return(-1);
}
if(!bed.IsEmpty())
{
myBedFile = ifopen(bed, "r");
}
if(params)
{
inputParameters.Status();
}
// Open the file for reading.
mySamIn.OpenForRead(inFile);
// Open the output file for writing.
SamFile samOut;
samOut.OpenForWrite(outFile);
// Open the bam index file for reading if a region was specified.
if((myRefName.Length() != 0) || (myRefID != UNSET_REF) || (myBedFile != NULL))
{
mySamIn.ReadBamIndex(indexFile);
}
// Read & write the sam header.
mySamIn.ReadHeader(mySamHeader);
samOut.WriteHeader(mySamHeader);
// Read the sam records.
SamRecord samRecord;
// Track the status.
int numSectionRecords = 0;
// Set returnStatus to success. It will be changed
// to the failure reason if any of the writes fail.
SamStatus::Status returnStatus = SamStatus::SUCCESS;
while(getNextSection())
{
// Keep reading records until they aren't anymore.
while(mySamIn.ReadRecord(mySamHeader, samRecord))
{
if(!readName.IsEmpty())
{
// Check for readname.
if(strcmp(samRecord.getReadName(), readName.c_str()) != 0)
{
// not a matching read name, so continue to the next record.
continue;
}
}
// Check to see if the read has already been processed.
if(myPrevEnd != UNSPECIFIED_INT)
{
// Because we already know that the bed was sorted,
// we know that the previous section started before
// this one, so if the previous end is greater than
// this record's end position we know that it
// was already written in the previous section.
// Note: can't be equal to the previous end since
// the end range was exclusive, while
// get0BasedAlignmentEnd is inclusive.
// myPrevEnd is reset by getNextSection when a new
// chromosome is hit.
if(samRecord.get0BasedAlignmentEnd() < myPrevEnd)
{
// This record was already written.
continue;
}
}
// Shift left if applicable.
if(lshift)
{
samRecord.shiftIndelsLeft();
}
// Successfully read a record from the file, so write it.
samOut.WriteRecord(mySamHeader, samRecord);
++numSectionRecords;
}
myWroteReg = true;
}
if(myBedFile != NULL)
{
ifclose(myBedFile);
}
std::cerr << "Wrote " << outFile << " with " << numSectionRecords
<< " records.\n";
return(returnStatus);
}
