//{{{ SV_Pair:: SV_Pair(const BamAlignment &bam_a,
// if both reads are on the same chrome, then read_l must map before read_r
// if the reads are on different strands then read_l must be on the lexo
// lesser chrom (using the string.compare() method)
SV_Pair::
SV_Pair(const BamAlignment &bam_a,
const BamAlignment &bam_b,
const RefVector &refs,
int _weight,
int _ev_id,
SV_PairReader *_reader)
{
reader = _reader;
if ( bam_a.MapQuality < bam_b.MapQuality )
min_mapping_quality = bam_a.MapQuality;
else
min_mapping_quality = bam_b.MapQuality;
struct interval tmp_a, tmp_b;
tmp_a.start = bam_a.Position;
tmp_a.end = bam_a.GetEndPosition(false, false) - 1;
tmp_a.chr = refs.at(bam_a.RefID).RefName;
if ( bam_a.IsReverseStrand() == true )
tmp_a.strand = '-';
else
tmp_a.strand = '+';
tmp_b.start = bam_b.Position;
tmp_b.end = bam_b.GetEndPosition(false, false) - 1;
tmp_b.chr = refs.at(bam_b.RefID).RefName;
if ( bam_b.IsReverseStrand() == true )
tmp_b.strand = '-';
else
tmp_b.strand = '+';
//if ( tmp_a.chr.compare(tmp_b.chr) > 0 ) {
if ( bam_a.RefID < bam_b.RefID ) {
read_l = tmp_a;
read_r = tmp_b;
//} else if ( tmp_a.chr.compare(tmp_b.chr) < 0 ) {
} else if ( bam_a.RefID > bam_b.RefID) {
read_l = tmp_b;
read_r = tmp_a;
} else { // ==
if (tmp_a.start > tmp_b.start) {
read_l = tmp_b;
read_r = tmp_a;
} else {
read_l = tmp_a;
read_r = tmp_b;
}
}
weight = _weight;
ev_id = _ev_id;
}
void BedCoverage::CollectCoverageBam(string bamFile) {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bedB->loadBedCovFileIntoMap();
// open the BAM file
BamReader reader;
reader.Open(bamFile);
// get header & reference information
string header = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// convert each aligned BAM entry to BED
// and compute coverage on B
BamAlignment bam;
while (reader.GetNextAlignment(bam)) {
if (bam.IsMapped()) {
// treat the BAM alignment as a single "block"
if (_obeySplits == false) {
// construct a new BED entry from the current BAM alignment.
BED a;
a.chrom = refs.at(bam.RefID).RefName;
a.start = bam.Position;
a.end = bam.GetEndPosition(false, false);
a.strand = "+";
if (bam.IsReverseStrand()) a.strand = "-";
_bedB->countHits(a, _sameStrand, _diffStrand, _countsOnly);
}
// split the BAM alignment into discrete blocks and
// look for overlaps only within each block.
else {
// vec to store the discrete BED "blocks" from a
bedVector bedBlocks;
// since we are counting coverage, we do want to split blocks when a
// deletion (D) CIGAR op is encountered (hence the true for the last parm)
GetBamBlocks(bam, refs.at(bam.RefID).RefName, bedBlocks, false, true);
// use countSplitHits to avoid over-counting each split chunk
// as distinct read coverage.
_bedB->countSplitHits(bedBlocks, _sameStrand, _diffStrand, _countsOnly);
}
}
}
// report the coverage (summary or histogram) for BED B.
if (_countsOnly == true)
ReportCounts();
else
ReportCoverage();
// close the BAM file
reader.Close();
}
//{{{ void process_pair(const BamAlignment &curr,
void
SV_Pair::
process_pair(const BamAlignment &curr,
const RefVector refs,
map<string, BamAlignment> &mapped_pairs,
UCSCBins<SV_BreakPoint*> &r_bin,
int weight,
int ev_id,
SV_PairReader *reader)
{
if (mapped_pairs.find(curr.Name) == mapped_pairs.end())
mapped_pairs[curr.Name] = curr;
else {
SV_Pair *new_pair = new SV_Pair(mapped_pairs[curr.Name],
curr,
refs,
weight,
ev_id,
reader);
//cerr << count_clipped(curr.CigarData) << "\t" <<
//count_clipped(mapped_pairs[curr.Name].CigarData) << endl;
if ( new_pair->is_sane() &&
new_pair->is_aberrant() &&
(count_clipped(curr.CigarData) > 0) &&
(count_clipped(mapped_pairs[curr.Name].CigarData) > 0) ) {
SV_BreakPoint *new_bp = new_pair->get_bp();
#ifdef TRACE
cerr << "READ\t" <<
refs.at(mapped_pairs[curr.Name].RefID).RefName << "," <<
mapped_pairs[curr.Name].Position << "," <<
(mapped_pairs[curr.Name].GetEndPosition(false, false) - 1)
<< "\t" <<
refs.at(curr.RefID).RefName << "," <<
curr.Position << "," <<
(curr.GetEndPosition(false, false) - 1)
<<
endl;
cerr << "\tPE\t" << *new_bp << endl;
#endif
new_bp->cluster(r_bin);
} else {
delete(new_pair);
}
mapped_pairs.erase(curr.Name);
}
}
long get_ref_lengths(int id, RefVector ref) {
long length = 0;
for (size_t i = 0; i < (size_t) id && i < ref.size(); i++) {
length += (long) ref[i].RefLength + (long) Parameter::Instance()->max_dist;
}
return length;
}
GenomeFile::GenomeFile(const RefVector &genome) {
for (size_t i = 0; i < genome.size(); ++i) {
string chrom = genome[i].RefName;
int length = genome[i].RefLength;
_chromSizes[chrom] = length;
_chromList.push_back(chrom);
}
}
long Breakpoint::calc_pos(long pos, RefVector ref) {
size_t i = 0;
pos -= ref[i].RefLength;
while (i < ref.size() && pos >= 0) {
i++;
pos -= ref[i].RefLength;
}
return pos + ref[i].RefLength;
}
void getBamBlocks(const BamAlignment &bam, const RefVector &refs,
vector<BED> &blocks, bool breakOnDeletionOps) {
CHRPOS currPosition = bam.Position;
CHRPOS blockStart = bam.Position;
string chrom = refs.at(bam.RefID).RefName;
string name = bam.Name;
string strand = "+";
string score = ToString(bam.MapQuality);
char prevOp = '\0';
if (bam.IsReverseStrand()) strand = "-";
bool blocksFound = false;
vector<CigarOp>::const_iterator cigItr = bam.CigarData.begin();
vector<CigarOp>::const_iterator cigEnd = bam.CigarData.end();
for ( ; cigItr != cigEnd; ++cigItr ) {
if (cigItr->Type == 'M') {
currPosition += cigItr->Length;
// we only want to create a new block if the current M op
// was preceded by an N op or a D op (and we are breaking on D ops)
if ((prevOp == 'D' && breakOnDeletionOps == true) || (prevOp == 'N')) {
blocks.push_back( BED(chrom, blockStart, currPosition, name, score, strand) );
blockStart = currPosition;
}
}
else if (cigItr->Type == 'D') {
if (breakOnDeletionOps == false)
currPosition += cigItr->Length;
else {
blocksFound = true;
currPosition += cigItr->Length;
blockStart = currPosition;
}
}
else if (cigItr->Type == 'N') {
blocks.push_back( BED(chrom, blockStart, currPosition, name, score, strand) );
blocksFound = true;
currPosition += cigItr->Length;
blockStart = currPosition;
}
else if (cigItr->Type == 'S' || cigItr->Type == 'H' || cigItr->Type == 'P' || cigItr->Type == 'I') {
// do nothing
}
else {
cerr << "Input error: invalid CIGAR type (" << cigItr->Type
<< ") for: " << bam.Name << endl;
exit(1);
}
prevOp = cigItr->Type;
}
// if there were no splits, we just create a block representing the contiguous alignment.
if (blocksFound == false) {
blocks.push_back( BED(chrom, bam.Position, currPosition, name, score, strand) );
}
}
long fuck_off(long pos, RefVector ref, std::string &chr) {
size_t i = 0;
pos -= (ref[i].RefLength + Parameter::Instance()->max_dist);
while (i < ref.size() && pos >= 0) {
i++;
pos -= ((long) ref[i].RefLength + (long) Parameter::Instance()->max_dist);
}
chr = ref[i].RefName;
return pos + ref[i].RefLength + (long) Parameter::Instance()->max_dist;
}
std::string Breakpoint::get_chr(long pos, RefVector ref) {
// std::cout << "pos: " << pos << std::endl;
size_t id = 0;
while (id < ref.size() && pos >= 0) {
pos -= (long) ref[id].RefLength;
// std::cout << id << std::endl;
id++;
}
return ref[id - 1].RefName;
}
//-------------------------------------------------------------------------
void XList::sortByElementNumber(String order){
// Get the number of sessions per speaker
LKVector spk(0,0);
for(unsigned long i=0;i<_vector.size();i++){
LKVector::type sps;
sps.idx = i;
sps.lk = _vector.getObject(i).getElementCount();
spk.addValue(sps);
}
// Sort Xlines of the temporary XList by element number
spk.descendingSort();
// Copy the current RefVector<XLine> into a temporary one
RefVector<XLine> tmpX;
for(unsigned long i=0;i<_vector.size();i++){
XLine *ll = new XLine(_vector.getObject(i));
tmpX.addObject(*ll);
}
// Remove all elements from the XList
_vector.deleteAllObjects();
// Fill the XList according to the number of elements
if(order == "descend"){
for(unsigned long i=0;i<tmpX.size();i++){
_vector.addObject(tmpX.getObject(spk[i].idx));
}
}
else if(order == "ascend"){
for(long i=tmpX.size()-1;i>=0;i--){
_vector.addObject(tmpX.getObject(spk[i].idx));
}
}
}
void getBamBlocks(const BamAlignment &bam, const RefVector &refs,
BedVec &blocks, bool breakOnDeletionOps) {
CHRPOS currPosition = bam.Position;
CHRPOS blockStart = bam.Position;
string chrom = refs.at(bam.RefID).RefName;
string name = bam.Name;
string strand = "+";
float score = bam.MapQuality;
if (bam.IsReverseStrand()) strand = "-";
vector<CigarOp>::const_iterator cigItr = bam.CigarData.begin();
vector<CigarOp>::const_iterator cigEnd = bam.CigarData.end();
for ( ; cigItr != cigEnd; ++cigItr ) {
if (cigItr->Type == 'M') {
currPosition += cigItr->Length;
blocks.push_back( Bed(chrom, blockStart, currPosition, name, score, strand) );
blockStart = currPosition;
}
else if (cigItr->Type == 'D') {
if (breakOnDeletionOps == false)
currPosition += cigItr->Length;
else {
currPosition += cigItr->Length;
blockStart = currPosition;
}
}
else if (cigItr->Type == 'N') {
currPosition += cigItr->Length;
blockStart = currPosition; }
else if (cigItr->Type == 'S' || cigItr->Type == 'H' || cigItr->Type == 'P' || cigItr->Type == 'I') {
// do nothing
}
else {
cerr << "Input error: invalid CIGAR type (" << cigItr->Type
<< ") for: " << bam.Name << endl;
exit(1);
}
}
}
void BedIntersect::IntersectBam(string bamFile) {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bedB = new BedFile(_bedBFile);
_bedB->loadBedFileIntoMap();
// create a dummy BED A file for printing purposes if not
// using BAM output.
if (_bamOutput == false) {
_bedA = new BedFile(_bedAFile);
_bedA->bedType = 12;
}
// open the BAM file
BamReader reader;
BamWriter writer;
reader.Open(bamFile);
// get header & reference information
string bamHeader = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// open a BAM output to stdout if we are writing BAM
if (_bamOutput == true) {
// set compression mode
BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
if ( _isUncompressedBam ) compressionMode = BamWriter::Uncompressed;
writer.SetCompressionMode(compressionMode);
// open our BAM writer
writer.Open("stdout", bamHeader, refs);
}
vector<BED> hits;
// reserve some space
hits.reserve(100);
BamAlignment bam;
// get each set of alignments for each pair.
while (reader.GetNextAlignment(bam)) {
// save an unaligned read if -v
if (!bam.IsMapped()) {
if (_noHit == true)
writer.SaveAlignment(bam);
continue;
}
// break alignment into discrete blocks,
bedVector bed_blocks;
string chrom = refs.at(bam.RefID).RefName;
GetBamBlocks(bam, chrom, bed_blocks, false, true);
// create a basic BED entry from the BAM alignment
BED bed;
MakeBedFromBam(bam, chrom, bed_blocks, bed);
bool overlapsFound = false;
if ((_bamOutput == true) && (_obeySplits == false))
{
overlapsFound = _bedB->anyHits(bed.chrom, bed.start, bed.end,
bed.strand, _sameStrand, _diffStrand,
_overlapFraction, _reciprocal);
}
else if ( ((_bamOutput == true) && (_obeySplits == true)) ||
((_bamOutput == false) && (_obeySplits == true)) )
{
// find the hits that overlap with the full span of the blocked BED
_bedB->allHits(bed.chrom, bed.start, bed.end, bed.strand,
hits, _sameStrand, _diffStrand,
_overlapFraction, _reciprocal);
// find the overlaps between the block in A and B
overlapsFound = FindBlockedOverlaps(bed, bed_blocks, hits, _bamOutput);
}
else if ((_bamOutput == false) && (_obeySplits == false))
{
FindOverlaps(bed, hits);
}
// save the BAM alignment if overlap reqs. were met
if (_bamOutput == true) {
if ((overlapsFound == true) && (_noHit == false))
writer.SaveAlignment(bam);
else if ((overlapsFound == false) && (_noHit == true))
writer.SaveAlignment(bam);
}
hits.clear();
}
// close the relevant BAM files.
reader.Close();
if (_bamOutput == true) {
writer.Close();
}
}
void TagBam::Tag() {
// open the annotations files for processing;
OpenAnnoFiles();
// open the BAM file
BamReader reader;
BamWriter writer;
if (!reader.Open(_bamFile)) {
cerr << "Failed to open BAM file " << _bamFile << endl;
exit(1);
}
// get header & reference information
string bamHeader = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// set compression mode
BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
// if ( _isUncompressedBam ) compressionMode = BamWriter::Uncompressed;
writer.SetCompressionMode(compressionMode);
// open our BAM writer
writer.Open("stdout", bamHeader, refs);
// rip through the BAM file and test for overlaps with each annotation file.
BamAlignment al;
vector<BED> hits;
while (reader.GetNextAlignment(al)) {
if (al.IsMapped() == true) {
BED a;
a.chrom = refs.at(al.RefID).RefName;
a.start = al.Position;
a.end = al.GetEndPosition(false, false);
a.strand = "+";
if (al.IsReverseStrand()) a.strand = "-";
ostringstream annotations;
// annotate the BAM file based on overlaps with the annotation files.
for (size_t i = 0; i < _annoFiles.size(); ++i)
{
// grab the current annotation file.
BedFile *anno = _annoFiles[i];
if (!_useNames && !_useScores && !_useIntervals) {
// add the label for this annotation file to tag if there is overlap
if (anno->anyHits(a.chrom, a.start, a.end, a.strand,
_sameStrand, _diffStrand, _overlapFraction, false))
{
annotations << _annoLabels[i] << ";";
}
}
// use the score field
else if (!_useNames && _useScores && !_useIntervals) {
anno->allHits(a.chrom, a.start, a.end, a.strand,
hits, _sameStrand, _diffStrand, 0.0, false);
for (size_t i = 0; i < hits.size(); ++i) {
annotations << hits[i].score;
if (i < hits.size() - 1) annotations << ",";
}
if (hits.size() > 0) annotations << ";";
hits.clear();
}
// use the name field from the annotation files to populate tag
else if (_useNames && !_useScores && !_useIntervals) {
anno->allHits(a.chrom, a.start, a.end, a.strand,
hits, _sameStrand, _diffStrand, 0.0, false);
for (size_t j = 0; j < hits.size(); ++j) {
annotations << hits[j].name;
if (j < hits.size() - 1) annotations << ",";
}
if (hits.size() > 0) annotations << ";";
hits.clear();
}
// use the full interval information annotation files to populate tag
else if (!_useNames && !_useScores && _useIntervals) {
anno->allHits(a.chrom, a.start, a.end, a.strand,
hits, _sameStrand, _diffStrand, 0.0, false);
for (size_t j = 0; j < hits.size(); ++j) {
annotations << _annoLabels[i] << ":" <<
hits[j].chrom << ":" <<
hits[j].start << "-" <<
hits[j].end << "," <<
hits[j].name << "," <<
hits[j].score << "," <<
hits[j].strand;
if (j < hits.size() - 1) annotations << ",";
}
if (hits.size() > 0) annotations << ";";
hits.clear();
}
}
// were there any overlaps with which to make a tag?
if (annotations.str().size() > 0) {
al.AddTag(_tag, "Z", annotations.str().substr(0, annotations.str().size() - 1)); // get rid of the last ";"
}
}
writer.SaveAlignment(al);
}
reader.Close();
writer.Close();
// close the annotations files;
CloseAnnoFiles();
}
// this has been copied from bamtools utilities, since it isn't in the API. Original file is bamtools_utilities.cpp.
// Like the rest of Bamtools, it is under the BSD license.
bool Filter::ParseRegionString(const string& regionString, BamRegion& region)
{
// -------------------------------
// parse region string
// check first for empty string
if ( regionString.empty() )
return false;
// non-empty string, look for a colom
size_t foundFirstColon = regionString.find(':');
// store chrom strings, and numeric positions
string chrom;
int startPos;
int stopPos;
// no colon found
// going to use entire contents of requested chromosome
// just store entire region string as startChrom name
// use BamReader methods to check if its valid for current BAM file
if ( foundFirstColon == string::npos ) {
chrom = regionString;
startPos = 0;
stopPos = -1;
}
// colon found, so we at least have some sort of startPos requested
else {
// store start chrom from beginning to first colon
chrom = regionString.substr(0,foundFirstColon);
// look for ".." after the colon
size_t foundRangeDots = regionString.find("..", foundFirstColon+1);
// no dots found
// so we have a startPos but no range
// store contents before colon as startChrom, after as startPos
if ( foundRangeDots == string::npos ) {
startPos = atoi( regionString.substr(foundFirstColon+1).c_str() );
stopPos = -1;
}
// ".." found, so we have some sort of range selected
else {
// store startPos between first colon and range dots ".."
startPos = atoi( regionString.substr(foundFirstColon+1, foundRangeDots-foundFirstColon-1).c_str() );
// look for second colon
size_t foundSecondColon = regionString.find(':', foundRangeDots+1);
// no second colon found
// so we have a "standard" chrom:start..stop input format (on single chrom)
if ( foundSecondColon == string::npos ) {
stopPos = atoi( regionString.substr(foundRangeDots+2).c_str() );
} else {
return false;
}
}
}
// -------------------------------
// validate reference IDs & genomic positions
const RefVector references = getReferences();
int RefID = -1;
for(int i = 0; i < references.size(); i++) {
if(references[i].RefName == chrom)
RefID = i;
}
// if startRefID not found, return false
if ( RefID == -1 ) {
cerr << "Can't find chromosome'" << chrom << "'" << endl;
return false;
}
// startPos cannot be greater than or equal to reference length
const RefData& startReference = references.at(RefID);
if ( startPos >= startReference.RefLength ) {
cerr << "Start position (" << startPos << ") after end of the reference sequence (" << startReference.RefLength << ")" << endl;
return false;
}
// stopPosition cannot be larger than reference length
const RefData& stopReference = references.at(RefID);
if ( stopPos > stopReference.RefLength ) {
cerr << "Start position (" << stopPos << ") after end of the reference sequence (" << stopReference.RefLength << ")" << endl;
return false;
}
// if no stopPosition specified, set to reference end
if ( stopPos == -1 ) stopPos = stopReference.RefLength;
// -------------------------------
// set up Region struct & return
region.LeftRefID = RefID;
region.LeftPosition = startPos;
region.RightRefID = RefID;;
region.RightPosition = stopPos;
return true;
}
bool check(const PropertyFilter& filter, const BamAlignment& al) {
bool keepAlignment = true;
const PropertyMap& properties = filter.Properties;
PropertyMap::const_iterator propertyIter = properties.begin();
PropertyMap::const_iterator propertyEnd = properties.end();
for ( ; propertyIter != propertyEnd; ++propertyIter ) {
// check alignment data field depending on propertyName
const string& propertyName = (*propertyIter).first;
const PropertyFilterValue& valueFilter = (*propertyIter).second;
if ( propertyName == ALIGNMENTFLAG_PROPERTY ) keepAlignment &= valueFilter.check(al.AlignmentFlag);
else if ( propertyName == CIGAR_PROPERTY ) {
stringstream cigarSs;
const vector<CigarOp>& cigarData = al.CigarData;
if ( !cigarData.empty() ) {
vector<CigarOp>::const_iterator cigarBegin = cigarData.begin();
vector<CigarOp>::const_iterator cigarIter = cigarBegin;
vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
for ( ; cigarIter != cigarEnd; ++cigarIter ) {
const CigarOp& op = (*cigarIter);
cigarSs << op.Length << op.Type;
}
keepAlignment &= valueFilter.check(cigarSs.str());
}
}
else if ( propertyName == INSERTSIZE_PROPERTY ) keepAlignment &= valueFilter.check(al.InsertSize);
else if ( propertyName == ISDUPLICATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsDuplicate());
else if ( propertyName == ISFAILEDQC_PROPERTY ) keepAlignment &= valueFilter.check(al.IsFailedQC());
else if ( propertyName == ISFIRSTMATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsFirstMate());
else if ( propertyName == ISMAPPED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMapped());
else if ( propertyName == ISMATEMAPPED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMateMapped());
else if ( propertyName == ISMATEREVERSESTRAND_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMateReverseStrand());
else if ( propertyName == ISPAIRED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsPaired());
else if ( propertyName == ISPRIMARYALIGNMENT_PROPERTY ) keepAlignment &= valueFilter.check(al.IsPrimaryAlignment());
else if ( propertyName == ISPROPERPAIR_PROPERTY ) keepAlignment &= valueFilter.check(al.IsProperPair());
else if ( propertyName == ISREVERSESTRAND_PROPERTY ) keepAlignment &= valueFilter.check(al.IsReverseStrand());
else if ( propertyName == ISSECONDMATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsSecondMate());
else if ( propertyName == ISSINGLETON_PROPERTY ) {
const bool isSingleton = al.IsPaired() && al.IsMapped() && !al.IsMateMapped();
keepAlignment &= valueFilter.check(isSingleton);
}
else if ( propertyName == MAPQUALITY_PROPERTY ) keepAlignment &= valueFilter.check(al.MapQuality);
else if ( propertyName == MATEPOSITION_PROPERTY ) keepAlignment &= ( al.IsPaired() && al.IsMateMapped() && valueFilter.check(al.MateRefID) );
else if ( propertyName == MATEREFERENCE_PROPERTY ) {
if ( !al.IsPaired() || !al.IsMateMapped() ) return false;
BAMTOOLS_ASSERT_MESSAGE( (al.MateRefID>=0 && (al.MateRefID<(int)filterToolReferences.size())), "Invalid MateRefID");
const string& refName = filterToolReferences.at(al.MateRefID).RefName;
keepAlignment &= valueFilter.check(refName);
}
else if ( propertyName == NAME_PROPERTY ) keepAlignment &= valueFilter.check(al.Name);
else if ( propertyName == POSITION_PROPERTY ) keepAlignment &= valueFilter.check(al.Position);
else if ( propertyName == QUERYBASES_PROPERTY ) keepAlignment &= valueFilter.check(al.QueryBases);
else if ( propertyName == REFERENCE_PROPERTY ) {
BAMTOOLS_ASSERT_MESSAGE( (al.RefID>=0 && (al.RefID<(int)filterToolReferences.size())), "Invalid RefID");
const string& refName = filterToolReferences.at(al.RefID).RefName;
keepAlignment &= valueFilter.check(refName);
}
else if ( propertyName == TAG_PROPERTY ) keepAlignment &= checkAlignmentTag(valueFilter, al);
else BAMTOOLS_ASSERT_UNREACHABLE;
// if alignment fails at ANY point, just quit and return false
if ( !keepAlignment ) return false;
}
BAMTOOLS_ASSERT_MESSAGE( keepAlignment, "Error in BamAlignmentChecker... keepAlignment should be true here");
return keepAlignment;
}
void BedGenomeCoverage::CoverageBam(string bamFile) {
ResetChromCoverage();
// open the BAM file
BamReader reader;
if (!reader.Open(bamFile)) {
cerr << "Failed to open BAM file " << bamFile << endl;
exit(1);
}
// get header & reference information
string header = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// load the BAM header references into a BEDTools "genome file"
_genome = new GenomeFile(refs);
// convert each aligned BAM entry to BED
// and compute coverage on B
BamAlignment bam;
while (reader.GetNextAlignment(bam)) {
// skip if the read is unaligned
if (bam.IsMapped() == false)
continue;
bool _isReverseStrand = bam.IsReverseStrand();
//changing second mate's strand to opposite
if( _dUTP && bam.IsPaired() && bam.IsMateMapped() && bam.IsSecondMate())
_isReverseStrand = !bam.IsReverseStrand();
// skip if we care about strands and the strand isn't what
// the user wanted
if ( (_filterByStrand == true) &&
((_requestedStrand == "-") != _isReverseStrand) )
continue;
// extract the chrom, start and end from the BAM alignment
string chrom(refs.at(bam.RefID).RefName);
CHRPOS start = bam.Position;
CHRPOS end = bam.GetEndPosition(false, false) - 1;
// are we on a new chromosome?
if ( chrom != _currChromName )
StartNewChrom(chrom);
if(_pair_chip_) {
// Skip if not a proper pair
if (bam.IsPaired() && (!bam.IsProperPair() or !bam.IsMateMapped()) )
continue;
// Skip if wrong coordinates
if( ( (bam.Position<bam.MatePosition) && bam.IsReverseStrand() ) ||
( (bam.MatePosition < bam.Position) && bam.IsMateReverseStrand() ) ) {
//chemically designed: left on positive strand, right on reverse one
continue;
}
/*if(_haveSize) {
if (bam.IsFirstMate() && bam.IsReverseStrand()) { //put fragmentSize in to the middle of pair end_fragment
int mid = bam.MatePosition+abs(bam.InsertSize)/2;
if(mid<_fragmentSize/2)
AddCoverage(0, mid+_fragmentSize/2);
else
AddCoverage(mid-_fragmentSize/2, mid+_fragmentSize/2);
}
else if (bam.IsFirstMate() && bam.IsMateReverseStrand()) { //put fragmentSize in to the middle of pair end_fragment
int mid = start+abs(bam.InsertSize)/2;
if(mid<_fragmentSize/2)
AddCoverage(0, mid+_fragmentSize/2);
else
AddCoverage(mid-_fragmentSize/2, mid+_fragmentSize/2);
}
} else */
if (bam.IsFirstMate() && bam.IsReverseStrand()) { //prolong to the mate to the left
AddCoverage(bam.MatePosition, end);
}
else if (bam.IsFirstMate() && bam.IsMateReverseStrand()) { //prolong to the mate to the right
AddCoverage(start, start + abs(bam.InsertSize) - 1);
}
} else if (_haveSize) {
if(bam.IsReverseStrand()) {
if(end<_fragmentSize) { //sometimes fragmentSize is bigger :(
AddCoverage(0, end);
} else {
AddCoverage(end + 1 - _fragmentSize, end );
}
} else {
AddCoverage(start,start+_fragmentSize - 1);
}
} else
// add coverage accordingly.
if (!_only_5p_end && !_only_3p_end) {
bedVector bedBlocks;
// we always want to split blocks when a D CIGAR op is found.
// if the user invokes -split, we want to also split on N ops.
if (_obeySplits) { // "D" true, "N" true
GetBamBlocks(bam, refs.at(bam.RefID).RefName, bedBlocks, true, true);
}
else { // "D" true, "N" false
GetBamBlocks(bam, refs.at(bam.RefID).RefName, bedBlocks, true, false);
}
AddBlockedCoverage(bedBlocks);
}
else if (_only_5p_end) {
CHRPOS pos = ( !bam.IsReverseStrand() ) ? start : end;
AddCoverage(pos,pos);
}
else if (_only_3p_end) {
CHRPOS pos = ( bam.IsReverseStrand() ) ? start : end;
AddCoverage(pos,pos);
}
}
// close the BAM
reader.Close();
// process the results of the last chromosome.
ReportChromCoverage(_currChromCoverage, _currChromSize,
_currChromName, _currChromDepthHist);
// report all empty chromsomes
PrintEmptyChromosomes();
// report the overall coverage if asked.
PrintFinalCoverage();
}
//{{{ SV_SplitRead:: SV_SplitRead(vector< BamAlignment > &block,
SV_SplitRead::
SV_SplitRead(const BamAlignment &bam_a,
const BamAlignment &bam_b,
const RefVector &refs,
int _weight,
int _id,
int _sample_id,
SV_SplitReadReader *_reader)
{
reader = _reader;
sample_id = _sample_id;
if ( bam_a.MapQuality < bam_b.MapQuality )
min_mapping_quality = bam_a.MapQuality;
else
min_mapping_quality = bam_b.MapQuality;
struct cigar_query query_a =
calc_query_pos_from_cigar(bam_a.CigarData,
bam_a.IsReverseStrand() );
struct cigar_query query_b =
calc_query_pos_from_cigar(bam_b.CigarData,
bam_b.IsReverseStrand() );
struct interval tmp_a, tmp_b;
tmp_a.strand = '+';
if (bam_a.IsReverseStrand())
tmp_a.strand = '-';
tmp_a.chr = refs.at(bam_a.RefID).RefName;
tmp_a.start = bam_a.Position;
tmp_a.end = bam_a.GetEndPosition();
tmp_b.strand = '+';
if (bam_b.IsReverseStrand())
tmp_b.strand = '-';
tmp_b.chr = refs.at(bam_b.RefID).RefName;
tmp_b.start = bam_b.Position;
tmp_b.end = bam_b.GetEndPosition();
//if ( ( tmp_a.chr.compare(tmp_b.chr) > 0 ) ||
//( ( tmp_a.chr.compare(tmp_b.chr) == 0 ) &&
//( tmp_a.start > tmp_b.start ) ) ) {
if ( (bam_a.RefID > bam_b.RefID) ||
( (bam_a.RefID == bam_b.RefID) &&
(tmp_a.start > tmp_b.start ) ) ) {
side_r = tmp_a;
side_l = tmp_b;
query_r = query_a;
query_l = query_b;
} else {
side_l = tmp_a;
side_r = tmp_b;
query_l = query_a;
query_r = query_b;
}
if (side_l.strand != side_r.strand)
type = SV_BreakPoint::INVERSION;
else if ( ( ( side_l.strand == '+' ) &&
( side_r.strand == '+' ) &&
( query_l.qs_pos < query_r.qs_pos ) ) ||
( ( side_l.strand == '-' ) &&
( side_r.strand == '-' ) &&
( query_l.qs_pos > query_r.qs_pos) ) )
type = SV_BreakPoint::DELETION;
else if ( ( ( side_l.strand == '+' ) &&
( side_r.strand == '+' ) &&
( query_l.qs_pos > query_r.qs_pos ) ) ||
( ( side_l.strand == '-' ) &&
( side_r.strand == '-' ) &&
( query_l.qs_pos < query_r.qs_pos) ) )
type = SV_BreakPoint::DUPLICATION;
else {
cerr << "ERROR IN BAM FILE. " <<
"TYPE not detected (DELETION,DUPLICATION,INVERSION)" <<
endl;
cerr << "\t" << query_l.qs_pos << "," << side_l.strand << "\t" <<
query_r.qs_pos << "," << side_r.strand << "\t" <<
tmp_a.chr << "," << tmp_a.start << "," << tmp_a.end << "\t" <<
tmp_b.chr << "," << tmp_b.start << "," << tmp_b.end << "\t" <<
endl;
throw(1);
}
weight = _weight;
id = _id;
}
// ValidateReaders checks that all the readers point to BAM files representing
// alignments against the same set of reference sequences, and that the
// sequences are identically ordered. If these checks fail the operation of
// the multireader is undefined, so we force program exit.
bool BamMultiReaderPrivate::ValidateReaders() const
{
m_errorString.clear();
// skip if 0 or 1 readers opened
if (m_readers.empty() || (m_readers.size() == 1)) return true;
// retrieve first reader
const MergeItem& firstItem = m_readers.front();
const BamReader* firstReader = firstItem.Reader;
if (firstReader == 0) return false;
// retrieve first reader's header data
const SamHeader& firstReaderHeader = firstReader->GetHeader();
const std::string& firstReaderSortOrder = firstReaderHeader.SortOrder;
// retrieve first reader's reference data
const RefVector& firstReaderRefData = firstReader->GetReferenceData();
const int firstReaderRefCount = firstReader->GetReferenceCount();
const int firstReaderRefSize = firstReaderRefData.size();
// iterate over all readers
std::vector<MergeItem>::const_iterator readerIter = m_readers.begin();
std::vector<MergeItem>::const_iterator readerEnd = m_readers.end();
for (; readerIter != readerEnd; ++readerIter) {
const MergeItem& item = (*readerIter);
BamReader* reader = item.Reader;
if (reader == 0) continue;
// get current reader's header data
const SamHeader& currentReaderHeader = reader->GetHeader();
const std::string& currentReaderSortOrder = currentReaderHeader.SortOrder;
// check compatible sort order
if (currentReaderSortOrder != firstReaderSortOrder) {
const std::string message =
std::string("mismatched sort order in ") + reader->GetFilename() + ", expected " +
firstReaderSortOrder + ", but found " + currentReaderSortOrder;
SetErrorString("BamMultiReader::ValidateReaders", message);
return false;
}
// get current reader's reference data
const RefVector currentReaderRefData = reader->GetReferenceData();
const int currentReaderRefCount = reader->GetReferenceCount();
const int currentReaderRefSize = currentReaderRefData.size();
// init reference data iterators
RefVector::const_iterator firstRefIter = firstReaderRefData.begin();
RefVector::const_iterator firstRefEnd = firstReaderRefData.end();
RefVector::const_iterator currentRefIter = currentReaderRefData.begin();
// compare reference counts from BamReader ( & container size, in case of BR error)
if ((currentReaderRefCount != firstReaderRefCount) ||
(firstReaderRefSize != currentReaderRefSize)) {
std::stringstream s;
s << "mismatched reference count in " << reader->GetFilename() << ", expected "
<< firstReaderRefCount << ", but found " << currentReaderRefCount;
SetErrorString("BamMultiReader::ValidateReaders", s.str());
return false;
}
// this will be ok; we just checked above that we have identically-sized sets of references
// here we simply check if they are all, in fact, equal in content
while (firstRefIter != firstRefEnd) {
const RefData& firstRef = (*firstRefIter);
const RefData& currentRef = (*currentRefIter);
// compare reference name & length
if ((firstRef.RefName != currentRef.RefName) ||
(firstRef.RefLength != currentRef.RefLength)) {
std::stringstream s;
s << "mismatched references found in" << reader->GetFilename()
<< "expected: " << std::endl;
// print first reader's reference data
RefVector::const_iterator refIter = firstReaderRefData.begin();
RefVector::const_iterator refEnd = firstReaderRefData.end();
for (; refIter != refEnd; ++refIter) {
const RefData& entry = (*refIter);
std::stringstream s;
s << entry.RefName << ' ' << std::endl;
}
s << "but found: " << std::endl;
// print current reader's reference data
refIter = currentReaderRefData.begin();
refEnd = currentReaderRefData.end();
for (; refIter != refEnd; ++refIter) {
const RefData& entry = (*refIter);
s << entry.RefName << ' ' << entry.RefLength << std::endl;
}
SetErrorString("BamMultiReader::ValidateReaders", s.str());
return false;
}
// update iterators
++firstRefIter;
++currentRefIter;
}
}
// if we get here, everything checks out
return true;
}
void BedWindow::WindowIntersectBam(string bamFile) {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bedB->loadBedFileIntoMap();
// open the BAM file
BamReader reader;
BamWriter writer;
reader.Open(bamFile);
// get header & reference information
string bamHeader = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// open a BAM output to stdout if we are writing BAM
if (_bamOutput == true) {
// set compression mode
BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
if ( _isUncompressedBam ) compressionMode = BamWriter::Uncompressed;
writer.SetCompressionMode(compressionMode);
// open our BAM writer
writer.Open("stdout", bamHeader, refs);
}
vector<BED> hits; // vector of potential hits
// reserve some space
hits.reserve(100);
_bedA->bedType = 6;
BamAlignment bam;
bool overlapsFound;
// get each set of alignments for each pair.
while (reader.GetNextAlignment(bam)) {
if (bam.IsMapped()) {
BED a;
a.chrom = refs.at(bam.RefID).RefName;
a.start = bam.Position;
a.end = bam.GetEndPosition(false, false);
// build the name field from the BAM alignment.
a.name = bam.Name;
if (bam.IsFirstMate()) a.name += "/1";
if (bam.IsSecondMate()) a.name += "/2";
a.score = ToString(bam.MapQuality);
a.strand = "+"; if (bam.IsReverseStrand()) a.strand = "-";
if (_bamOutput == true) {
overlapsFound = FindOneOrMoreWindowOverlaps(a);
if (overlapsFound == true) {
if (_noHit == false)
writer.SaveAlignment(bam);
}
else {
if (_noHit == true)
writer.SaveAlignment(bam);
}
}
else {
FindWindowOverlaps(a, hits);
hits.clear();
}
}
// BAM IsMapped() is false
else if (_noHit == true) {
writer.SaveAlignment(bam);
}
}
// close the relevant BAM files.
reader.Close();
if (_bamOutput == true) {
writer.Close();
}
}
int main(int argc, char* argv[])
{
//{{{ setup
double trim_threshold = 1e-10;
double merge_threshold = 1e-10;
int min_weight = 0;
int min_sample_weight = 0;
bool show_evidence = false;
bool has_pe_bams = false,
has_sr_bams = false,
has_bedpes = false;
CHR_POS window_size = 1000000;
string inter_chrom_file_prefix = "./";
int call_id = 0;
bool has_next = false;
vector<SV_EvidenceReader*>::iterator i_er;
UCSCBins<SV_BreakPoint*> r_bin;
srand (time(NULL));
string exclude_bed_file;
bool has_exclude = false;
string genome_file;
bool has_genome_file = false;
int print_prob = 0;
int bedpe_output = 0;
//vector<string> bam_files;
//}}}
//{{{ check to see if we should print out some help
if (argc == 1)
ShowHelp();
for(int i = 1; i < argc; i++) {
int parameterLength = (int)strlen(argv[i]);
if( (PARAMETER_CHECK("-h", 2, parameterLength)) ||
(PARAMETER_CHECK("--help", 6, parameterLength))) {
ShowHelp();
}
}
//}}}
//{{{ do some parsing and setup
vector<SV_EvidenceReader*> evidence_readers;
map<pair<string,string>, SV_EvidenceReader*> bam_evidence_readers;
for(int i = 1; i < argc; i++) {
int parameterLength = (int)strlen(argv[i]);
if(PARAMETER_CHECK("-pe", 3, parameterLength)) {
//{{{
has_pe_bams = true;
SV_PairReader *pe_r = new SV_PairReader();
if ((i+1) < argc) {
char *params = argv[i + 1];
char *param_val, *brka, *brkb;
for ( param_val = strtok_r(params, ",", &brka);
param_val;
param_val = strtok_r(NULL, ",", &brka)) {
char *param = strtok_r(param_val, ":", &brkb);
char *val = strtok_r(NULL, ":", &brkb);
if (val == NULL) {
cerr << "Parameter required for " << param << endl;
ShowHelp();
}
if ( ! pe_r->add_param(param, val) ) {
cerr << "Unknown pair end parameter:" << param << endl;
ShowHelp();
}
}
}
string msg = pe_r->check_params();
if ( msg.compare("") == 0 ) {
// Add to list of readers
// Set the distro map
pe_r->initialize();
SV_Evidence::distros[pe_r->ev_id] =
pair<log_space*,log_space*>(
SV_Pair::get_bp_interval_probability('+',
pe_r->distro_size,
pe_r->distro),
SV_Pair::get_bp_interval_probability('-',
pe_r->distro_size,
pe_r->distro));
SV_Evidence::distros_size[pe_r->ev_id] = pe_r->distro_size;
} else {
cerr << "missing pair end parameters:" << msg << endl;
ShowHelp();
}
// create SV_EvidenceReaders by (BAM, read_group) pairs
if (pe_r->read_group.size() == 0)
bam_evidence_readers[pair<string,string> (pe_r->get_source_file_name(),"")] = pe_r;
else {
for (vector<string>::iterator it = pe_r->read_group.begin();
it != pe_r->read_group.end();
++it) {
pair<string,string> ev_pair (pe_r->get_source_file_name(),*it);
bam_evidence_readers[ev_pair] = pe_r;
}
}
i++;
//}}}
}
else if(PARAMETER_CHECK("-bedpe", 6, parameterLength)) {
//{{{
has_bedpes = true;
SV_BedpeReader *be_r = new SV_BedpeReader();
if ((i+1) < argc) {
char *params = argv[i + 1];
char *param_val, *brka, *brkb;
for ( param_val = strtok_r(params, ",", &brka);
param_val;
param_val = strtok_r(NULL, ",", &brka)) {
char *param = strtok_r(param_val, ":", &brkb);
char *val = strtok_r(NULL, ":", &brkb);
if (val == NULL) {
cerr << "Parameter requied for " << param << endl;
ShowHelp();
}
if ( ! be_r->add_param(param, val) ) {
cerr << "Unknown bedpe parameter:" << param << endl;
ShowHelp();
}
}
}
string msg = be_r->check_params();
if ( msg.compare("") == 0 ) {
be_r->initialize();
SV_Evidence::distros[be_r->ev_id] =
pair<log_space*,log_space*>(
SV_Bedpe::
get_bp_interval_probability('+',
be_r->distro_size,
be_r->distro),
SV_Bedpe::
get_bp_interval_probability('-',
be_r->distro_size,
be_r->distro));
SV_Evidence::distros_size[be_r->ev_id] = be_r->distro_size;
evidence_readers.push_back(be_r);
} else {
cerr << "missing bedpe parameters:" << msg << endl;
ShowHelp();
}
i++;
//}}}
}
else if(PARAMETER_CHECK("-sr", 3, parameterLength)) {
//{{{
has_sr_bams = true;
SV_SplitReadReader *sr_r = new SV_SplitReadReader();
if ((i+1) < argc) {
char *params = argv[i + 1];
char *param_val, *brka, *brkb;
for ( param_val = strtok_r(params, ",", &brka);
param_val;
param_val = strtok_r(NULL, ",", &brka)) {
char *param = strtok_r(param_val, ":", &brkb);
char *val = strtok_r(NULL, ":", &brkb);
if (val == NULL) {
cerr << "Parameter required for " << param << endl;
ShowHelp();
}
if ( ! sr_r->add_param(param, val) ) {
cerr << "Unknown split read parameter:" << param << endl;
ShowHelp();
}
}
}
string msg = sr_r->check_params();
if ( msg.compare("") == 0 ) {
sr_r->initialize();
SV_Evidence::distros[sr_r->ev_id] =
pair<log_space*,log_space*>(
SV_SplitRead::
get_bp_interval_probability('+',
sr_r->back_distance),
SV_SplitRead::
get_bp_interval_probability('-',
sr_r->back_distance));
SV_Evidence::distros_size[sr_r->ev_id] =
sr_r->back_distance * 2 + 1;
} else {
cerr << "missing split read parameters:" << msg << endl;
ShowHelp();
}
// create SV_EvidenceReaders by (BAM, read_group) pairs
if (sr_r->read_group.size() == 0)
bam_evidence_readers[pair<string,string> (sr_r->get_source_file_name(),"")] = sr_r;
else {
for (vector<string>::iterator it = sr_r->read_group.begin();
it != sr_r->read_group.end();
++it) {
pair<string,string> ev_pair (sr_r->get_source_file_name(),*it);
bam_evidence_readers[ev_pair] = sr_r;
}
}
i++;
//}}}
}
else if(PARAMETER_CHECK("-tt", 3, parameterLength)) {
if ((i+1) < argc) {
trim_threshold = 1 - atof(argv[i + 1]);
i++;
}
}
else if(PARAMETER_CHECK("-mt", 3, parameterLength)) {
if ((i+1) < argc) {
merge_threshold = atof(argv[i + 1]);
i++;
}
}
else if(PARAMETER_CHECK("-mw", 3, parameterLength)) {
if ((i+1) < argc) {
min_weight = atoi(argv[i + 1]);
i++;
}
}
else if(PARAMETER_CHECK("-msw", 4, parameterLength)) {
if ((i+1) < argc) {
min_sample_weight = atoi(argv[i + 1]);
i++;
}
}
else if(PARAMETER_CHECK("-w", 2, parameterLength)) {
if ((i+1) < argc) {
window_size = atoi(argv[i + 1]);
i++;
}
}
else if(PARAMETER_CHECK("-x", 2, parameterLength)) {
if ((i+1) < argc) {
exclude_bed_file = argv[i + 1];
has_exclude = true;
i++;
}
}
else if(PARAMETER_CHECK("-g", 2, parameterLength)) {
if ((i+1) < argc) {
genome_file = argv[i + 1];
has_genome_file = true;
i++;
}
}
else if(PARAMETER_CHECK("-t", 2, parameterLength)) {
if ((i+1) < argc) {
inter_chrom_file_prefix = argv[i + 1];
i++;
}
}
else if(PARAMETER_CHECK("-e", 2, parameterLength)) {
show_evidence = true;
}
else if(PARAMETER_CHECK("-P", 2, parameterLength)) {
print_prob = 1;
}
else if(PARAMETER_CHECK("-b", 2, parameterLength)) {
bedpe_output = 1;
}
else {
cerr << endl << "*****ERROR: Unrecognized parameter: " <<
argv[i] << " *****" << endl << endl;
ShowHelp();
}
}
if (min_weight == 0 && min_sample_weight == 0) {
cerr << endl << "*****ERROR: must set min weight or min sample weight *****" <<
endl << endl;
ShowHelp();
}
SV_BreakPoint::p_trim_threshold = trim_threshold;
SV_BreakPoint::p_merge_threshold = merge_threshold;
// append rand number to the temp inter-chrom file
inter_chrom_file_prefix = inter_chrom_file_prefix + ToString(rand());
if (has_exclude)
parse_exclude_file(exclude_bed_file, SV_Evidence::exclude_regions);
SV_BamReader *bam_r;
if (has_pe_bams || has_sr_bams) {
bam_r = new SV_BamReader(&bam_evidence_readers);
bam_r->set_inter_chrom_file_name(inter_chrom_file_prefix + ".bam");
bam_r->initialize();
evidence_readers.push_back(bam_r);
}
map<string, int> genome_order;
if (has_genome_file) {
GenomeFile *genome;
genome = new GenomeFile(genome_file);
vector<string> chroms = genome->getChromList();
vector<string>::iterator chr_itr;
int chr_count = 0;
for (chr_itr = chroms.begin(); chr_itr != chroms.end(); ++chr_itr) {
genome_order[*chr_itr] = chr_count;
chr_count += 1;
}
} else if (has_pe_bams || has_sr_bams) {
//map<string, SV_EvidenceReader*> bam_evidence_readers;
//map<string, SV_EvidenceReader*>::iterator bam_itr;
//bam_itr = bam_evidence_readers.begin();
RefVector refs = bam_r->refs;
vector<RefData>::iterator ref_itr;
int chr_count = 0;
for (ref_itr = refs.begin(); ref_itr != refs.end(); ++ref_itr) {
RefData r = *ref_itr;
genome_order[r.RefName] = chr_count;
chr_count += 1;
}
} else {
cerr << endl << "*****ERROR: Unknown chromosome order. " <<
"Chromosome order must be\nspecified by either bam header " <<
"or a genome file *****" << endl << endl;
ShowHelp();
}
//}}} end parsing
//{{{ Test if there lines to process in each input file
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
has_next = has_next || er->has_next();
}
//}}}
// print VCF header
if (! bedpe_output) {
SV_VcfVariant *header_var = new SV_VcfVariant();
map<int,string>::iterator s_itr;
for (s_itr = SV_EvidenceReader::sample_names.begin();
s_itr != SV_EvidenceReader::sample_names.end();
++s_itr)
header_var->add_sample(s_itr->second);
// add appropriate fields to active_formats
header_var->set_sample_field(SV_EvidenceReader::
sample_names.begin()->second,
"GT",
"./.");
header_var->set_sample_field(SV_EvidenceReader::
sample_names.begin()->second,
"SU",
"0");
if (has_pe_bams)
header_var->set_sample_field(SV_EvidenceReader::
sample_names.begin()->second,
"PE",
"0");
if (has_sr_bams)
header_var->set_sample_field(SV_EvidenceReader::
sample_names.begin()->second,
"SR",
"0");
if (has_bedpes)
header_var->set_sample_field(SV_EvidenceReader::
sample_names.begin()->second,
"BD",
"0");
header_var->print_header();
delete(header_var);
}
//{{{ process the intra-chrom events that were saved to a file
CHR_POS max_pos = 0;
string last_min_chr = "";
while ( has_next ) {
string min_chr = "";
//{{{ find min_chr among all input files
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
if ( er->has_next() ) {
string curr_chr = er->get_curr_chr();
if ( ( min_chr.compare("") == 0 ) ||
( genome_order[curr_chr] < genome_order[min_chr] ) ) {
min_chr = curr_chr;
}
}
}
//}}}
//{{{ if the chrome switches, reset the max_pos
if (last_min_chr.compare(min_chr) != 0) {
max_pos = window_size;
last_min_chr = min_chr;
}
//}}}
cerr << min_chr << "\t" << max_pos << endl;
bool input_processed = true;
while (input_processed) {
input_processed = false;
//{{{ read the files
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
if ( er->has_next() ) {
string curr_chr = er->get_curr_chr();
CHR_POS curr_pos = er->get_curr_pos();
if ( ( genome_order[curr_chr] <= genome_order[min_chr] ) &&
( curr_pos < max_pos) ) {
er->process_input_chr_pos(curr_chr, max_pos, r_bin);
input_processed = true;
}
}
}
//}}}
//{{{ call breakpoints
vector< UCSCElement<SV_BreakPoint*> > values =
r_bin.values(min_chr, max_pos);
vector< UCSCElement<SV_BreakPoint*> >::iterator it;
for (it = values.begin(); it < values.end(); ++it) {
SV_BreakPoint *bp = it->value;
// Make sure both ends of the bp are less than or equal to the
// current chrom
if (bp->weight >= min_weight
&& bp->get_max_sample_weight() >= min_sample_weight) {
//bp->do_it();
// make sure there was not an error with trimming
if (bp->trim_intervals() > 0) {
if (bedpe_output) {
bp->print_bedpe(++call_id, print_prob);
if (show_evidence)
bp->print_evidence("\t");
}
else {
SV_VcfVariant *vcf_var =
new SV_VcfVariant(bp,
++call_id,
print_prob);
vcf_var->print_var();
delete(vcf_var);
}
}
}
if (r_bin.remove(*it, false, false, true) != 0) {
cerr << "Error removing element:" << *bp << endl;
abort();
}
bp->free_evidence();
delete bp;
}
//}}}
// move the window
max_pos = max_pos *2;
}
//{{{ Test if there is still input lines
has_next = false;
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
has_next = has_next || er->has_next();
}
//}}}
}
//}}}
//{{{ terminate input files
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
er->terminate();
}
//}}}
//{{{ Call remaining intra breakpoints
vector< UCSCElement<SV_BreakPoint*> > values = r_bin.values();
vector< UCSCElement<SV_BreakPoint*> >::iterator it;
for ( it = values.begin();
it != values.end(); ++it) {
SV_BreakPoint *bp = it->value;
if (bp->weight >= min_weight
&& bp->get_max_sample_weight() >= min_sample_weight) {
//bp->do_it();
if (bp->trim_intervals() > 0) {
if (bedpe_output) {
bp->print_bedpe(++call_id, print_prob);
if (show_evidence)
bp->print_evidence("\t");
}
else {
SV_VcfVariant *vcf_var =
new SV_VcfVariant(bp,
++call_id,
print_prob);
vcf_var->print_var();
}
}
}
if (r_bin.remove(*it, false, false, true) != 0) {
cerr << "Error removing element" << endl;
abort();
}
bp->free_evidence();
delete bp;
}
//}}}
//{{{ process the inter-chrom events that were saved to a file
string intra_bam_file_name = inter_chrom_file_prefix + ".bam";
ifstream intra_bam_file( intra_bam_file_name.c_str() );
if (intra_bam_file.good()) {
intra_bam_file.close();
sort_inter_chrom_bam( inter_chrom_file_prefix + ".bam",
inter_chrom_file_prefix + ".sort.bam");
SV_InterChromBamReader *ic_r = new SV_InterChromBamReader(
inter_chrom_file_prefix + ".sort.bam",
&bam_evidence_readers);
ic_r->initialize();
vector<SV_EvidenceReader*> inter_chrom_evidence_readers;
inter_chrom_evidence_readers.push_back(ic_r);
// There are two files containg all of the inter-chrom events, one bam
// and one bedpe, each line in the file corresponds to the properies
// set in one of the readers. Each line has a "LS" (lumpy source)
// property that gives its source file name. Using that entry, the
// line will be sent to the reader for processing.
// get new evidence readers for both bedpe and bam inter-chrom
// readers
has_next = true;
int32_t last_min_primary_refid = -1;
int32_t last_min_secondary_refid = -1;
max_pos = 0;
while ( has_next ) {
string min_primary_chr = "";
string min_secondary_chr = "";
int32_t min_primary_refid = -1;
int32_t min_secondary_refid = -1;
//{{{ find min_chr pair among all input files
for ( i_er = inter_chrom_evidence_readers.begin();
i_er != inter_chrom_evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
if ( er->has_next() ) {
int32_t curr_primary_refid = er->get_curr_primary_refid();
int32_t curr_secondary_refid =
er->get_curr_secondary_refid();
if ( (( min_primary_refid == -1 ) &&
( min_secondary_refid == -1 )) ||
(( curr_primary_refid < min_primary_refid) &&
( curr_secondary_refid < min_secondary_refid)) ) {
min_primary_refid = curr_primary_refid;
min_secondary_refid = curr_secondary_refid;
min_secondary_chr = er->get_curr_secondary_chr();
min_primary_chr = er->get_curr_primary_chr();
}
}
}
//}}}
// if the chrome pair switches, reset the max_pos
if ( (last_min_primary_refid != min_primary_refid) ||
(last_min_secondary_refid != min_secondary_refid) ) {
max_pos = window_size;
last_min_primary_refid = min_primary_refid;
last_min_secondary_refid = min_secondary_refid;
}
bool input_processed = true;
while (input_processed) {
input_processed = false;
//{{{ read the files, process anything in frame
for ( i_er = inter_chrom_evidence_readers.begin();
i_er != inter_chrom_evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
if ( er->has_next() ) {
int32_t curr_primary_refid =
er->get_curr_primary_refid();
int32_t curr_secondary_refid =
er->get_curr_secondary_refid();
CHR_POS curr_pos = er->get_curr_primary_pos();
if ( (curr_primary_refid <= min_primary_refid) &&
(curr_secondary_refid <= min_secondary_refid) &&
(curr_pos < max_pos) ) {
er->process_input_chr_pos(
er->get_curr_primary_chr(),
er->get_curr_secondary_chr(),
max_pos,
r_bin);
input_processed = true;
}
}
}
//}}}
//{{{ get breakpoints
// get anything that has ends in both chroms
vector< UCSCElement<SV_BreakPoint*> > values =
r_bin.values(min_secondary_chr);
vector< UCSCElement<SV_BreakPoint*> >::iterator it;
for (it = values.begin(); it < values.end(); ++it) {
SV_BreakPoint *bp = it->value;
if (bp->weight >= min_weight
&& bp->get_max_sample_weight() >= min_sample_weight) {
//bp->do_it();
if (bp->trim_intervals() > 0) {
if (bedpe_output) {
bp->print_bedpe(++call_id, print_prob);
if (show_evidence)
bp->print_evidence("\t");
}
else {
SV_VcfVariant *vcf_var =
new SV_VcfVariant(bp,
++call_id,
print_prob);
vcf_var->print_var();
}
}
}
if (r_bin.remove(*it, false, false, true) != 0) {
cerr << "Error removing element" << endl;
abort();
}
bp->free_evidence();
delete bp;
}
//}}}
max_pos = max_pos * 2;
}
has_next = false;
//{{{ Test if there is still input lines
for ( i_er = inter_chrom_evidence_readers.begin();
i_er != inter_chrom_evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
has_next = has_next || er->has_next();
}
//}}}
}
//{{{ Call remaining break points
values = r_bin.values();
for (it = values.begin(); it != values.end(); ++it) {
SV_BreakPoint *bp = it->value;
if (bp->weight >= min_weight
&& bp->get_max_sample_weight() >= min_sample_weight) {
//bp->do_it();
if (bp->trim_intervals() > 0) {
if (bedpe_output) {
bp->print_bedpe(++call_id, print_prob);
if (show_evidence)
bp->print_evidence("\t");
}
else {
SV_VcfVariant *vcf_var =
new SV_VcfVariant(bp,
++call_id,
print_prob);
vcf_var->print_var();
}
}
}
if (r_bin.remove(*it, false, false, true) != 0) {
cerr << "Error removing element" << endl;
abort();
}
bp->free_evidence();
delete bp;
}
//}}}
for ( i_er = inter_chrom_evidence_readers.begin();
i_er != inter_chrom_evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
delete(er);
}
}
//}}}
//{{{ free up stuff
string s = inter_chrom_file_prefix + ".bam";
remove(s.c_str());
s = inter_chrom_file_prefix + ".sort.bam";
remove(s.c_str());
map<int, pair<log_space*,log_space*> >::iterator e_it;
for( e_it = SV_Evidence::distros.begin();
e_it != SV_Evidence::distros.end();
++e_it) {
free(e_it->second.first);
free(e_it->second.second);
}
#if 0
for ( i_er = evidence_readers.begin();
i_er != evidence_readers.end();
++i_er) {
SV_EvidenceReader *er = *i_er;
delete(er);
}
#endif
evidence_readers.clear();
bam_evidence_readers.clear();
//}}}
return 0;
}
// Same as ParseRegionString() above, but accepts a BamMultiReader
bool ParseRegionString(const string& regionString,
const BamReader& reader,
BamRegion& region)
{
// -------------------------------
// parse region string
// check first for empty string
if ( regionString.empty() )
return false;
//cerr << "ParseRegionString Input: " << regionString << endl;
// non-empty string, look for a colom
size_t foundFirstColon = regionString.find(':');
// store chrom strings, and numeric positions
string startChrom;
string stopChrom;
int startPos;
int stopPos;
// no colon found
// going to use entire contents of requested chromosome
// just store entire region string as startChrom name
// use BamReader methods to check if its valid for current BAM file
if ( foundFirstColon == string::npos ) {
startChrom = regionString;
startPos = 0;
stopChrom = regionString;
stopPos = -1;
}
// colon found, so we at least have some sort of startPos requested
else {
// store start chrom from beginning to first colon
startChrom = regionString.substr(0,foundFirstColon);
// look for ".." after the colon
size_t foundRangeDots = regionString.find("..", foundFirstColon+1);
// no dots found
// so we have a startPos but no range
// store contents before colon as startChrom, after as startPos
if ( foundRangeDots == string::npos )
{
startPos = atoi( regionString.substr(foundFirstColon+1).c_str() );
stopChrom = startChrom;
stopPos = -1;
}
// ".." found, so we have some sort of range selected
else {
// store startPos between first colon and range dots ".."
startPos = atoi( regionString.substr(foundFirstColon+1, foundRangeDots-foundFirstColon-1).c_str() );
// look for second colon
size_t foundSecondColon = regionString.find(':', foundRangeDots+1);
// no second colon found
// so we have a "standard" chrom:start..stop input format (on single chrom)
if ( foundSecondColon == string::npos ) {
stopChrom = startChrom;
stopPos = atoi( regionString.substr(foundRangeDots+2).c_str() );
}
// second colon found
// so we have a range requested across 2 chrom's
else {
stopChrom = regionString.substr(foundRangeDots+2, foundSecondColon-(foundRangeDots+2));
stopPos = atoi( regionString.substr(foundSecondColon+1).c_str() );
}
}
}
// -------------------------------
// validate reference IDs & genomic positions
const RefVector references = reader.GetReferenceData();
// if startRefID not found, return false
int startRefID = reader.GetReferenceID(startChrom);
if ( startRefID == -1 ) return false;
// startPos cannot be greater than or equal to reference length
const RefData& startReference = references.at(startRefID);
if ( startPos >= startReference.RefLength ) return false;
// if stopRefID not found, return false
int stopRefID = reader.GetReferenceID(stopChrom);
if ( stopRefID == -1 ) return false;
// stopPosition cannot be larger than reference length
const RefData& stopReference = references.at(stopRefID);
if ( stopPos > stopReference.RefLength ) return false;
// if no stopPosition specified, set to reference end
if ( stopPos == -1 ) stopPos = stopReference.RefLength;
// -------------------------------
// set up Region struct & return
region.LeftRefID = startRefID;
region.LeftPosition = startPos;
region.RightRefID = stopRefID;;
region.RightPosition = stopPos;
//cerr << "ParseRegionString " << region.LeftRefID << " " << region.LeftPosition << " " << region.RightPosition << endl;
return true;
}
int main ( int argc, char *argv[] ) {
struct parameters *param = 0;
param = interface(param, argc, argv);
//region file input (the region file should be sorted as the same way as the bam file)
ifstream region_f;
region_f.open(param->region_f, ios_base::in); // the region file is opened
//bam input and generate index if not yet
//-------------------------------------------------------------------------------------------------------+
// BAM input (file or filenames?) |
//-------------------------------------------------------------------------------------------------------+
char *fof = param->mapping_f;
FILE *IN=NULL;
char linefof[5000];
int filecount=0;
vector <string> fnames;
if (strchr(fof,' ')!=NULL) {
char *ptr;
ptr=strtok(fof," ");
while (ptr!=NULL) {
fnames.push_back(ptr);
filecount++;
ptr=strtok(NULL," ");
}
} else {
IN=fopen(fof,"rt");
if (IN!=NULL) {
long linecount=0;
while (fgets(linefof,5000-1,IN)!=NULL) {
linecount++;
if (linefof[0]!='#' && linefof[0]!='\n') {
char *ptr=strchr(linefof,'\n');
if (ptr!=NULL && ptr[0]=='\n') {
ptr[0]='\0';
}
FILE *dummy=NULL;
dummy=fopen(linefof,"rt");
if (dummy!=NULL) { // seems to be a file of filenames...
fclose(dummy);
fnames.push_back(linefof);
filecount++;
} else if (filecount==0 || linecount>=1000-1) { // seems to be a single file
fnames.push_back(fof);
filecount++;
break;
}
}
}
fclose(IN);
}
} //file or file name decided and stored in vector "fnames"
cerr << "the input mapping files are:" << endl;
vector <string>::iterator fit = fnames.begin();
for(; fit != fnames.end(); fit++) {
cerr << *fit << endl;
}
//-------------------------------------------------------------------------------------------------------+
// end of file or filenames |
//-------------------------------------------------------------------------------------------------------+
// open the BAM file(s)
BamMultiReader reader;
reader.Open(fnames);
// get header & reference information
string header = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
if ( ! reader.LocateIndexes() ) // opens any existing index files that match our BAM files
reader.CreateIndexes(); // creates index files for BAM files that still lack one
// locus bias
struct lb empty_profile = {0,0,0,0};
vector <struct lb> locus_b(1000, empty_profile);
// output locus bias file
string locus_bias_set = param->lbias;
ofstream locus_bias;
if ( locus_bias_set != "" ) {
locus_bias.open(param->lbias);
if ( !locus_bias ) {
cerr << "can not open locus_bias file.\n";
exit(0);
}
}
//should decide which chromosome
string line;
string old_chr = "SRP";
string type = param->type;
//whether do some position-level pile-up stuff
bool posc = false;
ofstream posc_f;
ofstream chrmap_f;
string poscset = param->posc;
if ( poscset != "" ) {
posc = true;
posc_f.open(param->posc);
chrmap_f.open(param->chrmap);
}
bool noChr;
if ( param->nochr == 1 ){
noChr = true;
} else {
noChr = false;
}
//regions for the input of region file
deque <struct region> regions;
getline(region_f, line); //get the first line
eatline(line,regions,noChr);
deque <struct region>::iterator it = regions.begin();
while ( it->chr != old_chr ) {
old_chr = it->chr; // set the current chr as old chr
int chr_id = reader.GetReferenceID(it->chr);
if ( chr_id == -1 ) { //reference not found
for (; it != regions.end() && it->chr == old_chr; ) {
gene_processing(*it,locus_b); // print the old region info
it = regions.erase(it); // erase the current region
}
while ( regions.empty() ) {
getline(region_f, line);
if ( region_f.eof() ){
cerr << "finished: end of region file, zone 0" << endl;
break;
}
eatline(line, regions,noChr);
it = regions.begin();
if (it->chr == old_chr){
gene_processing(*it,locus_b);
regions.clear();
continue;
}
}
continue;
}
int chr_len = refs.at(chr_id).RefLength;
if ( !reader.SetRegion(chr_id, 1, chr_id, chr_len) ) // here set region
{
cerr << "bamtools count ERROR: Jump region failed " << it->chr << endl;
reader.Close();
exit(1);
}
//pile-up pos stats
set <string> fragment;
map <string, unsigned int> pileup;
bool isposPileup = false;
unsigned int old_start = 0;
unsigned int total_tags = 0;
unsigned int total_pos = 0;
unsigned int pileup_pos = 0;
BamAlignment bam;
while (reader.GetNextAlignment(bam)) {
if ( bam.IsMapped() == false ) continue; // skip unaligned reads
unsigned int unique;
bam.GetTag("NH", unique);
if (param->unique == 1) {
if (unique != 1) { // skipe uniquelly mapped reads
continue;
}
}
if (read_length == 0){
read_length = bam.Length;
}
//cout << bam.Name << endl;
string chrom = refs.at(bam.RefID).RefName;
string strand = "+";
if (bam.IsReverseStrand()) strand = "-";
unsigned int alignmentStart = bam.Position+1;
unsigned int mateStart;
if (type == "p") mateStart = bam.MatePosition+1;
unsigned int alignmentEnd = bam.GetEndPosition();
unsigned int cigarEnd;
vector <int> blockLengths;
vector <int> blockStarts;
blockStarts.push_back(0);
ParseCigar(bam.CigarData, blockStarts, blockLengths, cigarEnd);
// position check for unique mapped reads (because is paired-end reads, shoule base on fragment level for paired end reads)
if (posc == true && unique == 1) {
if (type == "p" && fragment.count(bam.Name) > 0)
fragment.erase(bam.Name);
else {
total_tags++;
if (type == "p"){
fragment.insert(bam.Name);
}
string alignSum;
if (type == "p") {
alignSum = int2str(alignmentStart) + "\t" + int2str(mateStart) + "\t.\t" + strand;
} else {
alignSum = int2str(alignmentStart) + "\t" + int2str(alignmentEnd) + "\t.\t" + strand;
}
if ( alignmentStart != old_start ) {
isposPileup = false;
map <string, unsigned int>::iterator pit = pileup.begin();
for (; pit != pileup.end(); pit++) {
posc_f << chrom << "\truping\tpileup\t" << pit->first << "\t.\t" << "Pileup=" << pit->second << endl; //print pileup
}
pileup.clear(); //clear pileup set
pileup.insert( pair <string, unsigned int> (alignSum, 1) ); //insert the new read
total_pos++;
}
else if ( alignmentStart == old_start ) { // same starts
if ( pileup.count(alignSum) > 0 ) { // pileup
if ( pileup[alignSum] == 1 && isposPileup == false ) {
pileup_pos++; isposPileup = true;
}
pileup[alignSum]++;
}
else {
pileup.insert( pair <string, unsigned int> (alignSum, 1) );
}
} //same starts
} //new fragment
old_start = alignmentStart;
} // do pos check
float incre = 1.;
if (blockStarts.size() > 1) incre = 0.5; // incre half for junction reads
incre /= static_cast < float >(unique); // for multi aligned reads
deque <struct region>::iterator iter = regions.begin();
if ( iter->start > alignmentEnd ) continue; // skip reads not overlapping with the first region
while ( iter->chr == old_chr && iter->start <= alignmentEnd && iter != regions.end() ) {
if (iter->end < alignmentStart) { // the region end is beyond the alignmentStart
gene_processing(*iter,locus_b); // processing
iter = regions.erase(iter); // this region should be removed
if ( regions.empty() ) {
getline(region_f, line); // get a line of region file
if ( ! region_f.eof() ) {
eatline(line, regions, noChr); // eat a line and put it into the duque
iter = regions.begin();
}
else { // it's reaching the end of the region file
cerr << "finished: end of region file, zone 3" << endl;
break;
}
}
continue;
}
if (iter->end >= alignmentStart && iter->start <= alignmentEnd) { //overlapping, should take action
vector <int>::iterator cigit = blockStarts.begin();
for (; cigit != blockStarts.end(); cigit++) {
unsigned int current_start = *cigit + alignmentStart;
int current_pos = current_start - (iter->start);
//cout << iter->chr << "\t" << iter->start << "\t" << iter->end << "\t" << current_start << endl;
if ( (iter->tags).count(current_pos) > 0 ) {
(iter->tags)[current_pos] += incre;
}
else
(iter->tags).insert( pair<int, float>(current_pos, incre) );
}
} // overlapping take action!
if ( (iter+1) != regions.end() )
iter++; // if this region is not the last element in the deque
else { // the last element
getline(region_f, line); // get a line of region file
if ( ! region_f.eof() ){
eatline(line, regions, noChr); // eat a line and put it into the duque
iter = regions.end();
iter--;
}
else { //it's reaching the end of the region file
cerr << "finished: end of region file, zone 4" << endl;
break;
}
}
} //while
} // read a bam
// print chr map
if (posc == true) {
chrmap_f << old_chr << "\t" << total_tags << "\t" << total_pos << "\t" << pileup_pos << endl;
}
//somehow to loop back
it = regions.begin(); //reset to begin
for (; it != regions.end() && it->chr == old_chr; ) {
gene_processing(*it,locus_b); // print the old region info
it = regions.erase(it); // erase the current region
}
while ( regions.empty() ) {
getline(region_f, line);
if ( region_f.eof() ){
cerr << "finished: end of region file, zone 5" << endl;
//print locus bias
for (unsigned int l = 0; l < 1000; l++){
locus_bias << l << "\t" << locus_b[l].ps << "\t" << locus_b[l].hs << "\t" << locus_b[l].pe << "\t" << locus_b[l].he << endl;
}
exit(0);
}
eatline(line, regions, noChr);
it = regions.begin();
if (it->chr == old_chr){
gene_processing(*it, locus_b);
regions.clear();
continue;
}
}
} // region chr != old chr
regions.clear();
reader.Close();
region_f.close();
return 0;
} //main
void BedIntersect::IntersectBam(string bamFile) {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bedB->loadBedFileIntoMap();
// open the BAM file
BamReader reader;
BamWriter writer;
reader.Open(bamFile);
// get header & reference information
string header = reader.GetHeaderText();
RefVector refs = reader.GetReferenceData();
// open a BAM output to stdout if we are writing BAM
if (_bamOutput == true) {
// open our BAM writer
writer.Open("stdout", header, refs, _isUncompressedBam);
}
vector<BED> hits;
// reserve some space
hits.reserve(100);
_bedA->bedType = 6;
BamAlignment bam;
// get each set of alignments for each pair.
while (reader.GetNextAlignment(bam)) {
if (bam.IsMapped()) {
BED a;
a.chrom = refs.at(bam.RefID).RefName;
a.start = bam.Position;
a.end = bam.GetEndPosition(false);
// build the name field from the BAM alignment.
a.name = bam.Name;
if (bam.IsFirstMate()) a.name += "/1";
if (bam.IsSecondMate()) a.name += "/2";
a.score = ToString(bam.MapQuality);
a.strand = "+";
if (bam.IsReverseStrand()) a.strand = "-";
if (_bamOutput == true) {
bool overlapsFound = false;
// treat the BAM alignment as a single "block"
if (_obeySplits == false) {
overlapsFound = FindOneOrMoreOverlap(a);
}
// split the BAM alignment into discrete blocks and
// look for overlaps only within each block.
else {
bool overlapFoundForBlock;
bedVector bedBlocks; // vec to store the discrete BED "blocks" from a
// we don't want to split on "D" ops, hence the "false"
getBamBlocks(bam, refs, bedBlocks, false);
vector<BED>::const_iterator bedItr = bedBlocks.begin();
vector<BED>::const_iterator bedEnd = bedBlocks.end();
for (; bedItr != bedEnd; ++bedItr) {
overlapFoundForBlock = FindOneOrMoreOverlap(a);
if (overlapFoundForBlock == true)
overlapsFound = true;
}
}
if (overlapsFound == true) {
if (_noHit == false)
writer.SaveAlignment(bam);
}
else {
if (_noHit == true) {
writer.SaveAlignment(bam);
}
}
}
else {
// treat the BAM alignment as a single BED "block"
if (_obeySplits == false) {
FindOverlaps(a, hits);
hits.clear();
}
// split the BAM alignment into discrete BED blocks and
// look for overlaps only within each block.
else {
bedVector bedBlocks; // vec to store the discrete BED "blocks" from a
getBamBlocks(bam, refs, bedBlocks, false);
vector<BED>::const_iterator bedItr = bedBlocks.begin();
vector<BED>::const_iterator bedEnd = bedBlocks.end();
for (; bedItr != bedEnd; ++bedItr) {
FindOverlaps(*bedItr, hits);
hits.clear();
}
}
}
}
}
// close the relevant BAM files.
reader.Close();
if (_bamOutput == true) {
writer.Close();
}
}
namespace BamTools {
// -------------------------------
// string literal constants
// property names
const string ALIGNMENTFLAG_PROPERTY = "alignmentFlag";
const string CIGAR_PROPERTY = "cigar";
const string INSERTSIZE_PROPERTY = "insertSize";
const string ISDUPLICATE_PROPERTY = "isDuplicate";
const string ISFAILEDQC_PROPERTY = "isFailedQC";
const string ISFIRSTMATE_PROPERTY = "isFirstMate";
const string ISMAPPED_PROPERTY = "isMapped";
const string ISMATEMAPPED_PROPERTY = "isMateMapped";
const string ISMATEREVERSESTRAND_PROPERTY = "isMateReverseStrand";
const string ISPAIRED_PROPERTY = "isPaired";
const string ISPRIMARYALIGNMENT_PROPERTY = "isPrimaryAlignment";
const string ISPROPERPAIR_PROPERTY = "isProperPair";
const string ISREVERSESTRAND_PROPERTY = "isReverseStrand";
const string ISSECONDMATE_PROPERTY = "isSecondMate";
const string ISSINGLETON_PROPERTY = "isSingleton";
const string MAPQUALITY_PROPERTY = "mapQuality";
const string MATEPOSITION_PROPERTY = "matePosition";
const string MATEREFERENCE_PROPERTY = "mateReference";
const string NAME_PROPERTY = "name";
const string POSITION_PROPERTY = "position";
const string QUERYBASES_PROPERTY = "queryBases";
const string REFERENCE_PROPERTY = "reference";
const string TAG_PROPERTY = "tag";
// boolalpha
const string TRUE_STR = "true";
const string FALSE_STR = "false";
RefVector filterToolReferences;
struct BamAlignmentChecker {
bool check(const PropertyFilter& filter, const BamAlignment& al) {
bool keepAlignment = true;
const PropertyMap& properties = filter.Properties;
PropertyMap::const_iterator propertyIter = properties.begin();
PropertyMap::const_iterator propertyEnd = properties.end();
for ( ; propertyIter != propertyEnd; ++propertyIter ) {
// check alignment data field depending on propertyName
const string& propertyName = (*propertyIter).first;
const PropertyFilterValue& valueFilter = (*propertyIter).second;
if ( propertyName == ALIGNMENTFLAG_PROPERTY ) keepAlignment &= valueFilter.check(al.AlignmentFlag);
else if ( propertyName == CIGAR_PROPERTY ) {
stringstream cigarSs;
const vector<CigarOp>& cigarData = al.CigarData;
if ( !cigarData.empty() ) {
vector<CigarOp>::const_iterator cigarBegin = cigarData.begin();
vector<CigarOp>::const_iterator cigarIter = cigarBegin;
vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
for ( ; cigarIter != cigarEnd; ++cigarIter ) {
const CigarOp& op = (*cigarIter);
cigarSs << op.Length << op.Type;
}
keepAlignment &= valueFilter.check(cigarSs.str());
}
}
else if ( propertyName == INSERTSIZE_PROPERTY ) keepAlignment &= valueFilter.check(al.InsertSize);
else if ( propertyName == ISDUPLICATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsDuplicate());
else if ( propertyName == ISFAILEDQC_PROPERTY ) keepAlignment &= valueFilter.check(al.IsFailedQC());
else if ( propertyName == ISFIRSTMATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsFirstMate());
else if ( propertyName == ISMAPPED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMapped());
else if ( propertyName == ISMATEMAPPED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMateMapped());
else if ( propertyName == ISMATEREVERSESTRAND_PROPERTY ) keepAlignment &= valueFilter.check(al.IsMateReverseStrand());
else if ( propertyName == ISPAIRED_PROPERTY ) keepAlignment &= valueFilter.check(al.IsPaired());
else if ( propertyName == ISPRIMARYALIGNMENT_PROPERTY ) keepAlignment &= valueFilter.check(al.IsPrimaryAlignment());
else if ( propertyName == ISPROPERPAIR_PROPERTY ) keepAlignment &= valueFilter.check(al.IsProperPair());
else if ( propertyName == ISREVERSESTRAND_PROPERTY ) keepAlignment &= valueFilter.check(al.IsReverseStrand());
else if ( propertyName == ISSECONDMATE_PROPERTY ) keepAlignment &= valueFilter.check(al.IsSecondMate());
else if ( propertyName == ISSINGLETON_PROPERTY ) {
const bool isSingleton = al.IsPaired() && al.IsMapped() && !al.IsMateMapped();
keepAlignment &= valueFilter.check(isSingleton);
}
else if ( propertyName == MAPQUALITY_PROPERTY ) keepAlignment &= valueFilter.check(al.MapQuality);
else if ( propertyName == MATEPOSITION_PROPERTY ) keepAlignment &= ( al.IsPaired() && al.IsMateMapped() && valueFilter.check(al.MateRefID) );
else if ( propertyName == MATEREFERENCE_PROPERTY ) {
if ( !al.IsPaired() || !al.IsMateMapped() ) return false;
BAMTOOLS_ASSERT_MESSAGE( (al.MateRefID>=0 && (al.MateRefID<(int)filterToolReferences.size())), "Invalid MateRefID");
const string& refName = filterToolReferences.at(al.MateRefID).RefName;
keepAlignment &= valueFilter.check(refName);
}
else if ( propertyName == NAME_PROPERTY ) keepAlignment &= valueFilter.check(al.Name);
else if ( propertyName == POSITION_PROPERTY ) keepAlignment &= valueFilter.check(al.Position);
else if ( propertyName == QUERYBASES_PROPERTY ) keepAlignment &= valueFilter.check(al.QueryBases);
else if ( propertyName == REFERENCE_PROPERTY ) {
BAMTOOLS_ASSERT_MESSAGE( (al.RefID>=0 && (al.RefID<(int)filterToolReferences.size())), "Invalid RefID");
const string& refName = filterToolReferences.at(al.RefID).RefName;
keepAlignment &= valueFilter.check(refName);
}
else if ( propertyName == TAG_PROPERTY ) keepAlignment &= checkAlignmentTag(valueFilter, al);
else BAMTOOLS_ASSERT_UNREACHABLE;
// if alignment fails at ANY point, just quit and return false
if ( !keepAlignment ) return false;
}
BAMTOOLS_ASSERT_MESSAGE( keepAlignment, "Error in BamAlignmentChecker... keepAlignment should be true here");
return keepAlignment;
}
bool checkAlignmentTag(const PropertyFilterValue& valueFilter, const BamAlignment& al) {
// ensure filter contains string data
Variant entireTagFilter = valueFilter.Value;
if ( !entireTagFilter.is_type<string>() ) return false;
// localize string from variant
const string& entireTagFilterString = entireTagFilter.get<string>();
// ensure we have at least "XX:x"
if ( entireTagFilterString.length() < 4 ) return false;
// get tagName & lookup in alignment
// if found, set tagType to tag type character
// if not found, return false
const string& tagName = entireTagFilterString.substr(0,2);
char tagType = '\0';
if ( !al.GetTagType(tagName, tagType) ) return false;
// remove tagName & ":" from beginning tagFilter
string tagFilterString = entireTagFilterString.substr(3);
// switch on tag type to set tag query value & parse filter token
int32_t intFilterValue, intQueryValue;
uint32_t uintFilterValue, uintQueryValue;
float realFilterValue, realQueryValue;
string stringFilterValue, stringQueryValue;
PropertyFilterValue tagFilter;
PropertyFilterValue::ValueCompareType compareType;
bool keepAlignment = false;
switch (tagType) {
// signed int tag type
case 'c' :
case 's' :
case 'i' :
if ( al.GetTag(tagName, intQueryValue) ) {
if ( FilterEngine<BamAlignmentChecker>::parseToken(tagFilterString, intFilterValue, compareType) ) {
tagFilter.Value = intFilterValue;
tagFilter.Type = compareType;
keepAlignment = tagFilter.check(intQueryValue);
}
}
break;
// unsigned int tag type
case 'C' :
case 'S' :
case 'I' :
if ( al.GetTag(tagName, uintQueryValue) ) {
if ( FilterEngine<BamAlignmentChecker>::parseToken(tagFilterString, uintFilterValue, compareType) ) {
tagFilter.Value = uintFilterValue;
tagFilter.Type = compareType;
keepAlignment = tagFilter.check(uintQueryValue);
}
}
break;
// 'real' tag type
case 'f' :
if ( al.GetTag(tagName, realQueryValue) ) {
if ( FilterEngine<BamAlignmentChecker>::parseToken(tagFilterString, realFilterValue, compareType) ) {
tagFilter.Value = realFilterValue;
tagFilter.Type = compareType;
keepAlignment = tagFilter.check(realQueryValue);
}
}
break;
// string tag type
case 'A':
case 'Z':
case 'H':
if ( al.GetTag(tagName, stringQueryValue) ) {
if ( FilterEngine<BamAlignmentChecker>::parseToken(tagFilterString, stringFilterValue, compareType) ) {
tagFilter.Value = stringFilterValue;
tagFilter.Type = compareType;
keepAlignment = tagFilter.check(stringQueryValue);
}
}
break;
// unknown tag type
default :
keepAlignment = false;
}
return keepAlignment;
}
};
} // namespace BamTools
bool RandomTool::RandomToolPrivate::Run(void) {
// set to default stdin if no input files provided
if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
// add files in the filelist to the input file list
if ( m_settings->HasInputFilelist ) {
ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
if ( !filelist.is_open() ) {
cerr << "bamtools random ERROR: could not open input BAM file list... Aborting." << endl;
return false;
}
string line;
while ( getline(filelist, line) )
m_settings->InputFiles.push_back(line);
}
// open our reader
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
cerr << "bamtools random ERROR: could not open input BAM file(s)... Aborting." << endl;
return false;
}
// look up index files for all BAM files
reader.LocateIndexes();
// make sure index data is available
if ( !reader.HasIndexes() ) {
cerr << "bamtools random ERROR: could not load index data for all input BAM file(s)... Aborting." << endl;
reader.Close();
return false;
}
// get BamReader metadata
const string headerText = reader.GetHeaderText();
const RefVector references = reader.GetReferenceData();
if ( references.empty() ) {
cerr << "bamtools random ERROR: no reference data available... Aborting." << endl;
reader.Close();
return false;
}
// determine compression mode for BamWriter
bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() &&
!m_settings->IsForceCompression );
BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
if ( writeUncompressed ) compressionMode = BamWriter::Uncompressed;
// open BamWriter
BamWriter writer;
writer.SetCompressionMode(compressionMode);
if ( !writer.Open(m_settings->OutputFilename, headerText, references) ) {
cerr << "bamtools random ERROR: could not open " << m_settings->OutputFilename
<< " for writing... Aborting." << endl;
reader.Close();
return false;
}
// if user specified a REGION constraint, attempt to parse REGION string
BamRegion region;
if ( m_settings->HasRegion && !Utilities::ParseRegionString(m_settings->Region, reader, region) ) {
cerr << "bamtools random ERROR: could not parse REGION: " << m_settings->Region << endl;
cerr << "Check that REGION is in valid format (see documentation) and that the coordinates are valid"
<< endl;
reader.Close();
writer.Close();
return false;
}
// seed our random number generator
srand( time(NULL) );
// grab random alignments
BamAlignment al;
unsigned int i = 0;
while ( i < m_settings->AlignmentCount ) {
int randomRefId = 0;
int randomPosition = 0;
// use REGION constraints to select random refId & position
if ( m_settings->HasRegion ) {
// select a random refId
randomRefId = getRandomInt(region.LeftRefID, region.RightRefID);
// select a random position based on randomRefId
const int lowerBoundPosition = ( (randomRefId == region.LeftRefID)
? region.LeftPosition
: 0 );
const int upperBoundPosition = ( (randomRefId == region.RightRefID)
? region.RightPosition
: (references.at(randomRefId).RefLength - 1) );
randomPosition = getRandomInt(lowerBoundPosition, upperBoundPosition);
}
// otherwise select from all possible random refId & position
else {
// select random refId
randomRefId = getRandomInt(0, (int)references.size() - 1);
// select random position based on randomRefId
const int lowerBoundPosition = 0;
const int upperBoundPosition = references.at(randomRefId).RefLength - 1;
randomPosition = getRandomInt(lowerBoundPosition, upperBoundPosition);
}
// if jump & read successful, save first alignment that overlaps random refId & position
if ( reader.Jump(randomRefId, randomPosition) ) {
while ( reader.GetNextAlignmentCore(al) ) {
if ( al.RefID == randomRefId && al.Position >= randomPosition ) {
writer.SaveAlignment(al);
++i;
break;
}
}
}
}
// cleanup & exit
reader.Close();
writer.Close();
return true;
}
/**
* Construct a Vector using a reference Vector. This will be used to do
* implicit casts from RefVector to Vector for most of the Vector operations.
*
* @param vec the RefVector to copy
*/
Vector::Vector(const RefVector& vec) :
data({vec.x(), vec.y(), vec.z(), 0.0}) { }
