//{{{ 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);
}
}
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) );
}
}
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);
}
}
}
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;
}
//{{{ 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;
}
// 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;
}
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 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();
}
}
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();
}
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();
}
}
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