void bamParser::parse(Reads & reads, string & filename,
vector<string> & chrs_to_parse) {
BamTools::BamReader bam;
BamTools::BamAlignment read;
string chr;
uint64_t readCnt = 0;
uint32_t meanReadLen = 0;
if (!(bam.Open(filename))) {
throw FileNotGood(filename);
}
const BamTools::RefVector refvec = bam.GetReferenceData();
while (bam.GetNextAlignment(read)) {
chr = getR1Chr(read, refvec);
if (isGoodRead(read)) {
if (isChrToParse(chrs_to_parse, chr)) {
updateAvgReadLength(readCnt, meanReadLen, read);
insertRead(read, reads, chr);
}
}
}
reads.setReadlength(meanReadLen);
}
position BamAlignmentReader::GetLastPositionInBam(const std::string& bamPath, Region::SharedPtr regionPtr)
{
BamTools::BamReader bamReader;
if (!bamReader.Open(bamPath))
{
throw "Unable to open bam file";
}
bamReader.LocateIndex();
int refID = bamReader.GetReferenceID(regionPtr->getReferenceID());
auto referenceData = bamReader.GetReferenceData();
bamReader.Close();
return referenceData[refID].RefLength;
}
inline int
run(Config const& c, TSingleHit)
{
// Create library objects
typedef std::map<std::string, LibraryInfo> TLibraryMap;
typedef std::map<std::string, TLibraryMap> TSampleLibrary;
TSampleLibrary sampleLib;
// Scan libraries
for(unsigned int file_c = 0; file_c < c.files.size(); ++file_c) {
// Get a sample name
std::string sampleName(c.files[file_c].stem().string());
// Check that all input bam files exist
BamTools::BamReader reader;
if ( ! reader.Open(c.files[file_c].string()) ) {
std::cerr << "Could not open input bam file: " << c.files[file_c].string() << std::endl;
reader.Close();
return -1;
}
// Check that all input bam files are indexed
reader.LocateIndex();
if ( !reader.HasIndex() ) {
std::cerr << "Missing bam index file: " << c.files[file_c].string() << std::endl;
reader.Close();
return -1;
}
// Get library parameters and overall maximum insert size
TLibraryMap libInfo;
getLibraryParams(c.files[file_c], libInfo, 0, 5);
sampleLib.insert(std::make_pair(sampleName, libInfo));
}
// Read all SV intervals
typedef std::vector<StructuralVariantRecord> TSVs;
TSVs svs;
std::map<unsigned int, std::string> idToName;
unsigned int intervalCount=1;
if (boost::filesystem::exists(c.int_file) && boost::filesystem::is_regular_file(c.int_file) && boost::filesystem::file_size(c.int_file)) {
Memory_mapped_file interval_file(c.int_file.string().c_str());
char interval_buffer[Memory_mapped_file::MAX_LINE_LENGTH];
while (interval_file.left_bytes() > 0) {
interval_file.read_line(interval_buffer);
// Read single interval line
StructuralVariantRecord sv;
Tokenizer token(interval_buffer, Memory_mapped_file::MAX_LINE_LENGTH);
std::string interval_rname;
token.getString(sv.chr);
sv.svStart = token.getUInt();
sv.svEnd = token.getUInt() + 1;
std::string svName;
token.getString(svName);
idToName.insert(std::make_pair(intervalCount, svName));
sv.id = intervalCount++;
svs.push_back(sv);
}
interval_file.close();
} else {
// Create artificial intervals
BamTools::BamReader readerRef;
if ( ! readerRef.Open(c.files[0].string()) ) return -1;
BamTools::RefVector references = readerRef.GetReferenceData();
typename BamTools::RefVector::const_iterator itRef = references.begin();
for(int refIndex=0;itRef!=references.end();++itRef, ++refIndex) {
int32_t pos = 0;
while (pos < references[refIndex].RefLength) {
int32_t window_len = pos+c.window_size;
if (window_len > references[refIndex].RefLength) window_len = references[refIndex].RefLength;
StructuralVariantRecord sv;
sv.chr = references[refIndex].RefName;
sv.svStart = pos;
sv.svEnd = window_len;
std::stringstream s;
s << sv.chr << ":" << sv.svStart << "-" << sv.svEnd;
idToName.insert(std::make_pair(intervalCount, s.str()));
sv.id = intervalCount++;
svs.push_back(sv);
pos += c.window_offset;
}
}
}
// Output data types
typedef std::pair<std::string, int> TSampleSVPair;
typedef std::pair<int, int> TBpRead;
typedef std::map<TSampleSVPair, TBpRead> TCountMap;
TCountMap countMap;
// Annotate coverage
annotateCoverage(c.files, c.minMapQual, c.inclCigar, sampleLib, svs, countMap, TSingleHit());
// Output library statistics
std::cout << "Library statistics" << std::endl;
TSampleLibrary::const_iterator sampleIt=sampleLib.begin();
for(;sampleIt!=sampleLib.end();++sampleIt) {
std::cout << "Sample: " << sampleIt->first << std::endl;
TLibraryMap::const_iterator libIt=sampleIt->second.begin();
for(;libIt!=sampleIt->second.end();++libIt) {
std::cout << "RG: ID=" << libIt->first << ",Median=" << libIt->second.median << ",MAD=" << libIt->second.mad << ",Orientation=" << (int) libIt->second.defaultOrient << ",MappedReads=" << libIt->second.mappedReads << ",DuplicatePairs=" << libIt->second.non_unique_pairs << ",UniquePairs=" << libIt->second.unique_pairs << std::endl;
}
}
// Output file
boost::iostreams::filtering_ostream dataOut;
dataOut.push(boost::iostreams::gzip_compressor());
dataOut.push(boost::iostreams::file_sink(c.outfile.string().c_str(), std::ios_base::out | std::ios_base::binary));
// Iterate all SVs
typename TSVs::const_iterator itSV = svs.begin();
typename TSVs::const_iterator itSVEnd = svs.end();
for(;itSV!=itSVEnd;++itSV) {
dataOut << itSV->chr << "\t" << itSV->svStart << "\t" << itSV->svEnd << "\t" << idToName.find(itSV->id)->second;
// Iterate all samples
for(unsigned int file_c = 0; file_c < c.files.size(); ++file_c) {
// Get the sample name
std::string sampleName(c.files[file_c].stem().string());
TSampleSVPair sampleSVPair = std::make_pair(sampleName, itSV->id);
typename TCountMap::iterator countMapIt=countMap.find(sampleSVPair);
dataOut << "\t";
if (c.avg_flag) dataOut << ( (countMapIt->second.first) / (double) (itSV->svEnd - itSV->svStart)) << "\t";
if (c.bp_flag) dataOut << countMapIt->second.first << "\t";
dataOut << countMapIt->second.second;
}
dataOut << std::endl;
}
// End
boost::posix_time::ptime now = boost::posix_time::second_clock::local_time();
std::cout << '[' << boost::posix_time::to_simple_string(now) << "] Done." << std::endl;;
return 0;
}
//
// Main
//
int filterBAMMain(int argc, char** argv)
{
parseFilterBAMOptions(argc, argv);
// Read the graph if distance-filtering mode is enabled
StringGraph* pGraph = NULL;
if(!opt::asqgFile.empty())
pGraph = SGUtil::loadASQG(opt::asqgFile, 0, false);
// Read the BWTs if depth-filtering mode is enabled
BWT* pBWT = NULL;
BWT* pRBWT = NULL;
if(!opt::fmIndexPrefix.empty())
{
pBWT = new BWT(opt::fmIndexPrefix + BWT_EXT, opt::sampleRate);
pRBWT = new BWT(opt::fmIndexPrefix + RBWT_EXT, opt::sampleRate);
}
Timer* pTimer = new Timer(PROGRAM_IDENT);
//
int numPairsTotal = 0;
int numPairsFilteredByDistance = 0;
int numPairsFilteredByER = 0;
int numPairsFilteredByQuality = 0;
int numPairsFilteredByDepth = 0;
int numPairsUnmapped = 0;
int numPairsWrote = 0;
// Open the bam files for reading/writing
BamTools::BamReader* pBamReader = new BamTools::BamReader;
pBamReader->Open(opt::bamFile);
BamTools::BamWriter* pBamWriter = new BamTools::BamWriter;
pBamWriter->Open(opt::outFile, pBamReader->GetHeaderText(), pBamReader->GetReferenceData());
const BamTools::RefVector& referenceVector = pBamReader->GetReferenceData();
BamTools::BamAlignment record1;
BamTools::BamAlignment record2;
bool done = false;
while(!done)
{
if(numPairsTotal++ % 200000 == 0)
printf("[sga filterBAM] Processed %d pairs\n", numPairsTotal);
done = !readAlignmentPair(pBamReader, record1, record2);
if(done)
break;
if(!record1.IsMapped() || !record2.IsMapped())
{
numPairsUnmapped += 1;
continue;
}
// Ensure the pairing is correct
if(record1.Name != record2.Name)
{
std::cout << "NAME FAIL: " << record1.Name << " " << record2.Name << "\n";
}
assert(record1.Name == record2.Name);
bool bPassedFilters = true;
// Check if the error rate is below the max
double er1 = getErrorRate(record1);
double er2 = getErrorRate(record2);
if(er1 > opt::maxError || er2 > opt::maxError)
{
bPassedFilters = false;
numPairsFilteredByER += 1;
}
if(record1.MapQuality < opt::minQuality || record2.MapQuality < opt::minQuality)
{
bPassedFilters = false;
numPairsFilteredByQuality += 1;
}
// Perform depth check for pairs aligning to different contigs
if(bPassedFilters && (pBWT != NULL && pRBWT != NULL && opt::maxKmerDepth > 0) && (record1.RefID != record2.RefID))
{
int maxDepth1 = getMaxKmerDepth(record1.QueryBases, pBWT, pRBWT);
int maxDepth2 = getMaxKmerDepth(record1.QueryBases, pBWT, pRBWT);
if(maxDepth1 > opt::maxKmerDepth || maxDepth2 > opt::maxKmerDepth)
{
bPassedFilters = false;
numPairsFilteredByDepth += 1;
}
}
// Perform short-insert pair check
if(pGraph != NULL)
{
bPassedFilters = bPassedFilters && filterByGraph(pGraph, referenceVector, record1, record2);
numPairsFilteredByDistance += 1;
}
if(bPassedFilters)
{
pBamWriter->SaveAlignment(record1);
pBamWriter->SaveAlignment(record2);
numPairsWrote += 1;
}
}
std::cout << "Total pairs: " << numPairsTotal << "\n";
std::cout << "Total pairs output: " << numPairsWrote << "\n";
std::cout << "Total filtered because one pair is unmapped: " << numPairsUnmapped << "\n";
std::cout << "Total filtered by distance: " << numPairsFilteredByDistance << "\n";
std::cout << "Total filtered by error rate: " << numPairsFilteredByER << "\n";
std::cout << "Total filtered by quality: " << numPairsFilteredByQuality << "\n";
std::cout << "Total filtered by depth: " << numPairsFilteredByDepth << "\n";
if(pGraph != NULL)
delete pGraph;
if(pBWT != NULL)
delete pBWT;
if(pRBWT != NULL)
delete pRBWT;
pBamWriter->Close();
pBamReader->Close();
delete pTimer;
delete pBamReader;
delete pBamWriter;
return 0;
}
inline int
run(Config const& c, TCoverageType covType)
{
// Create library objects
typedef boost::unordered_map<std::string, LibraryInfo> TLibraryMap;
typedef boost::unordered_map<std::string, TLibraryMap> TSampleLibrary;
TSampleLibrary sampleLib;
// Scan libraries
for(unsigned int file_c = 0; file_c < c.files.size(); ++file_c) {
// Get a sample name
std::string sampleName(c.files[file_c].stem().string());
// Check that all input bam files exist
BamTools::BamReader reader;
if ( ! reader.Open(c.files[file_c].string()) ) {
std::cerr << "Could not open input bam file: " << c.files[file_c].string() << std::endl;
reader.Close();
return -1;
}
// Check that all input bam files are indexed
reader.LocateIndex();
if ( !reader.HasIndex() ) {
std::cerr << "Missing bam index file: " << c.files[file_c].string() << std::endl;
reader.Close();
return -1;
}
// Get library parameters and overall maximum insert size
TLibraryMap libInfo;
getLibraryParams(c.files[file_c], libInfo, 0, 5);
sampleLib.insert(std::make_pair(sampleName, libInfo));
}
// Get references
BamTools::BamReader readerRef;
if ( ! readerRef.Open(c.files[0].string()) ) return -1;
BamTools::RefVector references = readerRef.GetReferenceData();
// Read all SV intervals
typedef std::vector<CovRecord> TSVs;
TSVs svs;
std::map<unsigned int, std::string> idToName;
unsigned int intervalCount=1;
if (boost::filesystem::exists(c.int_file) && boost::filesystem::is_regular_file(c.int_file) && boost::filesystem::file_size(c.int_file)) {
typedef boost::unordered_map<std::string, unsigned int> TMapChr;
TMapChr mapChr;
typename BamTools::RefVector::const_iterator itRef = references.begin();
for(unsigned int i = 0;itRef!=references.end();++itRef, ++i) mapChr[ itRef->RefName ] = i;
std::ifstream interval_file(c.int_file.string().c_str(), std::ifstream::in);
if (interval_file.is_open()) {
while (interval_file.good()) {
std::string intervalLine;
getline(interval_file, intervalLine);
typedef boost::tokenizer< boost::char_separator<char> > Tokenizer;
boost::char_separator<char> sep(" \t,;");
Tokenizer tokens(intervalLine, sep);
Tokenizer::iterator tokIter = tokens.begin();
if (tokIter!=tokens.end()) {
std::string chrName=*tokIter++;
TMapChr::const_iterator mapChrIt = mapChr.find(chrName);
if (mapChrIt != mapChr.end()) {
if (tokIter!=tokens.end()) {
CovRecord sv;
sv.chr = mapChrIt->second;
sv.svStart = boost::lexical_cast<int32_t>(*tokIter++);
sv.svEnd = boost::lexical_cast<int32_t>(*tokIter++) + 1;
std::string svName = *tokIter;
idToName.insert(std::make_pair(intervalCount, svName));
sv.id = intervalCount++;
svs.push_back(sv);
}
}
}
}
interval_file.close();
}
} else {
// Create artificial intervals
typename BamTools::RefVector::const_iterator itRef = references.begin();
for(int refIndex=0;itRef!=references.end();++itRef, ++refIndex) {
int32_t pos = 0;
unsigned int wSize = c.window_size;
unsigned int wOffset = c.window_offset;
if (c.window_num>0) {
wSize=(itRef->RefLength / c.window_num) + 1;
wOffset=wSize;
}
while (pos < references[refIndex].RefLength) {
int32_t window_len = pos+wSize;
if (window_len > references[refIndex].RefLength) window_len = references[refIndex].RefLength;
CovRecord sv;
sv.chr = refIndex;
sv.svStart = pos;
sv.svEnd = window_len;
std::stringstream s;
s << references[sv.chr].RefName << ":" << sv.svStart << "-" << sv.svEnd;
idToName.insert(std::make_pair(intervalCount, s.str()));
sv.id = intervalCount++;
svs.push_back(sv);
pos += wOffset;
}
}
}
// Output data types
typedef std::pair<std::string, int> TSampleSVPair;
typedef std::pair<int, int> TBpRead;
typedef std::map<TSampleSVPair, TBpRead> TCountMap;
TCountMap countMap;
// Annotate coverage
if (c.inclCigar) annotateCoverage(c.files, c.minGenoQual, sampleLib, svs, countMap, BpLevelType<BpLevelCount>(), covType);
else annotateCoverage(c.files, c.minGenoQual, sampleLib, svs, countMap, BpLevelType<NoBpLevelCount>(), covType);
// Output library statistics
std::cout << "Library statistics" << std::endl;
TSampleLibrary::const_iterator sampleIt=sampleLib.begin();
for(;sampleIt!=sampleLib.end();++sampleIt) {
std::cout << "Sample: " << sampleIt->first << std::endl;
TLibraryMap::const_iterator libIt=sampleIt->second.begin();
for(;libIt!=sampleIt->second.end();++libIt) {
std::cout << "RG: ID=" << libIt->first << ",Median=" << libIt->second.median << ",MAD=" << libIt->second.mad << ",Orientation=" << (int) libIt->second.defaultOrient << std::endl;
}
}
// Output file
boost::iostreams::filtering_ostream dataOut;
dataOut.push(boost::iostreams::gzip_compressor());
dataOut.push(boost::iostreams::file_sink(c.outfile.string().c_str(), std::ios_base::out | std::ios_base::binary));
// Print header
dataOut << "#chr\tstart\tend\tid";
for(unsigned int file_c = 0; file_c < c.files.size(); ++file_c) {
std::string sampleName(c.files[file_c].stem().string());
dataOut << "\t";
if (c.avg_flag) dataOut << sampleName << "_avgcov" << "\t";
if (c.bp_flag) dataOut << sampleName << "_bpcount" << "\t";
if ((c.bp_flag) || (c.avg_flag)) dataOut << sampleName << "_readcount";
else dataOut << sampleName;
}
dataOut << std::endl;
// Iterate all SVs
typename TSVs::const_iterator itSV = svs.begin();
typename TSVs::const_iterator itSVEnd = svs.end();
for(;itSV!=itSVEnd;++itSV) {
dataOut << references[itSV->chr].RefName << "\t" << itSV->svStart << "\t" << itSV->svEnd << "\t" << idToName.find(itSV->id)->second;
// Iterate all samples
for(unsigned int file_c = 0; file_c < c.files.size(); ++file_c) {
// Get the sample name
std::string sampleName(c.files[file_c].stem().string());
TSampleSVPair sampleSVPair = std::make_pair(sampleName, itSV->id);
typename TCountMap::iterator countMapIt=countMap.find(sampleSVPair);
dataOut << "\t";
if (c.avg_flag) dataOut << ( (countMapIt->second.first) / (double) (itSV->svEnd - itSV->svStart)) << "\t";
if (c.bp_flag) dataOut << countMapIt->second.first << "\t";
dataOut << countMapIt->second.second;
}
dataOut << std::endl;
}
// End
boost::posix_time::ptime now = boost::posix_time::second_clock::local_time();
std::cout << '[' << boost::posix_time::to_simple_string(now) << "] Done." << std::endl;;
return 0;
}