void InputStructures::DetectFlowOrderzAndKeyFromBam(const SamHeader &samHeader){
//TODO Need to handle multiple BAM files with different flowOrders, at least throw an error for now.
for (BamTools::SamReadGroupConstIterator itr = samHeader.ReadGroups.Begin(); itr != samHeader.ReadGroups.End(); ++itr) {
if (itr->HasFlowOrder()) {
string tmpflowOrder = itr->FlowOrder;
if (bamFlowOrderVector.empty()){
bamFlowOrderVector.push_back(tmpflowOrder);
flowOrder = tmpflowOrder; // first one free
}else { //check if the flowOrder is the same if not throw an error, for now we dont support bams with different flow orders
vector<string>::iterator it = std::find(bamFlowOrderVector.begin(), bamFlowOrderVector.end(), tmpflowOrder);
if (it == bamFlowOrderVector.end()) {
// check to see if flowOrder is a substring/superstring first
std::size_t found_me = std::string::npos; // assume bad
if (tmpflowOrder.length()>flowOrder.length()){
found_me = tmpflowOrder.find(flowOrder);
if (found_me==0){ // must find at first position
flowOrder = tmpflowOrder; // longer superstring
bamFlowOrderVector.push_back(tmpflowOrder);
//cout<< "Super: " << tmpflowOrder.length() << " " << tmpflowOrder << endl;
} else
found_me = std::string::npos;
}else{
found_me = flowOrder.find(tmpflowOrder);
if (found_me==0){ // must find at first position
// substring, so no need to update flowOrder
bamFlowOrderVector.push_back(tmpflowOrder);
//cout << "Sub: " << tmpflowOrder.length() << " "<< tmpflowOrder << endl;
} else
found_me = std::string::npos;
}
if (found_me==std::string::npos){
cerr << "FATAL ERROR: BAM files specified as input have different flow orders. Currently tvc supports only BAM files with same flow order. " << endl;
exit(-1);
}
}
}
flowKey = itr->KeySequence;
}
}
if (bamFlowOrderVector.size()>1)
cout << "Compatibly nested flow orders found: " << bamFlowOrderVector.size() << " using longest, nFlows= " << flowOrder.length() << endl;
//cout << "Final: " << flowOrder.length() << " " << flowOrder << endl;
nFlows = flowOrder.length();
if (nFlows > 0) {
flowSigPresent = true;
treePhaserFlowOrder.SetFlowOrder(flowOrder, nFlows);
key.Set(treePhaserFlowOrder, flowKey, "key");
}
}
int scanBam()
{
std::vector< std::map<std::string, ScanResults> > resultlist;
// std::ostream* pWriter;
// pWriter = &std::cout;
bool isExome = opt::exomebedfile.size()==0? false: true;
std::cout << opt::bamlist << "\n";
std::cout << opt::bamlist.size() << " BAMs" << std::endl;
for(std::size_t i=0; i<opt::bamlist.size(); i++) {
// storing results for each read group (RG tag). use
// read group ID as key.
std::map<std::string, ScanResults> resultmap;
// store where the overlap was last found in the case of exome seq
std::map<std::string, std::vector<range>::iterator> lastfound;
std::vector<range>::iterator searchhint;
std::cerr << "Start analysing BAM " << opt::bamlist[i] << "\n";
// Open the bam files for reading/writing
BamTools::BamReader* pBamReader = new BamTools::BamReader;
pBamReader->Open(opt::bamlist[i]);
// get bam headers
const BamTools::SamHeader header = pBamReader ->GetHeader();
// for(BamTools::SamSequenceConstIterator it = header.Sequences.Begin();
// it != header.Sequences.End();++it){
// std::cout << "Assembly ID:" << it->AssemblyID << ", Name:" << it->Name << std::endl;
// }
// exit(0);
bool rggroups=false;
if(opt::ignorerg){ // ignore read groups
std::cerr << "Treat all reads in BAM as if they were from a same sample" << std::endl;
ScanResults results;
results.sample = opt::unknown;
resultmap[opt::unknown]=results;
}else{
std::map <std::string, std::string> readgroups;
std::map <std::string, std::string> readlibs;
rggroups = header.HasReadGroups();
if(rggroups){
for(BamTools::SamReadGroupConstIterator it = header.ReadGroups.Begin();
it != header.ReadGroups.End();++it){
readgroups[it->ID]= it->Sample;
if(it->HasLibrary()){
readlibs[it->ID] = it -> Library;
}else{
readlibs[it->ID] = opt::unknown;
}
}
std::cerr<<"Specified BAM has "<< readgroups.size()<< " read groups" << std::endl;
for(std::map<std::string, std::string>::iterator it = readgroups.begin(); it != readgroups.end(); ++it){
ScanResults results;
std::string rgid = it -> first;
results.sample = it -> second;
results.lib = readlibs[rgid];
resultmap[rgid]=results; //results are identified by RG tag.
}
}else{
std::cerr << "Warning: can't find RG tag in the BAM header" << std::endl;
std::cerr << "Warning: treat all reads in BAM as if they were from a same sample" << std::endl;
ScanResults results;
results.sample = opt::unknown;
results.lib = opt::unknown;
resultmap[opt::unknown]=results;
}
}
BamTools::BamAlignment record1;
bool done = false;
int nprocessed=0; // number of reads analyzed
int ntotal=0; // number of reads scanned in bam (we skip some reads, see below)
while(!done)
{
ntotal ++;
done = !pBamReader -> GetNextAlignment(record1);
std::string tag = opt::unknown;
if(rggroups){
// skip reads that do not have read group tag
if(record1.HasTag("RG")){
record1.GetTag("RG", tag);
// std::cerr << c << " reads:{" << record1.QueryBases << "} tag:{" << tag << "}\n";
}else{
std::cerr << "can't find RG tag for read at position {" << record1.RefID << ":" << record1.Position << "}" << std::endl;
std::cerr << "skip this read" << std::endl;
continue;
}
}
// skip reads with readgroup not defined in BAM header
if(resultmap.find(tag) == resultmap.end()){
std::cerr << "RG tag {" << tag << "} for read at position ";
std::cerr << "{" << record1.RefID << ":" << record1.Position << "} doesn't exist in BAM header.";
continue;
}
// for exome, exclude reads mapped to the exome regions.
if(isExome){
range rg;
rg.first = record1.Position;
rg.second = record1.Position + record1.Length;
std::string chrm = refID2Name(record1.RefID);
if(chrm != "-1"){ // check if overlap exome when the read is mapped to chr1-22, X, Y
// std::cerr << "read: " << chrm << " " << rg << "\n" << std::endl;
std::map<std::string, std::vector<range> >::iterator chrmit = opt::exomebed.find(chrm);
if(chrmit == opt::exomebed.end())
{
// std::cerr<<"chromosome or reference sequence: " << chrm << " is not present in the specified exome bed file." <<std::endl;
// std::cerr<<"please check sequence name encoding, i.e. for chromosome one, is it chr1 or 1" << std::endl;
// unmapped reads can have chr names as a star (*). We also don't consider MT reads.
resultmap[tag].n_exreadsChrUnmatched +=1;
}else{
std::vector<range>::iterator itend = opt::exomebed[chrm].end();
std::map<std::string, std::vector<range>::iterator>::iterator lastfoundchrmit = lastfound.find(chrm);
if(lastfoundchrmit == lastfound.end()){ // first entry to this chrm
lastfound[chrm] = chrmit->second.begin();// start from begining
}
// set the hint to where the previous found is
searchhint = lastfound[chrm];
std::vector<range>::iterator itsearch = searchRange(searchhint, itend, rg);
// if found
if(itsearch != itend){// if found
searchhint = itsearch;
resultmap[tag].n_exreadsExcluded +=1;
lastfound[chrm] = searchhint; // update search hint
continue;
}
}
}
}
resultmap[tag].numTotal +=1;
if(record1.IsMapped())
{
resultmap[tag].numMapped += 1;
}
if(record1.IsDuplicate()){
resultmap[tag].numDuplicates +=1;
}
double gc = calcGC(record1.QueryBases);
int ptn_count = countMotif(record1.QueryBases, opt::PATTERN, opt::PATTERN_REV);
// when the read length exceeds 100bp, number of patterns might exceed the boundary
if (ptn_count > ScanParameters::TEL_MOTIF_N-1){
continue;
}
resultmap[tag].telcounts[ptn_count]+=1;
if(gc >= ScanParameters::GC_LOWERBOUND && gc <= ScanParameters::GC_UPPERBOUND){
// get index for GC bin.
int idx = floor((gc-ScanParameters::GC_LOWERBOUND)/ScanParameters::GC_BINSIZE);
assert(idx >=0 && idx <= ScanParameters::GC_BIN_N-1);
// std::cerr << c << " GC:{"<< gc << "} telcounts:{"<< ptn_count <<"} GC idx{" << idx << "}\n";
if(idx > ScanParameters::GC_BIN_N-1){
std::cerr << nprocessed << " GC:{"<< gc << "} telcounts:{"<< ptn_count <<"} GC bin index out of bound:" << idx << "\n";
exit(EXIT_FAILURE);
}
resultmap[tag].gccounts[idx]+=1;
}
// if(resultmap[tag].n_exreadsChrUnmatched > 1000){
// std::cerr<<"too many reads found with unmatched chromosome ID between BAM and exome BED. \n" << std::endl;
// }
nprocessed++;
if( nprocessed%10000000 == 0){
std::cerr << "[scan] processed " << nprocessed << " reads \n" ;
}
}
pBamReader->Close();
delete pBamReader;
// consider each BAM separately
resultlist.push_back(resultmap);
std::cerr << "[scan] total reads in BAM scanned " << ntotal << std::endl;
std::cerr << "Completed scanning BAM\n";
}
if(opt::onebam){
merge_results_by_readgroup(resultlist);
}
outputresults(resultlist);
if(isExome){
printlog(resultlist);
}
std::cerr << "Completed writing results\n";
return 0;
}
void InputStructures::DetectFlowOrderzAndKeyFromBam(const SamHeader &samHeader){
// We only store flow orders that are different from each other in the flow order vector.
// The same flow order but different total numbers of flow map to the same flow order object
// So multiple runs, even with different max flows, point to the same flow order object
// We assume that the read group name is written in the form <run_id>.<Barcode Name>
flow_order_vector.clear();
vector<string> temp_flow_order_vector;
int num_read_groups = 0;
for (BamTools::SamReadGroupConstIterator itr = samHeader.ReadGroups.Begin(); itr != samHeader.ReadGroups.End(); ++itr) {
num_read_groups++;
if (itr->ID.empty()){
cerr << "TVC ERROR: BAM file has a read group without ID." << endl;
exit(EXIT_FAILURE);
}
// We need a flow order to do variant calling so throw an error if there is none.
if (not itr->HasFlowOrder()) {
cerr << "TVC ERROR: read group " << itr->ID << " does not have a flow order." << endl;
exit(EXIT_FAILURE);
}
// Check for duplicate read group ID entries and throw an error if one is found
std::map<string,string>::const_iterator key_it = key_by_read_group.find(itr->ID);
if (key_it != key_by_read_group.end()) {
cerr << "TVC ERROR: Multiple read group entries with ID " << itr->ID << endl;
exit(EXIT_FAILURE);
}
// Store Key Sequence for each read group
// The read group key in the BAM file contains the full prefix: key sequence + barcode + barcode adapter
key_by_read_group[itr->ID] = itr->KeySequence;
// Get run id from read group name: convention <read group name> = <run_id>.<Barcode Name>
string run_id = itr->ID.substr(0,itr->ID.find('.'));
if (run_id.empty()) {
cerr << "TVC ERROR: Unable to extract run id from read group name " << itr->ID << endl;
exit(EXIT_FAILURE);
}
// Check whether an entry already exists for this run id and whether it is compatible
std::map<string,int>::const_iterator fo_it = flow_order_index_by_run_id.find(run_id);
if (fo_it != flow_order_index_by_run_id.end()) {
// Flow order for this run id may be equal or a subset of the stored one
if (temp_flow_order_vector.at(fo_it->second).length() < itr->FlowOrder.length()
or temp_flow_order_vector.at(fo_it->second).substr(0, itr->FlowOrder.length()) != itr->FlowOrder
or num_flows_by_run_id.at(run_id) != (int)(itr->FlowOrder).length())
{
cerr << "TVC ERROR: Flow order information extracted from read group name " << itr->ID
<< " does not match previous entry for run id " << run_id << ": " << endl;
cerr << "Exiting entry : " << temp_flow_order_vector.at(fo_it->second) << endl;
cerr << "Newly extracted: " << itr->FlowOrder << endl;
cerr << temp_flow_order_vector.at(fo_it->second) << endl;
exit(EXIT_FAILURE);
}
// Found matching entry and everything is OK.
continue;
}
// New run id: Check whether this flow order is the same or a sub/ superset of an existing flow order
unsigned int iFO = 0;
for (; iFO< temp_flow_order_vector.size(); iFO++){
// Is the new flow order a subset of an existing flow order?
if ( temp_flow_order_vector.at(iFO).length() >= itr->FlowOrder.length() ) {
if (temp_flow_order_vector.at(iFO).substr(0, itr->FlowOrder.length()) == itr->FlowOrder ) {
flow_order_index_by_run_id[run_id] = iFO;
num_flows_by_run_id[run_id] = itr->FlowOrder.length();
break;
}
else
continue;
}
// Is the new flow order a superset of an existing flow order?
if ( temp_flow_order_vector.at(iFO).length() < itr->FlowOrder.length() ) {
if ( itr->FlowOrder.substr(0, temp_flow_order_vector.at(iFO).length()) == temp_flow_order_vector.at(iFO) ) {
temp_flow_order_vector.at(iFO) = itr->FlowOrder;
flow_order_index_by_run_id[run_id] = iFO;
num_flows_by_run_id[run_id] = itr->FlowOrder.length();
break;
}
}
}
// Do we have a new flow order?
if (iFO == temp_flow_order_vector.size()) {
temp_flow_order_vector.push_back(itr->FlowOrder);
flow_order_index_by_run_id[run_id] = iFO;
num_flows_by_run_id[run_id] = itr->FlowOrder.length();
}
} // --- end loop over read groups
// Now we have amassed all the unique flow orders and can construct the FlowOrder objects
for (unsigned int iFO=0; iFO < temp_flow_order_vector.size(); iFO++){
ion::FlowOrder tempIonFlowOrder(temp_flow_order_vector.at(iFO), temp_flow_order_vector.at(iFO).length());
flow_order_vector.push_back(tempIonFlowOrder);
}
// Verbose output
cout << "TVC found a total of " << flow_order_vector.size() << " different flow orders of max flow lengths: ";
int iFO=0;
for (; iFO<(int)flow_order_vector.size()-1; iFO++)
cout << flow_order_vector.at(iFO).num_flows() << ',';
cout << flow_order_vector.at(iFO).num_flows() << endl;
}