void TargetsManager::Initialize(const ReferenceReader& ref_reader, const string& _targets, bool _trim_ampliseq_primers /*const ExtendParameters& parameters*/)
{
//
// Step 1. Retrieve raw target definitions
//
list<UnmergedTarget> raw_targets;
if (not _targets.empty()) {
LoadRawTargets(ref_reader, _targets, raw_targets);
} else {
for (int chr = 0; chr < ref_reader.chr_count(); ++chr) {
raw_targets.push_back(UnmergedTarget());
UnmergedTarget& target = raw_targets.back();
target.begin = 0;
target.end = ref_reader.chr_size(chr);
target.chr = chr;
}
}
//
// Step 2. Sort raw targets and transfer to the vector
//
int num_unmerged = raw_targets.size();
vector<UnmergedTarget*> raw_sort;
raw_sort.reserve(num_unmerged);
for (list<UnmergedTarget>::iterator I = raw_targets.begin(); I != raw_targets.end(); ++I)
raw_sort.push_back(&(*I));
sort(raw_sort.begin(), raw_sort.end(), CompareTargets);
unmerged.reserve(num_unmerged);
bool already_sorted = true;
list<UnmergedTarget>::iterator I = raw_targets.begin();
for (int idx = 0; idx < num_unmerged; ++idx, ++I) {
if (raw_sort[idx] != &(*I) and already_sorted) {
already_sorted = false;
cerr << "TargetsManager: BED not sorted at position " << idx;
cerr << " replaced " << I->name << ":" << I->chr << ":" << I->begin << "-" << I->end;
cerr << " with " << raw_sort[idx]->name << ":" << raw_sort[idx]->chr << ":" << raw_sort[idx]->begin << "-" << raw_sort[idx]->end << endl;
}
unmerged.push_back(*raw_sort[idx]);
}
//
// Step 3. Merge targets and link merged/unmerged entries
//
merged.reserve(num_unmerged);
bool already_merged = true;
for (int idx = 0; idx < num_unmerged; ++idx) {
if (idx and merged.back().chr == unmerged[idx].chr and merged.back().end >= unmerged[idx].begin) {
merged.back().end = max(merged.back().end, unmerged[idx].end);
already_merged = false;
} else {
merged.push_back(MergedTarget());
merged.back().chr = unmerged[idx].chr;
merged.back().begin = unmerged[idx].begin;
merged.back().end = unmerged[idx].end;
merged.back().first_unmerged = idx;
}
unmerged[idx].merged = merged.size();
}
if (_targets.empty()) {
cout << "TargetsManager: No targets file specified, processing entire reference" << endl;
} else {
cout << "TargetsManager: Loaded targets file " << _targets << endl;
cout << "TargetsManager: " << num_unmerged << " target(s)";
if (not already_merged)
cout << " (" << merged.size() << " after merging)";
cout << endl;
if (not already_sorted)
cout << "TargetsManager: Targets required sorting" << endl;
trim_ampliseq_primers = _trim_ampliseq_primers;
if (trim_ampliseq_primers)
cout << "TargetsManager: Trimming of AmpliSeq primers is enabled" << endl;
}
}
// open BAM input file
void BAMWalkerEngine::InitializeBAMs(const ReferenceReader& ref_reader, const vector<string>& bam_filenames)
{
if (not bam_reader_.SetExplicitMergeOrder(BamMultiReader::MergeByCoordinate)) {
cerr << "ERROR: Could not set merge order to BamMultiReader::MergeByCoordinate" << endl;
exit(1);
}
if (not bam_reader_.Open(bam_filenames)) {
cerr << "ERROR: Could not open input BAM file(s) : " << bam_reader_.GetErrorString() << endl;
exit(1);
}
if (not bam_reader_.LocateIndexes()) {
cerr << "ERROR: Could not open BAM index file(s) : " << bam_reader_.GetErrorString() << endl;
exit(1);
}
// BAM multi reader combines the read group information of the different BAMs but does not merge comment sections
bam_header_ = bam_reader_.GetHeader();
if (!bam_header_.HasReadGroups()) {
cerr << "ERROR: there is no read group in BAM files specified" << endl;
exit(1);
}
// Manually merge comment sections of BAM files if we have more than one BAM file
if (bam_filenames.size() > 1) {
unsigned int num_duplicates = 0;
unsigned int num_merged = 0;
for (unsigned int bam_idx = 0; bam_idx < bam_filenames.size(); bam_idx++) {
BamReader reader;
if (not reader.Open(bam_filenames.at(bam_idx))) {
cerr << "TVC ERROR: Failed to open input BAM file " << reader.GetErrorString() << endl;
exit(1);
}
SamHeader header = reader.GetHeader();
for (unsigned int i_co = 0; i_co < header.Comments.size(); i_co++) {
// Step 1: Check if this comment is already part of the merged header
unsigned int m_co = 0;
while (m_co < bam_header_.Comments.size() and bam_header_.Comments.at(m_co) != header.Comments.at(i_co))
m_co++;
if (m_co < bam_header_.Comments.size()){
num_duplicates++;
continue;
}
// Add comment line to merged header if it is a new one
num_merged++;
bam_header_.Comments.push_back(header.Comments.at(i_co));
}
}
// Verbose what we did
cout << "Merged " << num_merged << " unique comment lines into combined BAM header. Encountered " << num_duplicates << " duplicate comments." << endl;
}
//
// Reference sequences in the bam file must match that in the fasta file
//
vector<RefData> referenceSequences = bam_reader_.GetReferenceData();
if ((int)referenceSequences.size() != ref_reader.chr_count()) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " BAM has " << referenceSequences.size()
<< " chromosomes while fasta has " << ref_reader.chr_count() << endl;
exit(1);
}
for (int chr_idx = 0; chr_idx < ref_reader.chr_count(); ++chr_idx) {
if (referenceSequences[chr_idx].RefName != ref_reader.chr_str(chr_idx)) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " Chromosome #" << (chr_idx+1) << "in BAM is " << referenceSequences[chr_idx].RefName
<< " while fasta has " << ref_reader.chr_str(chr_idx) << endl;
exit(1);
}
if (referenceSequences[chr_idx].RefLength != ref_reader.chr_size(chr_idx)) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " Chromosome " << referenceSequences[chr_idx].RefName
<< "in BAM has length " << referenceSequences[chr_idx].RefLength
<< " while fasta has " << ref_reader.chr_size(chr_idx) << endl;
exit(1);
}
}
//
// Retrieve BaseCaller and TMAP version strings from BAM header
//
set<string> basecaller_versions;
set<string> tmap_versions;
for (SamProgramIterator I = bam_header_.Programs.Begin(); I != bam_header_.Programs.End(); ++I) {
if (I->ID.substr(0,2) == "bc")
basecaller_versions.insert(I->Version);
if (I->ID.substr(0,4) == "tmap")
tmap_versions.insert(I->Version);
}
basecaller_version_ = "";
for (set<string>::const_iterator I = basecaller_versions.begin(); I != basecaller_versions.end(); ++I) {
if (not basecaller_version_.empty())
basecaller_version_ += ", ";
basecaller_version_ += *I;
}
tmap_version_ = "";
for (set<string>::const_iterator I = tmap_versions.begin(); I != tmap_versions.end(); ++I) {
if (not tmap_version_.empty())
tmap_version_ += ", ";
tmap_version_ += *I;
}
}
// open BAM input file
void BAMWalkerEngine::InitializeBAMs(const ReferenceReader& ref_reader, const vector<string>& bam_filenames)
{
if (not bam_reader_.SetExplicitMergeOrder(BamMultiReader::MergeByCoordinate)) {
cerr << "ERROR: Could not set merge order to BamMultiReader::MergeByCoordinate" << endl;
exit(1);
}
if (not bam_reader_.Open(bam_filenames)) {
cerr << "ERROR: Could not open input BAM file(s) : " << bam_reader_.GetErrorString();
exit(1);
}
if (not bam_reader_.LocateIndexes()) {
cerr << "ERROR: Could not open BAM index file(s) : " << bam_reader_.GetErrorString();
exit(1);
}
bam_header_ = bam_reader_.GetHeader();
if (!bam_header_.HasReadGroups()) {
cerr << "ERROR: there is no read group in BAM files specified" << endl;
exit(1);
}
//
// Reference sequences in the bam file must match that in the fasta file
//
vector<RefData> referenceSequences = bam_reader_.GetReferenceData();
if ((int)referenceSequences.size() != ref_reader.chr_count()) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " BAM has " << referenceSequences.size()
<< " chromosomes while fasta has " << ref_reader.chr_count() << endl;
exit(1);
}
for (int chr_idx = 0; chr_idx < ref_reader.chr_count(); ++chr_idx) {
if (referenceSequences[chr_idx].RefName != ref_reader.chr_str(chr_idx)) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " Chromosome #" << (chr_idx+1) << "in BAM is " << referenceSequences[chr_idx].RefName
<< " while fasta has " << ref_reader.chr_str(chr_idx) << endl;
exit(1);
}
if (referenceSequences[chr_idx].RefLength != ref_reader.chr_size(chr_idx)) {
cerr << "ERROR: Reference in BAM file does not match fasta file" << endl
<< " Chromosome " << referenceSequences[chr_idx].RefName
<< "in BAM has length " << referenceSequences[chr_idx].RefLength
<< " while fasta has " << ref_reader.chr_size(chr_idx) << endl;
exit(1);
}
}
//
// Retrieve BaseCaller and TMAP version strings from BAM header
//
set<string> basecaller_versions;
set<string> tmap_versions;
for (SamProgramIterator I = bam_header_.Programs.Begin(); I != bam_header_.Programs.End(); ++I) {
if (I->ID.substr(0,2) == "bc")
basecaller_versions.insert(I->Version);
if (I->ID.substr(0,4) == "tmap")
tmap_versions.insert(I->Version);
}
basecaller_version_ = "";
for (set<string>::const_iterator I = basecaller_versions.begin(); I != basecaller_versions.end(); ++I) {
if (not basecaller_version_.empty())
basecaller_version_ += ", ";
basecaller_version_ += *I;
}
tmap_version_ = "";
for (set<string>::const_iterator I = tmap_versions.begin(); I != tmap_versions.end(); ++I) {
if (not tmap_version_.empty())
tmap_version_ += ", ";
tmap_version_ += *I;
}
}
