GenomicRegion::GenomicRegion(const std::string& tchr, const std::string& tpos1, const std::string& tpos2, const SeqLib::BamHeader& hdr)
{
strand = '*';
// convert the pos strings
// throws invalid_argument if conversion can't be performed
// or throws an out_of_range if it is too big for result
#ifdef HAVE_C11
pos1 = std::stoi(tpos1);
pos2 = std::stoi(tpos2);
#else
pos1 = std::atoi(tpos1.c_str());
pos2 = std::atoi(tpos2.c_str());
#endif
// if no header, assume that it is "standard"
if (hdr.isEmpty()) {
if (tchr == "X" || tchr == "chrX")
chr = 22;
else if (tchr == "Y" || tchr == "chrY")
chr = 23;
else
#ifdef HAVE_C11
chr = std::stoi(SeqLib::scrubString(tchr, "chr")) - 1;
#else
chr = std::atoi(SeqLib::scrubString(tchr, "chr").c_str());
#endif
return;
} else {
chr = hdr.Name2ID(tchr); //bam_name2id(hdr.get(), tchr.c_str());
}
}
int runFishhook(int argc, char** argv) {
parseFishOptions(argc, argv);
if (opt::verbose) {
std::cerr << "FishHook Params: " << std::endl
<< "\tWidth: " << SeqLib::AddCommas(opt::width) << std::endl
<< "\tEvents: " << opt::events << std::endl
<< "\tCoverage Mask: " << opt::coverage << std::endl
<< "\tSlop: " << SeqLib::AddCommas(opt::slop) << std::endl
<< "\tInterval Tracks: " << std::endl;
for (auto& i : opt::interval_files)
std::cerr << "\t-- " << i << std::endl;
std::cerr << "\tScored Tracks: " << std::endl;
for (auto& i : opt::scored_files)
std::cerr << "\t-- " << i << std::endl;
std::cerr << "\tSequence Features: " << std::endl;
for (auto& i : opt::seq_features)
std::cerr << "\t-- " << i << std::endl;
}
// read in the covariate tracks
SeqHashMap<std::string, Fractions> intervals;
// read a header for info
SeqLib::BamReader rdr;
if (!rdr.Open(opt::bam)) {
std::cerr << "Error: Could not read BAM supplied by -b: " << opt::bam << std::endl;
exit(EXIT_FAILURE);
}
hdr = rdr.Header();
// read in the reference genome
SeqLib::RefGenome ref;
if (!ref.LoadIndex(opt::refgenome)) {
if (opt::seq_features.size()) {
std::cerr << "Error: Could not read referene genome supplied by -G: " << opt::refgenome << std::endl;
exit(EXIT_FAILURE);
}
}
// read in the events
if (opt::verbose) std::cerr << "...reading events " << opt::events << std::endl;
EventList events;
if (!events.readFromBed(opt::events, hdr)) {
std::cerr << "Error: Could not read events BED: " << opt::events << std::endl;
exit(EXIT_FAILURE);
}
if (opt::verbose) std::cerr << "...read in " << SeqLib::AddCommas(events.size()) << " events" << std::endl;
events.CreateTreeMap();
// create the tiled regions
FishHookTiles fish(opt::width, opt::slop, hdr.GetHeaderSequenceVector());
if (opt::verbose)
std::cerr << "...constructed " << SeqLib::AddCommas(fish.size()) << " fishhook intervals" << std::endl;
fish.CreateTreeMap();
// read the coverage mask
SeqLib::GRC cov;
if (!opt::coverage.empty()) {
if (opt::verbose) std::cerr << "...reading coverage mask " << opt::coverage << std::endl;
cov.ReadBED(opt::coverage, hdr);
if (opt::verbose) std::cerr << "...read in " << SeqLib::AddCommas(cov.size()) << " covered regions " << std::endl;
if (!cov.size()) {
std::cerr << "Non-empty coverage track read with 0 regions. Check that is non-empty BED" << std::endl;
exit(EXIT_FAILURE);
}
if (opt::verbose) std::cerr << "...creating interval tree map on covered regions and overlapping with tiles" << std::endl;
cov.CreateTreeMap();
// find covered amount per tile
std::vector<int32_t> q, s;
SeqLib::GRC ovlp;
// fish is subject
if (fish.size() > cov.size()) // do in most efficient order
ovlp = cov.FindOverlaps(fish, q, s, false);
else
ovlp = fish.FindOverlaps(cov, s, q, false);
if (opt::verbose) std::cerr << "..." << SeqLib::AddCommas(ovlp.size()) << " regions are covered" << std::endl;
// set the amount covered by each
for (size_t i = 0; i < ovlp.size(); ++i) {
fish[s[i]].covered += (double)ovlp[i].Width() / fish[s[i]].Width();
}
// mask the events
q.clear(); s.clear(); ovlp.clear();
// events is subject
if (events.size() > cov.size()) // do in most efficient order
ovlp = cov.FindOverlaps(events, q, s, false);
else
ovlp = events.FindOverlaps(cov, s, q, false);
EventList newe;
// set the amount covered by each
for (size_t i = 0; i < ovlp.size(); ++i) {
newe.add(Event(ovlp[i], events.at(s[i]).id));
}
events = newe;
events.CreateTreeMap();
if (opt::verbose) std::cerr << "...kept " << SeqLib::AddCommas(events.size()) << " events after mask" << std::endl;
} else {
for (auto& i : fish)
i.covered = 1; // the entire thing is covered if no mask provided
}
// read in the interval tracks
for (auto& i : opt::interval_files)
read_track(i, intervals, cov, false);
for (auto& i : opt::scored_files)
read_track(i, intervals, cov, true);
// count events per tile (also de-dupes on patient per bin)
fish.CountEvents(events);
// overlap the covariates with the tiles
for (auto& i : intervals) {
fish.AddIntervalCovariate(i.first, i.second);
}
// make the matrix
FishModel fm;
fm.AddTiles(fish);
fm.SetNumThreads(opt::num_threads);
fm.EstimateOLS();
fm.CooksDistance(fm.GetOLS());
// write the covariates
fish.PrintBEDHeader(std::cout);
fish.PrintBED(std::cout, hdr);
return 0;
}
int main(int argc, char** argv) {
int c;
FastaReference reference;
bool has_ref = false;
bool suppress_output = false;
bool debug = false;
bool isuncompressed = true;
int maxiterations = 50;
if (argc < 2) {
printUsage(argv);
exit(1);
}
while (true) {
static struct option long_options[] =
{
{"help", no_argument, 0, 'h'},
{"debug", no_argument, 0, 'd'},
{"fasta-reference", required_argument, 0, 'f'},
{"max-iterations", required_argument, 0, 'm'},
{"suppress-output", no_argument, 0, 's'},
{"compressed", no_argument, 0, 'c'},
{0, 0, 0, 0}
};
int option_index = 0;
c = getopt_long (argc, argv, "hdcsf:m:",
long_options, &option_index);
/* Detect the end of the options. */
if (c == -1)
break;
switch (c) {
case 'f':
reference.open(optarg); // will exit on open failure
has_ref = true;
break;
case 'm':
maxiterations = atoi(optarg);
break;
case 'd':
debug = true;
break;
case 's':
suppress_output = true;
break;
case 'c':
isuncompressed = false;
break;
case 'h':
printUsage(argv);
exit(0);
break;
case '?':
printUsage(argv);
exit(1);
break;
default:
abort();
break;
}
}
if (!has_ref) {
cerr << "no FASTA reference provided, cannot realign" << endl;
exit(1);
}
BAMSINGLEREADER reader;
if (!reader.Open(STDIN)) {
cerr << "could not open stdin for reading" << endl;
exit(1);
}
#ifdef HAVE_BAMTOOLS
BamWriter writer;
if (isuncompressed) {
writer.SetCompressionMode(BamWriter::Uncompressed);
}
if (!suppress_output && !writer.Open("stdout", reader.GetHeaderText(), reader.GetReferenceData())) {
cerr << "could not open stdout for writing" << endl;
exit(1);
}
#else
SeqLib::BamWriter writer(isuncompressed ? SeqLib::SAM : SeqLib::BAM);
SeqLib::BamHeader hdr = reader.Header();
if (hdr.isEmpty()) {
cerr << "could not open header for input" << endl;
exit(1);
}
writer.SetHeader(hdr);
if (!suppress_output && !writer.Open("-")) {
cerr << "could not open stdout for writing" << endl;
exit(1);
}
#endif
// store the names of all the reference sequences in the BAM file
map<int, string> referenceIDToName;
REFVEC referenceSequences = reader.GETREFDATA;
int i = 0;
for (REFVEC::iterator r = referenceSequences.begin(); r != referenceSequences.end(); ++r) {
referenceIDToName[i] = r->REFNAME;
++i;
}
BAMALIGN alignment;
while (GETNEXT(reader, alignment)) {
DEBUG("--------------------------- read --------------------------" << endl);
DEBUG("| " << referenceIDToName[alignment.REFID] << ":" << alignment.POSITION << endl);
DEBUG("| " << alignment.QNAME << ":" << alignment.ENDPOSITION << endl);
DEBUG("| " << alignment.QNAME << ":" << (alignment.ISMAPPED ? " mapped" : " unmapped") << endl);
DEBUG("| " << alignment.QNAME << ":" << " cigar data size: " << alignment.GETCIGAR.size() << endl);
DEBUG("--------------------------- realigned --------------------------" << endl);
// skip unmapped alignments, as they cannot be left-realigned without CIGAR data
if (alignment.ISMAPPED) {
int endpos = alignment.ENDPOSITION;
int length = endpos - alignment.POSITION + 1;
if (alignment.POSITION >= 0 && length > 0) {
if (!stablyLeftAlign(alignment,
reference.getSubSequence(
referenceIDToName[alignment.REFID],
alignment.POSITION,
length),
maxiterations, debug)) {
cerr << "unstable realignment of " << alignment.QNAME
<< " at " << referenceIDToName[alignment.REFID] << ":" << alignment.POSITION << endl
<< alignment.QUERYBASES << endl;
}
}
}
DEBUG("----------------------------------------------------------------" << endl);
DEBUG(endl);
if (!suppress_output)
WRITEALIGNMENT(writer, alignment);
}
reader.Close();
if (!suppress_output)
writer.Close();
return 0;
}