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;
}
