void trigger(DataObject<D>& d) {
boost::shared_lock_guard<boost::shared_mutex> lock(d._mtx_links);
if (d._links.empty())
return;
for (auto &p : d._links)
_tbbExecutionQueue.push(p.second);
}
bool computeBiasFeaturesHelper(ParserT& parser,
tbb::concurrent_bounded_queue<TranscriptFeatures>& featQueue,
size_t& numComplete, size_t numThreads) {
using stream_manager = jellyfish::stream_manager<std::vector<std::string>::iterator>;
using sequence_parser = jellyfish::whole_sequence_parser<stream_manager>;
size_t merLen = 2;
Kmer lshift(2 * (merLen - 1));
Kmer masq((1UL << (2 * merLen)) - 1);
std::atomic<size_t> readNum{0};
size_t numActors = numThreads;
std::vector<std::thread> threads;
auto tstart = std::chrono::steady_clock::now();
for (auto i : boost::irange(size_t{0}, numActors)) {
threads.push_back(std::thread(
[&featQueue, &numComplete, &parser, &readNum, &tstart, lshift, masq, merLen, numActors]() -> void {
size_t cmlen, numKmers;
jellyfish::mer_dna_ns::mer_base_dynamic<uint64_t> kmer(merLen);
// while there are transcripts left to process
while (true) { //producer.nextRead(s)) {
sequence_parser::job j(parser);
// If this job is empty, then we're done
if (j.is_empty()) { return; }
for (size_t i=0; i < j->nb_filled; ++i) {
++readNum;
if (readNum % 100 == 0) {
auto tend = std::chrono::steady_clock::now();
auto sec = std::chrono::duration_cast<std::chrono::seconds>(tend-tstart);
auto nsec = sec.count();
auto rate = (nsec > 0) ? readNum / sec.count() : 0;
std::cerr << "processed " << readNum << " transcripts (" << rate << ") transcripts/s\r\r";
}
// we iterate over the entire read
const char* start = j->data[i].seq.c_str();
uint32_t readLen = j->data[i].seq.size();
const char* const end = start + readLen;
TranscriptFeatures tfeat{};
// reset all of the counts
numKmers = 0;
cmlen = 0;
kmer.polyA();
// the maximum number of kmers we'd have to store
uint32_t maxNumKmers = (readLen >= merLen) ? readLen - merLen + 1 : 0;
if (maxNumKmers == 0) { featQueue.push(tfeat); continue; }
// The transcript name
std::string fullHeader(j->data[i].header);
tfeat.name = fullHeader.substr(0, fullHeader.find(' '));
tfeat.length = readLen;
auto nfact = 1.0 / readLen;
// iterate over the read base-by-base
size_t offset{0};
size_t numChars{j->data[i].seq.size()};
while (offset < numChars) {
auto c = jellyfish::mer_dna::code(j->data[i].seq[offset]);
kmer.shift_left(c);
if (jellyfish::mer_dna::not_dna(c)) {
cmlen = 0;
++offset;
continue;
}
if (++cmlen >= merLen) {
size_t twomer = kmer.get_bits(0, 2*merLen);
tfeat.diNucleotides[twomer]++;
switch(c) {
case jellyfish::mer_dna::CODE_G:
case jellyfish::mer_dna::CODE_C:
tfeat.gcContent += nfact;
break;
}
}
++offset;
} // end while
char lastBase = j->data[i].seq.back();
auto c = jellyfish::mer_dna::code(lastBase);
switch(c) {
case jellyfish::mer_dna::CODE_G:
case jellyfish::mer_dna::CODE_C:
tfeat.gcContent += nfact;
break;
}
featQueue.push(tfeat);
} // end job
} // end while(true)
} // end lambda
));
} // actor loop
for (auto& t : threads) { t.join(); ++numComplete; }
return true;
}
