folly::Future<int32_t> service_with_special_namesSvIf::future_annotations() {
return apache::thrift::detail::si::future([&] { return annotations(); });
}
// perform classification in separate threads
void Classifier::runThread(FastaReader &reader)
{
DnaSequence seq;
boost::mt19937 randGen(0);
RandomGen generator(randGen);
while(seq = reader.readSequence())
{
// reproducible randomness
randGen.seed(reader.numRead());
// get a query sequence and convert to kmers
KmerSequence fwdSeq = kmerizer_.kmerize(seq);
// and the reverse complement
KmerSequence revSeq = kmerizer_.revComp(fwdSeq);
// search against database using both forward and reverse sequences
searchHit fwdHit = referenceData_.search(fwdSeq);
searchHit revHit = referenceData_.search(revSeq);
// use the direction which gave the highest scoring hit
KmerSequence &querySeq = fwdHit.score > revHit.score ? fwdSeq : revSeq;
searchHit &hit = fwdHit.score > revHit.score ? fwdHit : revHit;
std::vector<float> bootstraps(referenceData_.numLevels(), 0.f);
std::vector<std::string> annotations(referenceData_.numLevels(), "AMBIGUOUS");
// make each level annotations into a list of uniques
for (unsigned int i=0; i< hit.annotationIds.size(); i++)
{
std::sort(hit.annotationIds[i].begin(), hit.annotationIds[i].end());
std::vector<unsigned int>::iterator it;
it = std::unique(hit.annotationIds[i].begin(), hit.annotationIds[i].end());
hit.annotationIds[i].resize(std::distance(hit.annotationIds[i].begin(), it));
if (hit.annotationIds[i].size() == 1)
annotations[i] = referenceData_.annotationFromId(*hit.annotationIds[i].begin());
}
// bootstrap
if(numBootstrap_ > 0)
bootstraps = getBootstrap(querySeq, generator, hit);
// output
std::stringstream s;
s << std::setprecision(2) << std::fixed;
s << querySeq.header.substr(0, querySeq.header.find("\t")) << "\t";
s << hit.score << "\t";
//for(unsigned int i=0; i<bootstraps.size(); i++)
unsigned int i=bootstraps.size();
while(i--)
{
s << annotations[i] << "\t";
s << bootstraps[i];
if(i>0)
{
s << "\t";
}
else
{
// dump ambiguous species
if (hit.annotationIds[i].size() > 1 && outputAmbiguous_)
{
s << "\t";
for(std::vector<unsigned int>::iterator it = hit.annotationIds[i].begin(); it != hit.annotationIds[i].end(); ++it)
{
if (it != hit.annotationIds[i].begin())
s << ",";
s << referenceData_.annotationFromId(*it);
}
}
s << std::endl;
}
}
boost::mutex::scoped_lock lock(mutex_);
std::cout << s.str();
}
}
