Function loadReads

RResolver/BloomFilters.cpp:139–209

}

static void
loadReads(const std::vector<std::string>& readFilepaths, int r)
{
	ReadSize::readsSampleSize = 0;
	ReadSize::current.sampleCount = 0;
	int i = 0;
	for (auto& b : ReadSize::readSizes) {
		b.sampleCount = 0;
		i++;
	}

	for (const auto& path : readFilepaths) {
		uint64_t currentReadCount = 0;
		uint64_t totalReadCount = 0;
		size_t readSizesNum = ReadSize::readSizes.size();
		uint64_t *readSizeCounts = new uint64_t[readSizesNum];
		for (size_t i = 0; i < readSizesNum; i++) {
			readSizeCounts[i] = 0;
		}

		assert(!path.empty());
		if (opt::verbose) {
			std::cerr << "Loading reads from `" << path << "'..." << std::endl;;
		}

		btllib::SeqReader reader(path, btllib::SeqReader::Flag::SHORT_MODE);
#pragma omp parallel reduction(+:currentReadCount,totalReadCount,readSizeCounts[:readSizesNum])
		for (btllib::SeqReader::Record record; (record = reader.read());) {
			totalReadCount++;
			int i = 0;
			for (const auto& b : ReadSize::readSizes) {
				if (b.sizeAndMergedSizes.find(int(record.seq.size())) != b.sizeAndMergedSizes.end()) {
					readSizeCounts[i]++;
					break;
				}
				i++;
			}
			if (ReadSize::current.sizeAndMergedSizes.find(int(record.seq.size())) == ReadSize::current.sizeAndMergedSizes.end()) {
				continue;
			}
			currentReadCount++;

			std::string seq = record.seq.substr(0, r + opt::extract - 1);
			if (seq.size() >= g_vanillaBloom->get_k()) {
				g_vanillaBloom->insert(seq);
				if (opt::errorCorrection) {
					g_spacedSeedsBloom->insert(seq);
				}
			}
		}

		ReadSize::readsSampleSize += totalReadCount;
		ReadSize::current.sampleCount += currentReadCount;
		int i = 0;
		for (auto& b : ReadSize::readSizes) {
			b.sampleCount += readSizeCounts[i];
			i++;
		}