Function buildFilters

RResolver/BloomFilters.cpp:211–297

}

void
buildFilters(
    const std::vector<std::string>& readFilepaths,
    const int r,
    const size_t bloomBytesTotal)
{
	assert(bloomBytesTotal > 0);
	try {
		if (opt::verbose) {
			std::cerr << "Building Bloom filter(s) for r value " << r << '\n';
		}

		delete g_vanillaBloom;
		delete g_spacedSeedsBloom;

		size_t bloomBytesVanilla = size_t(bloomBytesTotal);
		size_t bloomBytesSpacedSeeds = 0;

		if (opt::errorCorrection) {
			double vanillaRatio =
			    VANILLA_TO_SEEDS_MEM_RATIO * double(HASH_NUM) /
			    (double(HASH_NUM) + double(SPACED_SEEDS_COUNT * SPACED_SEEDS_HASHES_PER_SEED));
			bloomBytesVanilla = size_t(vanillaRatio * bloomBytesTotal);
			bloomBytesSpacedSeeds = bloomBytesTotal - bloomBytesVanilla;
		}

		if (opt::verbose > 1) {
			if (opt::errorCorrection) {
				std::cerr << "Total Bloom filter memory = " << bytesToSI(bloomBytesTotal) << '\n';
				std::cerr << "Vanilla Bloom filter memory = " << bytesToSI(bloomBytesVanilla)
				          << '\n';
				std::cerr << "Spaced seeds Bloom filter memory = "
				          << bytesToSI(bloomBytesSpacedSeeds) << '\n';
			} else {
				std::cerr << "Vanilla Bloom filter memory = " << bytesToSI(bloomBytesVanilla)
				          << '\n';
			}
		}

		g_vanillaBloom = new btllib::KmerBloomFilter(bloomBytesVanilla, HASH_NUM, r);
		if (opt::errorCorrection) {
			const auto patterns =
				generateSpacedSeedsPatterns(SPACED_SEEDS_COUNT, r, SPACED_SEEDS_MISSES);
			for (const auto& pattern : patterns) {
				(void)pattern;
				assert(pattern.size() == size_t(r));
			}
			if (SPACED_SEEDS_QC) {
				QCSpacedSeedsPatterns(patterns);
			}
			g_spacedSeedsBloom = new btllib::SeedBloomFilter(bloomBytesSpacedSeeds, r, patterns, SPACED_SEEDS_HASHES_PER_SEED);
		}

		loadReads(readFilepaths, r);

		if (opt::verbose > 1) {
			const auto vanillaFPR = g_vanillaBloom->get_fpr();