Tune parameters (df39e2a9) · Commits · github_fork / Chromap

src/chromap.cc

+3 −2

Original line number	Diff line number	Diff line
		@@ -753,8 +753,9 @@ void Chromap<MappingRecord>::MapPairedEndReads() {
		{
		num_loaded_pairs_for_loading = LoadPairedEndReadsWithBarcodes(&read_batch1_for_loading, &read_batch2_for_loading, &barcode_batch_for_loading);
		} // end of openmp loading task
		int grain_size = 10000;
		#pragma omp taskloop grainsize(grain_size) //num_tasks(num_threads_* 50)
		//int grain_size = 5000;
		//#pragma omp taskloop grainsize(grain_size) //num_tasks(num_threads_* 50)
		#pragma omp taskloop num_tasks(num_threads_* num_threads_)
		for (uint32_t pair_index = 0; pair_index < num_loaded_pairs; ++pair_index) {
		read_batch1.PrepareNegativeSequenceAt(pair_index);
		read_batch2.PrepareNegativeSequenceAt(pair_index);

src/chromap.h

+2 −2

Original line number	Diff line number	Diff line
		@@ -218,7 +218,7 @@ class Chromap {
		bool output_mapping_in_BED_;
		bool output_mapping_in_TagAlign_;
		bool output_mapping_in_PAF_;
		uint32_t read_batch_size_ = 1000000; // default batch size, # reads for single-end reads, # read pairs for paired-end reads
		uint32_t read_batch_size_ = 500000; // default batch size, # reads for single-end reads, # read pairs for paired-end reads
		bool low_memory_mode_;
		bool cell_by_bin_;
		int bin_size_;
		@@ -258,7 +258,7 @@ class Chromap {
		uint64_t num_reads_ = 0;
		uint64_t num_duplicated_reads_ = 0; // # identical reads
		// For barcode stats
		uint64_t initial_num_sample_barcodes_ = 100000000;
		uint64_t initial_num_sample_barcodes_ = 50000000;
		uint64_t num_sample_barcodes_ = 0;
		uint64_t num_barcode_in_whitelist_ = 0;
		uint64_t num_corrected_barcode_ = 0;

src/index.cc

+14 −11

Original line number	Diff line number	Diff line
		@@ -492,12 +492,11 @@ void Index::GenerateCandidatesFromRepetitiveReadWithMateInfo(int error_threshold
		++best_candidate_num;
		}
		}
		if (best_candidate_num >= 500
		\|\| (max_count < min_num_seeds_required_for_mapping_ && best_candidate_num >= 5)
		\|\| (max_count == min_num_seeds_required_for_mapping_ && best_candidate_num >=50))
		if (best_candidate_num >= 300 \|\| (max_count <= min_num_seeds_required_for_mapping_ && best_candidate_num >= 200)) {
		return;
		}
		std::vector<std::pair<uint64_t, uint64_t> > boundaries;
		boundaries.reserve(500);
		boundaries.reserve(300);
		for (uint32_t ci = 0; ci < mate_candidates_size; ++ci) {
		if (mate_candidates->at(ci).count == max_count) {
		std::pair<uint64_t, uint64_t> r;
		@@ -552,7 +551,7 @@ void Index::GenerateCandidatesFromRepetitiveReadWithMateInfo(int error_threshold
		} else {
		uint32_t offset = value >> 32;
		uint32_t num_occurrences = value;
		uint32_t prev_l = 0;
		int32_t prev_l = 0;
		for (uint32_t bi = 0; bi < boundary_size; ++bi) {
		// use binary search to locate the coordinate near mate position
		int32_t l = prev_l, m = 0, r = num_occurrences - 1;
		@@ -619,14 +618,17 @@ void Index::GenerateCandidates(int error_threshold, const std::vector<std::pair<
		int repetitive_seed_count = CollectCandidates(max_seed_frequencies_[0], max_seed_frequencies_[0], minimizers, repetitive_seed_length, positive_hits, negative_hits, false);
		//std::cerr << "rep seed: " << repetitive_seed_count << "\n";
		//if ((repetitive_seed_count > (int)minimizers.size() / 2 && minimizers.size() >= 10)) {
		bool checkmore = false;
		bool use_high_frequency_minimizers = false;
		if (positive_hits->size() + negative_hits->size() == 0) {
		positive_hits->clear();
		negative_hits->clear();
		*repetitive_seed_length = 0;
		repetitive_seed_count = CollectCandidates(max_seed_frequencies_[1], max_seed_frequencies_[0], minimizers, repetitive_seed_length, positive_hits, negative_hits, true);
		//recollect = false;
		checkmore = true;
		use_high_frequency_minimizers = true;
		if (positive_hits->size() == 0 \|\| negative_hits->size() == 0) {
		use_high_frequency_minimizers = false;
		}
		}
		//if ((positive_candidates->size() == 0 \|\| negative_candidates->size() == 0) && recollect) {
		////if (positive_candidates->size() + negative_candidates->size() == 0 && recollect) {
		@@ -658,8 +660,9 @@ void Index::GenerateCandidates(int error_threshold, const std::vector<std::pair<
		int num_required_seeds = minimizers.size() - repetitive_seed_count;
		num_required_seeds = num_required_seeds > 1 ? num_required_seeds : 1;
		num_required_seeds = num_required_seeds > min_num_seeds_required_for_mapping_ ? min_num_seeds_required_for_mapping_ : num_required_seeds;
		if (checkmore)
		if (use_high_frequency_minimizers) {
		num_required_seeds = min_num_seeds_required_for_mapping_;
		}
		GenerateCandidatesOnOneDirection(error_threshold, num_required_seeds, positive_hits, positive_candidates);
		GenerateCandidatesOnOneDirection(error_threshold, num_required_seeds, negative_hits, negative_candidates);
		//fprintf(stderr, "p+n: %d\n", positive_candidates->size() + negative_candidates->size()) ;

Admin message