Clean up the code for generating custom rid rank. (4237c1c0) · Commits · github_fork / Chromap

src/chromap.cc

+25 −26

Original line number	Diff line number	Diff line
		@@ -717,46 +717,45 @@ void Chromap::ParseReadFormat(const std::string &read_format) {
		memcpy(barcode_format_, fields, sizeof(fields));
		}

		void Chromap::GenerateCustomizedRidRank(const std::string &rid_order_path,
		uint32_t num_reference_sequences,
		const SequenceBatch &reference,
		std::vector<int> &rid_rank) {
		rid_rank.resize(num_reference_sequences);
		void Chromap::GenerateCustomRidRanks(
		const std::string &custom_rid_order_file_path,
		uint32_t num_reference_sequences, const SequenceBatch &reference,
		std::vector<int> &rid_ranks) {
		for (uint32_t i = 0; i < num_reference_sequences; ++i) {
		rid_rank[i] = i;
		rid_ranks.emplace_back(i);
		}

		if (rid_order_path.length() == 0) {
		if (custom_rid_order_file_path.empty()) {
		return;
		}

		std::unordered_map<std::string, int> rname_to_rank;
		std::ifstream file_stream(rid_order_path);
		std::string line;
		uint32_t i = 0;
		while (getline(file_stream, line)) {
		rname_to_rank[line] = i;
		i += 1;
		std::unordered_map<std::string, int> ref_name_to_rank;
		std::ifstream custom_rid_order_file_stream(custom_rid_order_file_path);
		std::string ref_name;
		uint32_t ref_rank = 0;
		while (getline(custom_rid_order_file_stream, ref_name)) {
		ref_name_to_rank[ref_name] = ref_rank;
		ref_rank += 1;
		}
		file_stream.close();
		custom_rid_order_file_stream.close();

		// First put the chrosomes in the list provided by user.
		// First, rank the chromosomes in the custom order provided by users.
		for (uint32_t i = 0; i < num_reference_sequences; ++i) {
		std::string rname(reference.GetSequenceNameAt(i));
		if (rname_to_rank.find(rname) != rname_to_rank.end()) {
		rid_rank[i] = rname_to_rank[rname];
		std::string ref_name(reference.GetSequenceNameAt(i));
		if (ref_name_to_rank.find(ref_name) != ref_name_to_rank.end()) {
		rid_ranks[i] = ref_name_to_rank[ref_name];
		} else {
		rid_rank[i] = -1;
		rid_ranks[i] = -1;
		}
		}

		// There might be some rank without any rid associated with. This helps if
		// cutstom list contains rid not in the reference.
		uint32_t k = rname_to_rank.size();
		// Put the remaining chrosomes
		// There might be some rids without any custom order. We just order them based
		// on their original order in the reference file.
		uint32_t k = ref_name_to_rank.size();
		// Rank the remaining chromosomes.
		for (uint32_t i = 0; i < num_reference_sequences; ++i) {
		if (rid_rank[i] == -1) {
		rid_rank[i] = k;
		if (rid_ranks[i] == -1) {
		rid_ranks[i] = k;
		++k;
		}
		}

src/chromap.h

+22 −18

Original line number	Diff line number	Diff line
		@@ -109,10 +109,16 @@ class Chromap {

		void ParseReadFormat(const std::string &read_format);

		void GenerateCustomizedRidRank(const std::string &rid_order_path,
		// User custom rid order file contains a column of reference sequence names
		// and there is one name on each row. The reference sequence name on the ith
		// row means the rank of this sequence is i. This function loads the custom
		// rid order file and generates a mapping from the original rids to their
		// custom ranks, e.g., rid_ranks[i] is the custom rank of the ith rid in the
		// reference.
		void GenerateCustomRidRanks(const std::string &custom_rid_order_file_path,
		uint32_t num_reference_sequences,
		const SequenceBatch &reference,
		std::vector<int> &rid_rank);
		std::vector<int> &rid_ranks);

		// TODO: generate reranked candidates directly.
		void RerankCandidatesRid(std::vector<Candidate> &candidates);
		@@ -167,7 +173,7 @@ void Chromap::MapSingleEndReads() {
		reference.InitializeLoading(mapping_parameters_.reference_file_path);
		uint32_t num_reference_sequences = reference.LoadAllSequences();
		if (mapping_parameters_.custom_rid_order_path.length() > 0) {
		GenerateCustomizedRidRank(mapping_parameters_.custom_rid_order_path,
		GenerateCustomRidRanks(mapping_parameters_.custom_rid_order_path,
		num_reference_sequences, reference,
		custom_rid_rank_);
		reference.ReorderSequences(custom_rid_rank_);
		@@ -217,9 +223,8 @@ void Chromap::MapSingleEndReads() {
		MappingGenerator<MappingRecord> mapping_generator(mapping_parameters_,
		pairs_custom_rid_rank_);

		MappingWriter<MappingRecord> mapping_writer(mapping_parameters_,
		barcode_length_,
		pairs_custom_rid_rank_);
		MappingWriter<MappingRecord> mapping_writer(
		mapping_parameters_, barcode_length_, pairs_custom_rid_rank_);

		mapping_writer.OutputHeader(num_reference_sequences, reference);

		@@ -537,13 +542,13 @@ void Chromap::MapPairedEndReads() {
		reference.InitializeLoading(mapping_parameters_.reference_file_path);
		uint32_t num_reference_sequences = reference.LoadAllSequences();
		if (mapping_parameters_.custom_rid_order_path.length() > 0) {
		GenerateCustomizedRidRank(mapping_parameters_.custom_rid_order_path,
		GenerateCustomRidRanks(mapping_parameters_.custom_rid_order_path,
		num_reference_sequences, reference,
		custom_rid_rank_);
		reference.ReorderSequences(custom_rid_rank_);
		}
		if (mapping_parameters_.mapping_output_format == MAPPINGFORMAT_PAIRS) {
		GenerateCustomizedRidRank(mapping_parameters_.pairs_custom_rid_order_path,
		GenerateCustomRidRanks(mapping_parameters_.pairs_custom_rid_order_path,
		num_reference_sequences, reference,
		pairs_custom_rid_rank_);
		}
		@@ -608,8 +613,7 @@ void Chromap::MapPairedEndReads() {
		pairs_custom_rid_rank_);

		MappingWriter<MappingRecord> mapping_writer(
		mapping_parameters_, barcode_length_,
		pairs_custom_rid_rank_);
		mapping_parameters_, barcode_length_, pairs_custom_rid_rank_);
		mapping_writer.OutputHeader(num_reference_sequences, reference);

		uint32_t num_mappings_in_mem = 0;

Admin message