Check whether the barcode whitelist matches the data and add the option to check this step. (eeee74c4) · Commits · github_fork / Chromap

src/chromap.cc

+17 −3

Original line number	Diff line number	Diff line
		@@ -749,6 +749,13 @@ void Chromap<MappingRecord>::ComputeBarcodeAbundance(
		++num_sample_barcodes_;
		}
		}

		if (!skip_barcode_check_
		&& num_sample_barcodes_ * 20 < num_loaded_barcodes) {
		// Since num_loaded_pairs is a constant, this if is actuaclly only effective in the first iteration
		Chromap<>::ExitWithMessage("Less than 5\% barcodes can be found or corrected based on the barcode whitelist.\nPlease check whether the barcode whitelist matches the data, e.g. length, reverse-complement. If this is a false positive warning, please run Chromap with the option --skip-barcode-check.");
		}

		if (num_sample_barcodes_ >= max_num_sample_barcodes) {
		break;
		}
		@@ -759,6 +766,7 @@ void Chromap<MappingRecord>::ComputeBarcodeAbundance(
		break;
		}
		}

		std::cerr << "Compute barcode abundance using " << num_sample_barcodes_
		<< " in " << Chromap<>::GetRealTime() - real_start_time << "s.\n";
		}
		@@ -5968,8 +5976,8 @@ void ChromapDriver::ParseArgsAndRun(int argc, char *argv[]) {
		//("p,matrix-output-prefix", "Prefix of matrix output files",
		// cxxopts::value<std::string>(), "FILE")
		("output-mappings-not-in-whitelist",
		"Output mappings with barcode not in the whitelist")(
		"chr-order", "custom chromsome order", cxxopts::value<std::string>(),
		"Output mappings with barcode not in the whitelist")
		("chr-order", "custom chromsome order", cxxopts::value<std::string>(),
		"FILE")("BED", "Output mappings in BED/BEDPE format")(
		"TagAlign", "Output mappings in TagAlign/PairedTagAlign format")(
		"SAM", "Output mappings in SAM format")(
		@@ -5999,7 +6007,9 @@ void ChromapDriver::ParseArgsAndRun(int argc, char *argv[]) {
		"drop-repetitive-reads",
		"Drop reads with too many best mappings [500000]", cxxopts::value<int>(),
		"INT")("allocate-multi-mappings", "Allocate multi-mappings")(
		"PAF", "Output mappings in PAF format (only for test)");
		"PAF", "Output mappings in PAF format (only for test)")
		("skip-barcode-check", "Do not check whether too few barcodes are in the whitelist")
		;

		auto result = options.parse(argc, argv);
		if (result.count("h")) {
		@@ -6277,6 +6287,10 @@ void ChromapDriver::ParseArgsAndRun(int argc, char *argv[]) {
		result["pairs-natural-chr-order"].as<std::string>();
		}

		if (result.count("skip-barcode-check")) {
		mapping_parameters.skip_barcode_check = true;
		}

		// std::cerr << "Parameters: error threshold: " << error_threshold << ",
		// match score: " << match_score << ", mismatch_penalty: " <<
		// mismatch_penalty << ", gap open penalties for deletions and insertions: "

src/chromap.h

+6 −2

Original line number	Diff line number	Diff line
		@@ -16,7 +16,7 @@
		#include "output_tools.h"
		#include "sequence_batch.h"

		#define CHROMAP_VERSION "0.1.3-r256"
		#define CHROMAP_VERSION "0.1.3-r257"

		namespace chromap {
		struct uint128_t {
		@@ -117,6 +117,7 @@ struct MappingParameters {
		std::string matrix_output_prefix;
		std::string custom_rid_order_path;
		std::string pairs_custom_rid_order_path;
		bool skip_barcode_check = false;
		};

		#define SortMappingWithoutBarcode(m) \
		@@ -202,7 +203,8 @@ class Chromap {
		matrix_output_prefix_(mapping_parameters.matrix_output_prefix),
		custom_rid_order_path_(mapping_parameters.custom_rid_order_path),
		pairs_custom_rid_order_path_(
		mapping_parameters.pairs_custom_rid_order_path) {
		mapping_parameters.pairs_custom_rid_order_path),
		skip_barcode_check_(mapping_parameters.skip_barcode_check){
		barcode_lookup_table_ = kh_init(k64_seq);
		barcode_whitelist_lookup_table_ = kh_init(k64_seq);
		barcode_histogram_ = kh_init(k64_seq);
		@@ -681,6 +683,8 @@ class Chromap {
		uint32_t barcode_length_ = 0;
		khash_t(k64_seq) * barcode_histogram_;
		khash_t(k64_seq) * barcode_index_table_;
		bool skip_barcode_check_ = false ;

		// For peak calling
		std::vector<std::vector<uint16_t> > pileup_on_diff_ref_seqs_;
		std::vector<std::vector<Peak> > peaks_on_diff_ref_seqs_;

Admin message