Make char int conversion functions free. (c7ba742a) · Commits · github_fork / Chromap

src/alignment.cc

+44 −62

Original line number	Diff line number	Diff line
		@@ -9,8 +9,7 @@ int GetLongestMatchLength(const char pattern, const char text,
		int max_match = 0;
		int tmp = 0;
		for (int i = 0; i < read_length; ++i) {
		if (SequenceBatch::CharToUint8(pattern[i]) ==
		SequenceBatch::CharToUint8(text[i])) {
		if (CharToUint8(pattern[i]) == CharToUint8(text[i])) {
		++tmp;
		} else if (tmp > max_match) {
		max_match = tmp;
		@@ -148,7 +147,7 @@ int BandedAlignPatternToText(int error_threshold, const char *pattern,
		int *mapping_end_position) {
		uint32_t Peq[5] = {0, 0, 0, 0, 0};
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base = SequenceBatch::CharToUint8(pattern[i]);
		uint8_t base = CharToUint8(pattern[i]);
		Peq[base] = Peq[base] \| (1 << i);
		}
		uint32_t highest_bit_in_band_mask = 1 << (2 * error_threshold);
		@@ -161,10 +160,9 @@ int BandedAlignPatternToText(int error_threshold, const char *pattern,
		uint32_t HP = 0;
		int num_errors_at_band_start_position = 0;
		for (int i = 0; i < read_length; i++) {
		uint8_t pattern_base =
		SequenceBatch::CharToUint8(pattern[i + 2 * error_threshold]);
		uint8_t pattern_base = CharToUint8(pattern[i + 2 * error_threshold]);
		Peq[pattern_base] = Peq[pattern_base] \| highest_bit_in_band_mask;
		X = Peq[SequenceBatch::CharToUint8(text[i])] \| VN;
		X = Peq[CharToUint8(text[i])] \| VN;
		D0 = ((VP + (X & VP)) ^ VP) \| X;
		HN = VP & D0;
		HP = VN \| ~(VP \| D0);
		@@ -207,7 +205,7 @@ int BandedAlignPatternToTextWithDropOff(int error_threshold,
		int *read_mapping_length) {
		uint32_t Peq[5] = {0, 0, 0, 0, 0};
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base = SequenceBatch::CharToUint8(pattern[i]);
		uint8_t base = CharToUint8(pattern[i]);
		Peq[base] = Peq[base] \| (1 << i);
		}
		uint32_t highest_bit_in_band_mask = 1 << (2 * error_threshold);
		@@ -225,10 +223,9 @@ int BandedAlignPatternToTextWithDropOff(int error_threshold,
		int fail_beginning = 0; // the alignment failed at the beginning part
		int prev_num_errors_at_band_start_position = 0;
		for (; i < read_length; i++) {
		uint8_t pattern_base =
		SequenceBatch::CharToUint8(pattern[i + 2 * error_threshold]);
		uint8_t pattern_base = CharToUint8(pattern[i + 2 * error_threshold]);
		Peq[pattern_base] = Peq[pattern_base] \| highest_bit_in_band_mask;
		X = Peq[SequenceBatch::CharToUint8(text[i])] \| VN;
		X = Peq[CharToUint8(text[i])] \| VN;
		D0 = ((VP + (X & VP)) ^ VP) \| X;
		HN = VP & D0;
		HP = VN \| ~(VP \| D0);
		@@ -297,8 +294,8 @@ int BandedAlignPatternToTextWithDropOffFrom3End(int error_threshold,
		int *read_mapping_length) {
		uint32_t Peq[5] = {0, 0, 0, 0, 0};
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base = SequenceBatch::CharToUint8(
		pattern[read_length + 2 * error_threshold - 1 - i]);
		uint8_t base =
		CharToUint8(pattern[read_length + 2 * error_threshold - 1 - i]);
		Peq[base] = Peq[base] \| (1 << i);
		}
		uint32_t highest_bit_in_band_mask = 1 << (2 * error_threshold);
		@@ -318,10 +315,9 @@ int BandedAlignPatternToTextWithDropOffFrom3End(int error_threshold,
		for (; i < read_length; i++) {
		// printf("%c %c %d\n", pattern[read_length - 1 - i], pattern[read_length -
		// 1 - i + error_threshold], text[read_length - 1 - i]);
		uint8_t pattern_base =
		SequenceBatch::CharToUint8(pattern[read_length - 1 - i]);
		uint8_t pattern_base = CharToUint8(pattern[read_length - 1 - i]);
		Peq[pattern_base] = Peq[pattern_base] \| highest_bit_in_band_mask;
		X = Peq[SequenceBatch::CharToUint8(text[read_length - 1 - i])] \| VN;
		X = Peq[CharToUint8(text[read_length - 1 - i])] \| VN;
		D0 = ((VP + (X & VP)) ^ VP) \| X;
		HN = VP & D0;
		HP = VN \| ~(VP \| D0);
		@@ -407,10 +403,10 @@ void BandedAlign4PatternsToText(int error_threshold, const char **patterns,
		Peq[ai] = _mm_setzero_si128();
		}
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base0 = SequenceBatch::CharToUint8(reference_sequence0[i]);
		uint8_t base1 = SequenceBatch::CharToUint8(reference_sequence1[i]);
		uint8_t base2 = SequenceBatch::CharToUint8(reference_sequence2[i]);
		uint8_t base3 = SequenceBatch::CharToUint8(reference_sequence3[i]);
		uint8_t base0 = CharToUint8(reference_sequence0[i]);
		uint8_t base1 = CharToUint8(reference_sequence1[i]);
		uint8_t base2 = CharToUint8(reference_sequence2[i]);
		uint8_t base3 = CharToUint8(reference_sequence3[i]);
		Peq[base0] = _mm_or_si128(highest_bit_in_band_mask_vpu0, Peq[base0]);
		Peq[base1] = _mm_or_si128(highest_bit_in_band_mask_vpu1, Peq[base1]);
		Peq[base2] = _mm_or_si128(highest_bit_in_band_mask_vpu2, Peq[base2]);
		@@ -432,19 +428,15 @@ void BandedAlign4PatternsToText(int error_threshold, const char **patterns,
		__m128i num_errors_at_band_start_position_vpu = _mm_setzero_si128();
		__m128i early_stop_threshold_vpu = _mm_set1_epi32(error_threshold * 3);
		for (int i = 0; i < read_length; i++) {
		uint8_t base0 = SequenceBatch::CharToUint8(
		reference_sequence0[i + 2 * error_threshold]);
		uint8_t base1 = SequenceBatch::CharToUint8(
		reference_sequence1[i + 2 * error_threshold]);
		uint8_t base2 = SequenceBatch::CharToUint8(
		reference_sequence2[i + 2 * error_threshold]);
		uint8_t base3 = SequenceBatch::CharToUint8(
		reference_sequence3[i + 2 * error_threshold]);
		uint8_t base0 = CharToUint8(reference_sequence0[i + 2 * error_threshold]);
		uint8_t base1 = CharToUint8(reference_sequence1[i + 2 * error_threshold]);
		uint8_t base2 = CharToUint8(reference_sequence2[i + 2 * error_threshold]);
		uint8_t base3 = CharToUint8(reference_sequence3[i + 2 * error_threshold]);
		Peq[base0] = _mm_or_si128(highest_bit_in_band_mask_vpu0, Peq[base0]);
		Peq[base1] = _mm_or_si128(highest_bit_in_band_mask_vpu1, Peq[base1]);
		Peq[base2] = _mm_or_si128(highest_bit_in_band_mask_vpu2, Peq[base2]);
		Peq[base3] = _mm_or_si128(highest_bit_in_band_mask_vpu3, Peq[base3]);
		X = _mm_or_si128(Peq[SequenceBatch::CharToUint8(text[i])], VN);
		X = _mm_or_si128(Peq[CharToUint8(text[i])], VN);
		D0 = _mm_and_si128(X, VP);
		D0 = _mm_add_epi32(D0, VP);
		D0 = _mm_xor_si128(D0, VP);
		@@ -548,14 +540,14 @@ void BandedAlign8PatternsToText(int error_threshold, const char **patterns,
		Peq[ai] = _mm_setzero_si128();
		}
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base0 = SequenceBatch::CharToUint8(reference_sequence0[i]);
		uint8_t base1 = SequenceBatch::CharToUint8(reference_sequence1[i]);
		uint8_t base2 = SequenceBatch::CharToUint8(reference_sequence2[i]);
		uint8_t base3 = SequenceBatch::CharToUint8(reference_sequence3[i]);
		uint8_t base4 = SequenceBatch::CharToUint8(reference_sequence4[i]);
		uint8_t base5 = SequenceBatch::CharToUint8(reference_sequence5[i]);
		uint8_t base6 = SequenceBatch::CharToUint8(reference_sequence6[i]);
		uint8_t base7 = SequenceBatch::CharToUint8(reference_sequence7[i]);
		uint8_t base0 = CharToUint8(reference_sequence0[i]);
		uint8_t base1 = CharToUint8(reference_sequence1[i]);
		uint8_t base2 = CharToUint8(reference_sequence2[i]);
		uint8_t base3 = CharToUint8(reference_sequence3[i]);
		uint8_t base4 = CharToUint8(reference_sequence4[i]);
		uint8_t base5 = CharToUint8(reference_sequence5[i]);
		uint8_t base6 = CharToUint8(reference_sequence6[i]);
		uint8_t base7 = CharToUint8(reference_sequence7[i]);
		Peq[base0] = _mm_or_si128(highest_bit_in_band_mask_vpu0, Peq[base0]);
		Peq[base1] = _mm_or_si128(highest_bit_in_band_mask_vpu1, Peq[base1]);
		Peq[base2] = _mm_or_si128(highest_bit_in_band_mask_vpu2, Peq[base2]);
		@@ -581,22 +573,14 @@ void BandedAlign8PatternsToText(int error_threshold, const char **patterns,
		__m128i num_errors_at_band_start_position_vpu = _mm_setzero_si128();
		__m128i early_stop_threshold_vpu = _mm_set1_epi16(error_threshold * 3);
		for (int i = 0; i < read_length; i++) {
		uint8_t base0 = SequenceBatch::CharToUint8(
		reference_sequence0[i + 2 * error_threshold]);
		uint8_t base1 = SequenceBatch::CharToUint8(
		reference_sequence1[i + 2 * error_threshold]);
		uint8_t base2 = SequenceBatch::CharToUint8(
		reference_sequence2[i + 2 * error_threshold]);
		uint8_t base3 = SequenceBatch::CharToUint8(
		reference_sequence3[i + 2 * error_threshold]);
		uint8_t base4 = SequenceBatch::CharToUint8(
		reference_sequence4[i + 2 * error_threshold]);
		uint8_t base5 = SequenceBatch::CharToUint8(
		reference_sequence5[i + 2 * error_threshold]);
		uint8_t base6 = SequenceBatch::CharToUint8(
		reference_sequence6[i + 2 * error_threshold]);
		uint8_t base7 = SequenceBatch::CharToUint8(
		reference_sequence7[i + 2 * error_threshold]);
		uint8_t base0 = CharToUint8(reference_sequence0[i + 2 * error_threshold]);
		uint8_t base1 = CharToUint8(reference_sequence1[i + 2 * error_threshold]);
		uint8_t base2 = CharToUint8(reference_sequence2[i + 2 * error_threshold]);
		uint8_t base3 = CharToUint8(reference_sequence3[i + 2 * error_threshold]);
		uint8_t base4 = CharToUint8(reference_sequence4[i + 2 * error_threshold]);
		uint8_t base5 = CharToUint8(reference_sequence5[i + 2 * error_threshold]);
		uint8_t base6 = CharToUint8(reference_sequence6[i + 2 * error_threshold]);
		uint8_t base7 = CharToUint8(reference_sequence7[i + 2 * error_threshold]);
		Peq[base0] = _mm_or_si128(highest_bit_in_band_mask_vpu0, Peq[base0]);
		Peq[base1] = _mm_or_si128(highest_bit_in_band_mask_vpu1, Peq[base1]);
		Peq[base2] = _mm_or_si128(highest_bit_in_band_mask_vpu2, Peq[base2]);
		@@ -605,7 +589,7 @@ void BandedAlign8PatternsToText(int error_threshold, const char **patterns,
		Peq[base5] = _mm_or_si128(highest_bit_in_band_mask_vpu5, Peq[base5]);
		Peq[base6] = _mm_or_si128(highest_bit_in_band_mask_vpu6, Peq[base6]);
		Peq[base7] = _mm_or_si128(highest_bit_in_band_mask_vpu7, Peq[base7]);
		X = _mm_or_si128(Peq[SequenceBatch::CharToUint8(text[i])], VN);
		X = _mm_or_si128(Peq[CharToUint8(text[i])], VN);
		D0 = _mm_and_si128(X, VP);
		D0 = _mm_add_epi16(D0, VP);
		D0 = _mm_xor_si128(D0, VP);
		@@ -694,8 +678,8 @@ void BandedTraceback(int error_threshold, int min_num_errors,
		// if not then there are gaps so that we have to traceback with edit distance.
		uint32_t Peq[5] = {0, 0, 0, 0, 0};
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base = SequenceBatch::CharToUint8(
		pattern[read_length - 1 + 2 * error_threshold - i]);
		uint8_t base =
		CharToUint8(pattern[read_length - 1 + 2 * error_threshold - i]);
		Peq[base] = Peq[base] \| (1 << i);
		}
		uint32_t highest_bit_in_band_mask = 1 << (2 * error_threshold);
		@@ -708,10 +692,9 @@ void BandedTraceback(int error_threshold, int min_num_errors,
		uint32_t HP = 0;
		int num_errors_at_band_start_position = 0;
		for (int i = 0; i < read_length; i++) {
		uint8_t pattern_base =
		SequenceBatch::CharToUint8(pattern[read_length - 1 - i]);
		uint8_t pattern_base = CharToUint8(pattern[read_length - 1 - i]);
		Peq[pattern_base] = Peq[pattern_base] \| highest_bit_in_band_mask;
		X = Peq[SequenceBatch::CharToUint8(text[read_length - 1 - i])] \| VN;
		X = Peq[CharToUint8(text[read_length - 1 - i])] \| VN;
		D0 = ((VP + (X & VP)) ^ VP) \| X;
		HN = VP & D0;
		HP = VN \| ~(VP \| D0);
		@@ -759,7 +742,7 @@ void BandedTracebackToEnd(int error_threshold, int min_num_errors,
		// if not then there are gaps so that we have to traceback with edit distance.
		uint32_t Peq[5] = {0, 0, 0, 0, 0};
		for (int i = 0; i < 2 * error_threshold; i++) {
		uint8_t base = SequenceBatch::CharToUint8(pattern[i]);
		uint8_t base = CharToUint8(pattern[i]);
		Peq[base] = Peq[base] \| (1 << i);
		}
		uint32_t highest_bit_in_band_mask = 1 << (2 * error_threshold);
		@@ -774,10 +757,9 @@ void BandedTracebackToEnd(int error_threshold, int min_num_errors,
		for (int i = 0; i < read_length; i++) {
		// printf("=>%d %d %c %c\n", i, num_errors_at_band_start_position, pattern[i
		// + 2 * error_threshold], text[i]) ;
		uint8_t pattern_base =
		SequenceBatch::CharToUint8(pattern[i + 2 * error_threshold]);
		uint8_t pattern_base = CharToUint8(pattern[i + 2 * error_threshold]);
		Peq[pattern_base] = Peq[pattern_base] \| highest_bit_in_band_mask;
		X = Peq[SequenceBatch::CharToUint8(text[i])] \| VN;
		X = Peq[CharToUint8(text[i])] \| VN;
		D0 = ((VP + (X & VP)) ^ VP) \| X;
		HN = VP & D0;
		HP = VN \| ~(VP \| D0);

src/barcode_translator.h

+2 −2

Original line number	Diff line number	Diff line
		@@ -91,8 +91,8 @@ class BarcodeTranslator {
		sequence.reserve(seed_length);
		uint64_t mask_ = 3;
		for (uint32_t i = 0; i < seed_length; ++i) {
		sequence.push_back(SequenceBatch::Uint8ToChar(
		(seed >> ((seed_length - 1 - i) * 2)) & mask_));
		sequence.push_back(
		Uint8ToChar((seed >> ((seed_length - 1 - i) * 2)) & mask_));
		}
		return sequence;
		}

src/chromap.cc

+15 −16

Original line number	Diff line number	Diff line
		@@ -526,9 +526,9 @@ bool Chromap::CorrectBarcodeAt(uint32_t barcode_index,
		adjusted_qual = adjusted_qual < 3 ? 3 : adjusted_qual;
		double score =
		pow(10.0, ((-adjusted_qual) / 10.0)) * barcode_abundance;
		corrected_barcodes_with_quals.emplace_back(BarcodeWithQual{
		barcode_length - 1 - i,
		SequenceBatch::Uint8ToChar(base_to_change1), 0, 0, score});
		corrected_barcodes_with_quals.emplace_back(
		BarcodeWithQual{barcode_length - 1 - i,
		Uint8ToChar(base_to_change1), 0, 0, score});
		// std::cerr << "1score: " << score << " pos1: " << barcode_length - 1
		// - i << " b1: " << base_to_change1 << " pos2: " << 0 << " b2: " <<
		// (char)0 << "\n";
		@@ -579,10 +579,9 @@ bool Chromap::CorrectBarcodeAt(uint32_t barcode_index,
		double score =
		pow(10.0, ((-adjusted_qual) / 10.0)) * barcode_abundance;
		corrected_barcodes_with_quals.emplace_back(BarcodeWithQual{
		barcode_length - 1 - i,
		SequenceBatch::Uint8ToChar(base_to_change1),
		barcode_length - 1 - j,
		SequenceBatch::Uint8ToChar(base_to_change2), score});
		barcode_length - 1 - i, Uint8ToChar(base_to_change1),
		barcode_length - 1 - j, Uint8ToChar(base_to_change2),
		score});
		// std::cerr << "2score: " << score << " pos1: " <<
		// barcode_length - 1 - i << " b1: " << base_to_change1 << "
		// pos2: " << barcode_length - 1 -j << " b2: " <<
		@@ -724,14 +723,14 @@ void Chromap::ParseReadFormat(const std::string &read_format) {
		;
		bool parse_success = true;
		if (read_format[i] == 'r' && read_format[i + 1] == '1') {
		parse_success = read1_format_.ParseEffectiveRange(
		read_format.c_str() + i, j - i);
		parse_success =
		read1_format_.ParseEffectiveRange(read_format.c_str() + i, j - i);
		} else if (read_format[i] == 'r' && read_format[i + 1] == '2') {
		parse_success = read2_format_.ParseEffectiveRange(
		read_format.c_str() + i, j - i);
		parse_success =
		read2_format_.ParseEffectiveRange(read_format.c_str() + i, j - i);
		} else if (read_format[i] == 'b' && read_format[i + 1] == 'c') {
		parse_success = barcode_format_.ParseEffectiveRange(
		read_format.c_str() + i, j - i);
		parse_success =
		barcode_format_.ParseEffectiveRange(read_format.c_str() + i, j - i);
		} else {
		parse_success = false;
		}

src/chromap.h

+7 −5

Original line number	Diff line number	Diff line
		@@ -312,7 +312,8 @@ void Chromap::MapSingleEndReads() {
		read_batch_for_loading, barcode_batch_for_loading);
		} // end of openmp loading task
		uint32_t history_update_threshold =
		mm_to_candidates_cache.GetUpdateThreshold(num_loaded_reads, num_reads_, false);
		mm_to_candidates_cache.GetUpdateThreshold(num_loaded_reads,
		num_reads_, false);
		// int grain_size = 10000;
		//#pragma omp taskloop grainsize(grain_size) //num_tasks(num_threads_* 50)
		#pragma omp taskloop num_tasks( \
		@@ -702,7 +703,8 @@ void Chromap::MapPairedEndReads() {

		int grain_size = 5000;
		uint32_t history_update_threshold =
		mm_to_candidates_cache.GetUpdateThreshold(num_loaded_pairs, num_reads_, true);
		mm_to_candidates_cache.GetUpdateThreshold(num_loaded_pairs,
		num_reads_, true);
		#pragma omp taskloop grainsize(grain_size)
		for (uint32_t pair_index = 0; pair_index < num_loaded_pairs;
		++pair_index) {

src/index.cc

+1 −1

Original line number	Diff line number	Diff line
		@@ -622,7 +622,7 @@ void Index::GenerateMinimizerSketch(
		int min_position = 0;

		for (uint32_t position = 0; position < sequence_length; ++position) {
		uint8_t current_base = SequenceBatch::CharToUint8(sequence[position]);
		uint8_t current_base = CharToUint8(sequence[position]);
		std::pair<uint64_t, uint64_t> current_seed = {UINT64_MAX, UINT64_MAX};
		if (current_base < 4) {
		// Not an ambiguous base.

Admin message