r1285: better --spsc support

Author: lh3

index.c

@@ -967,7 +967,7 @@ typedef struct mm_idx_spsc_s {
 	uint64_t *a; // pos<<56 | score<<1 | acceptor
 } mm_idx_spsc_t;
 
-int32_t mm_idx_spsc_read(mm_idx_t *idx, const char *fn, int32_t max_sc)
+int32_t mm_idx_spsc_read2(mm_idx_t *idx, const char *fn, int32_t max_sc, float scale)
 {
 	gzFile fp;
 	kstring_t str = {0,0,0};
@@ -1007,6 +1007,8 @@ int32_t mm_idx_spsc_read(mm_idx_t *idx, const char *fn, int32_t max_sc)
 			}
 		}
 		if (i < 4) continue; // not enough fields
+		if (scale > 0.0f && scale < 1.0f)
+			score = score > 0.0f? (int)(score * scale + .499) : (int)(score * scale - .499);
 		if (score > max_sc) score = max_sc;
 		if (score < -max_sc) score = -max_sc;
 		cid = mm_idx_name2id(idx, name);
@@ -1030,6 +1032,11 @@ int32_t mm_idx_spsc_read(mm_idx_t *idx, const char *fn, int32_t max_sc)
 	return 0;
 }
 
+int32_t mm_idx_spsc_read(mm_idx_t *idx, const char *fn, int32_t max_sc)
+{
+	return mm_idx_spsc_read2(idx, fn, max_sc, 1.0f);
+}
+
 static int32_t mm_idx_find_intv(int32_t n, const uint64_t *a, int64_t x)
 {
 	int32_t s = 0, e = n;

main.c

@@ -85,6 +85,8 @@ static ko_longopt_t long_options[] = {
 	{ "jump-min-match", ko_required_argument, 360 },
 	{ "write-junc",     ko_no_argument,       361 },
 	{ "pass1",          ko_required_argument, 362 },
+	{ "spsc-scale",     ko_required_argument, 363 },
+	{ "spsc0",          ko_required_argument, 364 },
 	{ "dbg-seed-occ",   ko_no_argument,       501 },
 	{ "help",           ko_no_argument,       'h' },
 	{ "max-intron-len", ko_required_argument, 'G' },
@@ -134,6 +136,7 @@ int main(int argc, char *argv[])
 	mm_mapopt_t opt;
 	mm_idxopt_t ipt;
 	int i, c, n_threads = 3, n_parts, old_best_n = -1;
+	float spsc_scale = 0.7f;
 	char *fnw = 0, *rg = 0, *fn_bed_junc = 0, *fn_bed_jump = 0, *fn_bed_pass1 = 0, *fn_spsc = 0, *s, *alt_list = 0;
 	FILE *fp_help = stderr;
 	mm_idx_reader_t *idx_rdr;
@@ -240,7 +243,6 @@ int main(int argc, char *argv[])
 		else if (c == 338) opt.max_qlen = mm_parse_num(o.arg); // --max-qlen
 		else if (c == 340) fn_bed_junc = o.arg; // --junc-bed
 		else if (c == 341) opt.junc_bonus = atoi(o.arg); // --junc-bonus
-		else if (c == 358) opt.junc_pen = atoi(o.arg); // --junc-pen
 		else if (c == 342) opt.flag |= MM_F_SAM_HIT_ONLY; // --sam-hit-only
 		else if (c == 343) opt.chain_gap_scale = atof(o.arg); // --chain-gap-scale
 		else if (c == 351) opt.chain_skip_scale = atof(o.arg); // --chain-skip-scale
@@ -260,6 +262,8 @@ int main(int argc, char *argv[])
 		else if (c == 361) opt.flag |= MM_F_OUT_JUNC | MM_F_CIGAR; // --write-junc
 		else if (c == 362) fn_bed_pass1 = o.arg; // --jump-pass1
 		else if (c == 501) mm_dbg_flag |= MM_DBG_SEED_FREQ; // --dbg-seed-occ
+		else if (c == 363) spsc_scale = atof(o.arg); // --spsc-scale
+		else if (c == 358 || c == 364) opt.junc_pen = atoi(o.arg); // --junc-pen or --spsc0
 		else if (c == 330) {
 			fprintf(stderr, "[WARNING] \033[1;31m --lj-min-ratio has been deprecated.\033[0m\n");
 		} else if (c == 313) { // --sr
@@ -476,7 +480,7 @@ int main(int argc, char *argv[])
 				fprintf(stderr, "[WARNING] failed to load the pass-1 jump BED file\n");
 		}
 		if (fn_spsc) {
-			mm_idx_spsc_read(mi, fn_spsc, mm_max_spsc_bonus(&opt));
+			mm_idx_spsc_read2(mi, fn_spsc, mm_max_spsc_bonus(&opt), spsc_scale);
 			if (mi->spsc == 0 && mm_verbose >= 2)
 				fprintf(stderr, "[WARNING] failed to load the splice score file\n");
 		}

minimap.h

@@ -5,7 +5,7 @@
 #include <stdio.h>
 #include <sys/types.h>
 
-#define MM_VERSION "2.29-r1283"
+#define MM_VERSION "2.29-r1285-dirty"
 
 #define MM_F_NO_DIAG       (0x001LL) // no exact diagonal hit
 #define MM_F_NO_DUAL       (0x002LL) // skip pairs where query name is lexicographically larger than target name
@@ -163,7 +163,8 @@ typedef struct {
 	int transition; // transition mismatch score (A:G, C:T)
 	int sc_ambi; // score when one or both bases are "N"
 	int noncan;      // cost of non-canonical splicing sites
-	int junc_bonus, junc_pen;
+	int junc_bonus;  // bonus for a splice site in annotation
+	int junc_pen;    // penalty for GT- or -AG not scored in --spsc
 	int zdrop, zdrop_inv;   // break alignment if alignment score drops too fast along the diagonal
 	int end_bonus;
 	int min_dp_max;  // drop an alignment if the score of the max scoring segment is below this threshold
@@ -420,6 +421,7 @@ int mm_idx_bed_junc(const mm_idx_t *mi, int32_t ctg, int32_t st, int32_t en, uin
 
 int mm_max_spsc_bonus(const mm_mapopt_t *mo);
 int32_t mm_idx_spsc_read(mm_idx_t *idx, const char *fn, int32_t max_sc);
+int32_t mm_idx_spsc_read2(mm_idx_t *idx, const char *fn, int32_t max_sc, float scale);
 int64_t mm_idx_spsc_get(const mm_idx_t *db, int32_t cid, int64_t st0, int64_t en0, int32_t rev, uint8_t *sc);
 
 // deprecated APIs for backward compatibility

minimap2.1

@@ -443,14 +443,23 @@ line corresponds to a donor site and `A' for an acceptor site.
 A positive score suggests the junction is preferred and a negative score
 suggests the junction is not preferred.
 .TP
-.BR --junc-pen \ INT
-Penalty for a position not in FILE specified by
+.BR --spsc0 \ INT
+Penalty for positions not in
+.I FILE
+specified by
 .B --spsc
 [5]. Effective with
 .B --spsc
 but not
 .BR --junc-bed .
 .TP
+.BR --spsc-scale \ FLOAT
+Scale splice scores in
+.B --spsc
+by
+.IR FLOAT
+rounded to the nearest integer [0.7].
+.TP
 .BR --junc-bed \ FILE
 Junctions to prefer during base alignment [].
 Same format as