Merge pull request #503 from chhylp123/hifiasm_dev_debug

disable multiple assertions

CommandLines.cpp

@@ -61,6 +61,7 @@ static ko_longopt_t long_options[] = {
     { "path-max",     ko_required_argument, 346},
     { "path-min",     ko_required_argument, 347},
     { "trio-dual",     ko_no_argument, 348},
+    { "ul-cut",     ko_required_argument, 349},
     // { "path-round",     ko_required_argument, 348},
 	{ 0, 0, 0 }
 };
@@ -174,6 +175,8 @@ void Print_H(hifiasm_opt_t* asm_opt)
     fprintf(stderr, "    --path-min   FLOAT\n");
     fprintf(stderr, "                 min path drop ratio [%.2g]; higher number may make the assembly cleaner\n", asm_opt->min_path_drop_rate);
     fprintf(stderr, "                 but may lead to more misassemblies\n");
+    fprintf(stderr, "    --ul-cut     INT\n");
+    fprintf(stderr, "                 filter out <INT UL reads during the UL assembly [%d]\n", asm_opt->ul_min_base);
     // fprintf(stderr, "    --low-het    enable it for genomes with very low het heterozygosity rate (<0.0001%%)\n");
 
     fprintf(stderr, "Example: ./hifiasm -o NA12878.asm -t 32 NA12878.fq.gz\n");
@@ -290,6 +293,7 @@ void init_opt(hifiasm_opt_t* asm_opt)
     asm_opt->hifi_pst_join = 1;
     asm_opt->ul_pst_join = 1;
     asm_opt->trio_cov_het_ovlp = -1;
+    asm_opt->ul_min_base = 0;
 }
 
 void destory_enzyme(enzyme* f)
@@ -835,6 +839,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
         else if (c == 346) asm_opt->max_path_drop_rate = atof(opt.arg);
         else if (c == 347) asm_opt->min_path_drop_rate = atof(opt.arg);
         else if (c == 348) asm_opt->trio_cov_het_ovlp = 1;
+        else if (c == 349) asm_opt->ul_min_base = atol(opt.arg);
         else if (c == 'l') {   ///0: disable purge_dup; 1: purge containment; 2: purge overlap
             asm_opt->purge_level_primary = asm_opt->purge_level_trio = atoi(opt.arg);
         }

CommandLines.h

@@ -5,7 +5,7 @@
 #include <pthread.h>
 #include <stdint.h>
 
-#define HA_VERSION "0.19.5-r593"
+#define HA_VERSION "0.19.6-r595"
 
 #define VERBOSE 0
 
@@ -146,6 +146,7 @@ typedef struct {
     int32_t integer_correct_round;
     uint8_t dbg_ovec_cal;
     uint8_t hifi_pst_join, ul_pst_join;
+    uint32_t ul_min_base;
 } hifiasm_opt_t;
 
 extern hifiasm_opt_t asm_opt;

inter.cpp

@@ -12692,7 +12692,7 @@ static void *worker_ul_scall_pipeline(void *data, int step, void *in) // callbac
 		s->opt = p->opt; s->uu = p->uu; s->uopt = p->uopt; s->rg = p->rg;
         while ((ret = kseq_read(p->ks)) >= 0) 
         {
-            if (p->ks->seq.l < (uint64_t)p->opt->k) continue;
+            if ((p->ks->seq.l < (uint64_t)p->opt->k) || (p->ks->seq.l < asm_opt.ul_min_base)) continue;
             if (s->n == s->m) {
                 s->m = s->m < 16? 16 : s->m + (s->n>>1);
                 REALLOC(s->len, s->m);
@@ -12823,7 +12823,7 @@ static void *worker_ul_rescall_pipeline(void *data, int step, void *in) // callb
 		s->opt = p->opt; s->uu = p->uu; s->uopt = p->uopt; s->rg = p->rg; s->mm = p->mm;
         while ((ret = kseq_read(p->ks)) >= 0) 
         {
-            if (p->ks->seq.l < (uint64_t)p->opt->k) continue;
+			if ((p->ks->seq.l < (uint64_t)p->opt->k) || (p->ks->seq.l < asm_opt.ul_min_base)) continue;
             if (s->n == s->m) {
                 s->m = s->m < 16? 16 : s->m + (s->n>>1);
                 REALLOC(s->len, s->m);
@@ -15956,7 +15956,7 @@ uint32_t refine_rid_chain(const asg_t *rg, mg_tbuf_t *b, ul_vec_t *rch, uint64_t
 {
 	if(rch->bb.n == 1 && rch->bb.a[0].base) return 1;///no alignment
 	if(rch->bb.n == 0) return 1;///no alignment
-	uint64_t i, m, c[2], nc, cc, ni[2]; 
+	uint64_t i, m, c[2], nc, cc/**, ni[2]**/; 
 	for (i = c[1] = rch->bb.n-1, m = nc = 0; i != (uint32_t)-1; i = rch->bb.a[i].pidx) {
 		m++; c[0] = i; if(!(rg->seq[rch->bb.a[i].hid].del)) nc++; 
 	}
@@ -15967,16 +15967,14 @@ uint32_t refine_rid_chain(const asg_t *rg, mg_tbuf_t *b, ul_vec_t *rch, uint64_t
 			if(rg->seq[rch->bb.a[i].hid].del) cc++;
 			else break;
 		}
-		ni[1] = i;
+		// ni[1] = i;
 		for (i = c[0]; i != (uint32_t)-1; i = rch->bb.a[i].aidx) {
 			if(rg->seq[rch->bb.a[i].hid].del) cc++;
 			else break;
 		}
-		ni[0] = i;
+		// ni[0] = i;
 		if(nc + cc == m) return 1;
-		assert(ni[1] > ni[0] && ni[0] != (uint32_t)-1 && ni[1] != (uint32_t)-1);
-		// fprintf(stderr, "[M::%s::%.*s(id:%ld), len:%u] aln::%lu, m::%lu, nc::%lu, cc::%lu\n", __func__, 
-		// UL_INF.nid.a[ulid].n, UL_INF.nid.a[ulid].a, ulid, rch->rlen, (uint64_t)rch->bb.n, m, nc, cc);
+		// assert(ni[1] > ni[0] && ni[0] != (uint32_t)-1 && ni[1] != (uint32_t)-1);
 		return 2;
 	}
 	return 0;
@@ -16042,7 +16040,7 @@ uint32_t refine_contain_consensus_chain(const asg_t *rg, ul_vec_t *rch, R_to_U *
 		if(!is_end) continue;
 		i = m = k; occ = 0;
 		while (i != ((uint32_t)-1)) {
-			assert(bu->a[i] == ((uint64_t)-1));
+			//assert(bu->a[i] == ((uint64_t)-1));
 			if(rg->seq[rch->bb.a[i].hid].del) {
 				if(occ > 1) {
 					bu->a[m] = occ;