disable integer correction; disable contain dedup

CommandLines.cpp

@@ -56,6 +56,7 @@ static ko_longopt_t long_options[] = {
     { "bin-only",     ko_no_argument, 341},
     { "ul-round",     ko_required_argument, 342},
     { "prt-raw",     ko_no_argument, 343},
+    { "integer-correct",     ko_required_argument, 344},
 	{ 0, 0, 0 }
 };
 
@@ -267,6 +268,7 @@ void init_opt(hifiasm_opt_t* asm_opt)
     asm_opt->bin_only = 0;
     asm_opt->ul_clean_round = 1;
     asm_opt->prt_dbg_gfa = 0;
+    asm_opt->integer_correct_round = 0;
 }
 
 void destory_enzyme(enzyme* f)
@@ -801,6 +803,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
         else if (c == 341) asm_opt->bin_only = 1;
         else if (c == 342) asm_opt->ul_clean_round = atol(opt.arg);
         else if (c == 343) asm_opt->prt_dbg_gfa = 1;
+        else if (c == 344) asm_opt->integer_correct_round = 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 "0.19.0-r543"
+#define HA_VERSION "0.19.0-r550"
 
 #define VERBOSE 0
 
@@ -140,6 +140,7 @@ typedef struct {
     uint8_t bin_only;
     int32_t ul_clean_round;
     int32_t prt_dbg_gfa;
+    int32_t integer_correct_round;
 } hifiasm_opt_t;
 
 extern hifiasm_opt_t asm_opt;

Overlaps.cpp

@@ -33683,10 +33683,15 @@ ma_hit_t_alloc* src, uint64_t* readLen, R_to_U* ruIndex, bub_label_t *b_mask_t,
         sg = ma_sg_gen_ul(src, n_read, *cov, ruIndex, max_hang_length, mini_overlap_length, UL_COV_THRES);
         if(asm_opt.prt_dbg_gfa) prt_dbg_gfa(sg, "raw", *cov, src, ruIndex, max_hang_length, mini_overlap_length);
         gen_ug_opt_t(&uopt, src, NULL, max_hang_length, mini_overlap_length, gap_fuzz, min_dp, readLen, *cov, ruIndex, -1, -1, -1, -1, -1, b_mask_t);
+        
+        ///debug
+        // asg_symm(sg);
+        // dedup_contain_g(&uopt, sg);
+        
         if(clean_contain_g(&uopt, sg, 0)) update_sg_uo(sg, src);
         // prt_specfic_sge(sg, 10498, 10505, "--*--");
         asg_arc_del_trans_ul(sg, gap_fuzz);
-        // prt_specfic_sge(sg, 10498, 10505, "--#--");
+        // prt_specfic_sge(sg, 10498, 10505, "--#--");        
     }
     
     init_bub_label_t(b_mask_t, MIN(10, asm_opt.thread_num), sg->n_seq);

gfa_ut.cpp

@@ -1922,6 +1922,8 @@ void flex_asg_t_cleanup(flex_asg_t *fg)
         fg->g->is_srt = 0;
     }
     asg_cleanup(fg->g);
+    // asg_symm(fg->g);
+    // asg_arc_del_trans_ul(fg->g, fg->gap_fuzz);
 }
 
 void asg_arc_cut_contain(flex_asg_t *fg, asg64_v *in, asg64_v *in0, R_to_U* rI, float ou_rat, uint32_t only_trans_nn)
@@ -2770,7 +2772,7 @@ double ou_drop_rate, int64_t max_tip, int64_t gap_fuzz, bub_label_t *b_mask_t, i
 
 	asg_arc_cut_tips(sg, max_tip, &bu, is_ou, is_ou?rI:NULL);
     // fprintf(stderr, "[M::%s] count_edges_v_w(sg, 49778, 49847)->%ld\n", __func__, count_edges_v_w(sg, 49778, 49847));
-    
+    // if(is_ou) dedup_contain_g(uopt, sg);
     for (i = 0; i < clean_round; i++, drop += step) {
         if(drop > max_ovlp_drop_ratio) drop = max_ovlp_drop_ratio;
         if(is_ou) {
@@ -2800,7 +2802,10 @@ double ou_drop_rate, int64_t max_tip, int64_t gap_fuzz, bub_label_t *b_mask_t, i
 
         // prt_specfic_sge(sg, 10531, 10519, "--3--");
         // if(is_ou) asg_arc_cut_contain(fg, &bu, &ba, rI, ((i+1)<clean_round)?ou_drop_rate:-1);
-        if(is_ou) asg_arc_cut_contain(fg, &bu, &ba, rI, ou_drop_rate, 0);
+        if(is_ou) {
+            asg_arc_cut_contain(fg, &bu, &ba, rI, ou_drop_rate, 0);
+            // dedup_contain_g(uopt, sg);
+        }
 
         asg_arc_identify_simple_bubbles_multi(sg, b_mask_t, 1);
         asg_arc_cut_bub_links(sg, &bu, HARD_OL_DROP, HARD_OL_SEC_DROP, HARD_OU_DROP, is_ou, asm_opt.large_pop_bubble_size, rev, rI, max_tip);
@@ -2818,9 +2823,7 @@ double ou_drop_rate, int64_t max_tip, int64_t gap_fuzz, bub_label_t *b_mask_t, i
 
         if(is_ou) {
             if(ul_refine_alignment(uopt, sg)) update_sg_uo(sg, src);
-            // debug_out = 1;
             if(clean_contain_g(uopt, sg, 1)) update_sg_uo(sg, src);
-            // exit(1);
         }
     }
 
@@ -2831,6 +2834,7 @@ double ou_drop_rate, int64_t max_tip, int64_t gap_fuzz, bub_label_t *b_mask_t, i
     if(is_ou) {
         asg_arc_cut_contain(fg, &bu, &ba, rI, ou_drop_rate, 0);
         asg_arc_cut_contain(fg, &bu, &ba, rI, -1, 1);
+        // dedup_contain_g(uopt, sg);
         if(clean_contain_g(uopt, sg, 1)) update_sg_uo(sg, src);
     }
     if(!is_ou) asg_iterative_semi_circ(sg, src, &bu, max_tip, 1);
@@ -2874,7 +2878,11 @@ double ou_drop_rate, int64_t max_tip, int64_t gap_fuzz, bub_label_t *b_mask_t, i
     
     ug_ext_gfa(uopt, sg, ug_ext_len);
 
+    // if(is_ou) dedup_contain_g(uopt, sg);
+    // exit(1)
+
     output_unitig_graph(sg, uopt->coverage_cut, o_file, src, rI, uopt->max_hang, uopt->min_ovlp);
+    // exit(1);
     // flat_bubbles(sg, ruIndex->is_het); free(ruIndex->is_het); ruIndex->is_het = NULL;
     flat_soma_v(sg, src, rI);
 
@@ -17154,7 +17162,7 @@ ul_renew_t *ropt, const char *bin_file, uint64_t free_uld, uint64_t is_bridg, ui
         print_raw_uls_aln(uidx, asm_opt.output_file_name);
     // }
     
-    ul_re_correct(uidx, 3); 
+    ul_re_correct(uidx, asm_opt.integer_correct_round/**3**/); 
     init_ulg_opt_t(&uu, uopt, clean_round, 0.2, 0.6, min_ovlp_drop_ratio, max_ovlp_drop_ratio, 0.55, max_tip, max_ul_tip, b_mask_t, is_trio);
     // print_debug_gfa(sg, init_ug, uopt->coverage_cut, "UL.debug0", uopt->sources, uopt->ruIndex, uopt->max_hang, uopt->min_ovlp, 0, 0, 1);
     /**ul2ul_idx_t *u2o = **/gen_ul2ul(uidx, uopt, &uu, 0, is_bridg);

gfa_ut.h

@@ -39,5 +39,6 @@ bubble_type *gen_bubble_chain(asg_t *sg, ma_ug_t *ug, ug_opt_t *uopt, uint8_t **
 void filter_sg_by_ug(asg_t *rg, ma_ug_t *ug, ug_opt_t *uopt);
 void ug_ext_gfa(ug_opt_t *uopt, asg_t *sg, uint32_t max_len);
 void update_sg_uo(asg_t *g, ma_hit_t_alloc *src);
+uint32_t get_arcs(asg_t *g, uint32_t v, uint32_t* idx, uint32_t idx_n);
 
 #endif

inter.h

@@ -125,5 +125,6 @@ uint64_t check_ul_ov_t_consist(ul_ov_t *x, ul_ov_t *y, int64_t ql, int64_t tl, d
 uint32_t infer_se(uint32_t qs, uint32_t qe, uint32_t ts, uint32_t te, uint32_t rev, 
 uint32_t rqs, uint32_t rqe, uint32_t *rts, uint32_t *rte);
 uint32_t clean_contain_g(const ug_opt_t *uopt, asg_t *sg, uint32_t push_trans);
+void dedup_contain_g(const ug_opt_t *uopt, asg_t *sg);
 
 #endif