xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-10-11 03:40:57 +08:00
Code Issues Packages Projects Releases Wiki Activity

UL graph done

Browse Source
This commit is contained in:
chhylp123
2022-07-24 18:13:16 -04:00
parent 138610b049
commit 33afa25398
5 changed files with 3581 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Overlaps.cpp
+10 -8
View File
@@ -11180,7 +11180,6 @@ void debug_contig_end(ma_ug_t *ug, asg_t* read_g, kvec_asg_arc_t_warp* edge)
} }
} }
void renew_utg(ma_ug_t **ug, asg_t* read_g, kvec_asg_arc_t_warp* edge);
void push_sub_unitig(ma_ug_t *n_ug, ma_utg_t *src_u, asg_t *read_g, kvec_asg_arc_t_warp* edge, void push_sub_unitig(ma_ug_t *n_ug, ma_utg_t *src_u, asg_t *read_g, kvec_asg_arc_t_warp* edge,
uint32_t beg_idx, uint32_t occ) uint32_t beg_idx, uint32_t occ)
{ {
@@ -14819,14 +14818,14 @@ const char* command)
uint32_t print_debug_gfa(asg_t *read_g, ma_ug_t *ug, ma_sub_t* coverage_cut, const char* output_file_name, uint32_t print_debug_gfa(asg_t *read_g, ma_ug_t *ug, ma_sub_t* coverage_cut, const char* output_file_name,
ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp) ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp, int is_polish, int is_update_ou, int is_check_alter_lable)
{ {
kvec_asg_arc_t_warp new_rtg_edges; kvec_asg_arc_t_warp new_rtg_edges;
kv_init(new_rtg_edges.a); kv_init(new_rtg_edges.a);
if(ug == NULL) { if(ug == NULL) {
ug = ma_ug_gen(read_g); ug = ma_ug_gen(read_g);
} else { } else if(is_check_alter_lable) {
uint32_t i; uint32_t i;
for (i = 0; i < ug->u.n; ++i) for (i = 0; i < ug->u.n; ++i)
{ {
@@ -14844,8 +14843,8 @@ ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp)
} }
} }
update_ug_ou(ug, read_g); if(is_update_ou) update_ug_ou(ug, read_g);
ma_ug_seq(ug, read_g, coverage_cut, sources, &new_rtg_edges, max_hang, min_ovlp, 0, 0); if(is_polish) ma_ug_seq(ug, read_g, coverage_cut, sources, &new_rtg_edges, max_hang, min_ovlp, 0, 0);
fprintf(stderr, "Writing raw unitig GFA to disk... \n"); fprintf(stderr, "Writing raw unitig GFA to disk... \n");
char* gfa_name = (char*)malloc(strlen(output_file_name)+25); char* gfa_name = (char*)malloc(strlen(output_file_name)+25);
@@ -16627,7 +16626,7 @@ int min_ovlp, hap_cov_t *cov)
resolve_tangles(ug, read_g, reverse_sources, 20, 100, 0.05, 0.2, ruIndex, cov->is_r_het, (uint32_t)-1, drop_ratio); resolve_tangles(ug, read_g, reverse_sources, 20, 100, 0.05, 0.2, ruIndex, cov->is_r_het, (uint32_t)-1, drop_ratio);
drop_semi_circle(ug, g, read_g, reverse_sources, ruIndex, cov->is_r_het); drop_semi_circle(ug, g, read_g, reverse_sources, ruIndex, cov->is_r_het);
print_debug_gfa(read_g, ug, coverage_cut, "debug_clean_end", sources, ruIndex, asm_opt.max_hang_Len, asm_opt.min_overlap_Len); // print_debug_gfa(read_g, ug, coverage_cut, "debug_clean_end", sources, ruIndex, asm_opt.max_hang_Len, asm_opt.min_overlap_Len);
unitig_arc_del_short_diploid_by_length_topo(g, ug, drop_ratio, asm_opt.max_short_tip, reverse_sources, 0, 1); unitig_arc_del_short_diploid_by_length_topo(g, ug, drop_ratio, asm_opt.max_short_tip, reverse_sources, 0, 1);
if(round > 0) if(round > 0)
@@ -31549,8 +31548,11 @@ ma_sub_t **coverage_cut_ptr, int debug_g)
gen_ug_opt_t(&uopt, sources, reverse_sources, max_hang_length, mini_overlap_length, gap_fuzz, min_dp, readLen, coverage_cut, ruIndex, gen_ug_opt_t(&uopt, sources, reverse_sources, max_hang_length, mini_overlap_length, gap_fuzz, min_dp, readLen, coverage_cut, ruIndex,
(asm_opt.max_short_tip*2), 0.15, 3, 0.05, 0.9, &b_mask_t); (asm_opt.max_short_tip*2), 0.15, 3, 0.05, 0.9, &b_mask_t);
} }
if(asm_opt.ar) ul_realignment_gfa(&uopt, sg); if(asm_opt.ar) {
print_debug_gfa(sg, NULL, coverage_cut, "UL.debug", sources, ruIndex, max_hang_length, mini_overlap_length); ul_realignment_gfa(&uopt, sg, clean_round, min_ovlp_drop_ratio, max_ovlp_drop_ratio,
asm_opt.max_short_tip, &b_mask_t, ha_opt_triobin(&asm_opt));
}
print_debug_gfa(sg, NULL, coverage_cut, "UL.debug", sources, ruIndex, max_hang_length, mini_overlap_length, 0, 0, 0);
/** /**
asg_cut_tip(sg, asm_opt.max_short_tip); asg_cut_tip(sg, asm_opt.max_short_tip);
///debug_info_of_specfic_node("m64043_200505_112554/8849050/ccs", sg, "inner_1"); ///debug_info_of_specfic_node("m64043_200505_112554/8849050/ccs", sg, "inner_1");
Overlaps.h
+3 -1
View File
@@ -259,6 +259,7 @@ typedef struct {
// kv_ul_ov_t *ov; // kv_ul_ov_t *ov;
} ul_idx_t; } ul_idx_t;
#define MA_HT_DOUBLE (-1024)
#define MA_HT_INT (-1) #define MA_HT_INT (-1)
#define MA_HT_QCONT (-2) #define MA_HT_QCONT (-2)
#define MA_HT_TCONT (-3) #define MA_HT_TCONT (-3)
@@ -1079,7 +1080,7 @@ int asg_arc_del_triangular_directly(asg_t *g, long long min_edge_length,
ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex); ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex);
int asg_arc_del_short_diploid_by_exact(asg_t *g, int max_ext, ma_hit_t_alloc* sources); int asg_arc_del_short_diploid_by_exact(asg_t *g, int max_ext, ma_hit_t_alloc* sources);
uint32_t print_debug_gfa(asg_t *read_g, ma_ug_t *ug, ma_sub_t* coverage_cut, const char* output_file_name, uint32_t print_debug_gfa(asg_t *read_g, ma_ug_t *ug, ma_sub_t* coverage_cut, const char* output_file_name,
ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp); ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp, int is_polish, int is_update_ou, int is_check_alter_lable);
void debug_info_of_specfic_node(const char* name, asg_t *g, R_to_U* ruIndex, const char* command); void debug_info_of_specfic_node(const char* name, asg_t *g, R_to_U* ruIndex, const char* command);
ma_ug_t *gen_polished_ug(const ug_opt_t *uopt, asg_t *sg); ma_ug_t *gen_polished_ug(const ug_opt_t *uopt, asg_t *sg);
void output_unitig_graph(asg_t *sg, ma_sub_t* coverage_cut, char* output_file_name, void output_unitig_graph(asg_t *sg, ma_sub_t* coverage_cut, char* output_file_name,
@@ -1087,6 +1088,7 @@ ma_hit_t_alloc* sources, R_to_U* ruIndex, int max_hang, int min_ovlp);
void flat_soma_v(asg_t *sg, ma_hit_t_alloc* sources, R_to_U* ruIndex); void flat_soma_v(asg_t *sg, ma_hit_t_alloc* sources, R_to_U* ruIndex);
void hic_clean(asg_t* read_g); void hic_clean(asg_t* read_g);
int64_t count_edges_v_w(asg_t *g, uint32_t v, uint32_t w); int64_t count_edges_v_w(asg_t *g, uint32_t v, uint32_t w);
void renew_utg(ma_ug_t **ug, asg_t* read_g, kvec_asg_arc_t_warp* edge);
#define JUNK_COV 5 #define JUNK_COV 5
#define DISCARD_RATE 0.8 #define DISCARD_RATE 0.8
gfa_ut.cpp
+3564 -16
View File
File diff suppressed because it is too large Load Diff
gfa_ut.h
+2 -1
View File
@@ -14,5 +14,6 @@ void asg_arc_cut_bub_links(asg_t *g, asg64_v *in, float len_rat, float sec_len_r
void asg_arc_cut_complex_bub_links(asg_t *g, asg64_v *in, float len_rat, float ou_rat, uint32_t is_ou, bub_label_t *b_mask_t); void asg_arc_cut_complex_bub_links(asg_t *g, asg64_v *in, float len_rat, float ou_rat, uint32_t is_ou, bub_label_t *b_mask_t);
uint32_t asg_cut_large_indel(asg_t *g, asg64_v *in, int32_t max_ext, float ou_rat, uint32_t is_ou); uint32_t asg_cut_large_indel(asg_t *g, asg64_v *in, int32_t max_ext, float ou_rat, uint32_t is_ou);
uint32_t asg_cut_semi_circ(asg_t *g, uint32_t lim_len, uint32_t is_clean); uint32_t asg_cut_semi_circ(asg_t *g, uint32_t lim_len, uint32_t is_clean);
void ul_realignment_gfa(ug_opt_t *uopt, asg_t *sg); void ul_realignment_gfa(ug_opt_t *uopt, asg_t *sg, int64_t clean_round, double min_ovlp_drop_ratio,
double max_ovlp_drop_ratio, int64_t max_tip, bub_label_t *b_mask_t, uint32_t is_trio);
#endif #endif
inter.cpp
+1 -14
View File
@@ -8913,18 +8913,9 @@ static void filter_short_ulalignments(void *data, long i, int tid) // callback f
const ma_ug_t *ug = (ma_ug_t *)data; const ma_ug_t *ug = (ma_ug_t *)data;
uc_block_t *a = NULL; uc_block_t *p; int64_t k, a_n; uint32_t z, fz, lz, l, bz; uc_block_t *a = NULL; uc_block_t *p; int64_t k, a_n; uint32_t z, fz, lz, l, bz;
a = UL_INF.a[i].bb.a; a_n = UL_INF.a[i].bb.n; a = UL_INF.a[i].bb.a; a_n = UL_INF.a[i].bb.n;
// if(i == 4) {
// fprintf(stderr, "[M::%s::i->%ld] a_n->%ld\n", __func__, i, a_n);
// }
for (k = a_n - 1; k >= 0; k--) { for (k = a_n - 1; k >= 0; k--) {
p = &(a[k]); p = &(a[k]);
// if(i == 27512) {
// fprintf(stderr, "+[M::%s::k->%ld::a_n->%ld] p->ts::%u, p->te::%u, p->pchain::%u, p->pidx::%u, p->aidx::%u, p->qs::%u, p->qe::%u\n",
// __func__, k, a_n, p->ts, p->te, p->pchain, p->pidx, p->aidx, p->qs, p->qe);
// }
// if(p->qs > p->qe) {
// fprintf(stderr, "[M::%s] id->%ld, qs->%u, qe->%u\n", __func__, i, p->qs, p->qe);
// }
if(p->base || (!p->el) || (!p->pchain)) continue; if(p->base || (!p->el) || (!p->pchain)) continue;
if(p->pidx == (uint32_t)-1) { if(p->pidx == (uint32_t)-1) {
if(!ugl_cover_check(p->ts, p->te, &(ug->u.a[p->hid]))) { if(!ugl_cover_check(p->ts, p->te, &(ug->u.a[p->hid]))) {
@@ -8956,10 +8947,6 @@ static void filter_short_ulalignments(void *data, long i, int tid) // callback f
for (k = a_n - 1; k >= 0; k--) { for (k = a_n - 1; k >= 0; k--) {
p = &(a[k]); p = &(a[k]);
// if(i == 27512) {
// fprintf(stderr, "-[M::%s::k->%ld::a_n->%ld] p->ts::%u, p->te::%u, p->pchain::%u, p->pidx::%u, p->aidx::%u, p->qs::%u, p->qe::%u\n",
// __func__, k, a_n, p->ts, p->te, p->pchain, p->pidx, p->aidx, p->qs, p->qe);
// }
if(p->base || (!p->el) || (!p->pchain)) continue; if(p->base || (!p->el) || (!p->pchain)) continue;
if(p->pidx != (uint32_t)-1) { if(p->pidx != (uint32_t)-1) {
assert(a[p->pidx].aidx == (uint32_t)k); assert(a[p->pidx].aidx == (uint32_t)k);