mirror of
https://github.com/chhylp123/hifiasm.git
synced 2026-10-02 02:58:12 +08:00
unsucessful hpc mark
This commit is contained in:
chhylp123
+148
-17
@@ -178,6 +178,12 @@ typedef struct {
|
||||
int64_t min_sc, penalty, max_drop;
|
||||
} telo_end_pip_t;
|
||||
|
||||
typedef struct {
|
||||
ma_hit_t_alloc* src;
|
||||
ma_hit_t_alloc* rec;
|
||||
int64_t n_thread, n_a;
|
||||
} rclean_aux;
|
||||
|
||||
///this value has been updated at the first line of build_string_graph_without_clean
|
||||
long long min_thres;
|
||||
|
||||
@@ -188,6 +194,24 @@ kv_u_trans_t *get_utg_ovlp(ma_ug_t **ug, asg_t* read_g, ma_hit_t_alloc* sources,
|
||||
R_to_U* ruIndex, int max_hang, int min_ovlp, kvec_asg_arc_t_warp* new_rtg_edges, bub_label_t* b_mask_t, uint8_t* r_het);
|
||||
void delete_useless_nodes(ma_ug_t **ug);
|
||||
|
||||
static void reset_ma_hit_t_0(void *data, long i, int tid) // callback for kt_for()
|
||||
{
|
||||
rclean_aux *p = (rclean_aux *)data; uint64_t k;
|
||||
for (k = 0; k < p->src[i].length; k++) {
|
||||
p->src[i].buffer[k].del = 0;
|
||||
}
|
||||
for (k = 0; k < p->rec[i].length; k++) {
|
||||
p->rec[i].buffer[k].del = 0;
|
||||
}
|
||||
}
|
||||
|
||||
void inital_ovlap(ma_hit_t_alloc *src, ma_hit_t_alloc *rec, int64_t n_a, int64_t n_thre)
|
||||
{
|
||||
rclean_aux p; memset(&p, 0, sizeof(p));
|
||||
p.n_thread = n_thre; p.src = src; p.rec = rec; p.n_a = n_a;
|
||||
kt_for(p.n_thread, reset_ma_hit_t_0, &p, p.n_a);
|
||||
}
|
||||
|
||||
static void mark_telo_ends(void *data, long i, int tid) // callback for kt_for()
|
||||
{
|
||||
telo_end_pip_t *sl = (telo_end_pip_t *)data;
|
||||
@@ -23758,7 +23782,7 @@ void write_all_data_to_disk(ma_hit_t_alloc* sources, ma_hit_t_alloc* reverse_sou
|
||||
}
|
||||
|
||||
int load_debug_graph(asg_t** sg, ma_hit_t_alloc** sources, ma_sub_t** coverage_cut,
|
||||
char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf);
|
||||
char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf, uint8_t **cmk);
|
||||
int load_all_data_from_disk(ma_hit_t_alloc **sources, ma_hit_t_alloc **reverse_sources, char* output_file_name)
|
||||
{
|
||||
char* gfa_name = (char*)malloc(strlen(output_file_name)+25);
|
||||
@@ -23776,7 +23800,7 @@ int load_all_data_from_disk(ma_hit_t_alloc **sources, ma_hit_t_alloc **reverse_s
|
||||
}
|
||||
}
|
||||
|
||||
if((asm_opt.flag & HA_F_VERBOSE_GFA) && load_debug_graph(NULL, NULL, NULL, output_file_name, NULL, NULL, NULL)) {
|
||||
if((asm_opt.flag & HA_F_VERBOSE_GFA) && load_debug_graph(NULL, NULL, NULL, output_file_name, NULL, NULL, NULL, NULL)) {
|
||||
(*sources) = NULL;
|
||||
(*reverse_sources) = NULL;
|
||||
free(gfa_name);
|
||||
@@ -32191,6 +32215,42 @@ int load_coverage_cut(ma_sub_t** coverage_cut, char* read_file_name)
|
||||
}
|
||||
|
||||
|
||||
int write_cmk(uint8_t* cmk, char* read_file_name, uint64_t n_read)
|
||||
{
|
||||
char* index_name = (char*)malloc(strlen(read_file_name)+15);
|
||||
sprintf(index_name, "%s.bin", read_file_name);
|
||||
FILE* fp = fopen(index_name, "w");
|
||||
fwrite(&n_read, sizeof(n_read), 1, fp);
|
||||
fwrite(cmk, sizeof((*(cmk))), n_read, fp);
|
||||
free(index_name);
|
||||
fflush(fp);
|
||||
fclose(fp);
|
||||
|
||||
return 1;
|
||||
}
|
||||
|
||||
int load_cmk(uint8_t** cmk, char* read_file_name)
|
||||
{
|
||||
char* index_name = (char*)malloc(strlen(read_file_name)+15);
|
||||
sprintf(index_name, "%s.bin", read_file_name);
|
||||
FILE* fp = fopen(index_name, "r");
|
||||
if(!fp)
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
int f_flag = 0;
|
||||
uint64_t n_read;
|
||||
f_flag += fread(&n_read, sizeof(n_read), 1, fp);
|
||||
(*cmk) = (uint8_t*)malloc(sizeof((*(*(cmk))))*n_read);
|
||||
|
||||
fread((*cmk), sizeof((*((*cmk)))), n_read, fp);
|
||||
|
||||
free(index_name);
|
||||
fflush(fp);
|
||||
fclose(fp);
|
||||
return 1;
|
||||
}
|
||||
|
||||
int write_coverage_cut(ma_sub_t* coverage_cut, char* read_file_name, uint64_t n_read)
|
||||
{
|
||||
char* index_name = (char*)malloc(strlen(read_file_name)+15);
|
||||
@@ -32441,7 +32501,7 @@ int load_asg_t(asg_t **sg, char* read_file_name)
|
||||
}
|
||||
|
||||
int write_debug_graph(asg_t *sg, ma_hit_t_alloc* sources, ma_sub_t* coverage_cut,
|
||||
char* output_file_name, ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf)
|
||||
char* output_file_name, ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf, uint8_t *cmk)
|
||||
{
|
||||
|
||||
char* gfa_name = (char*)malloc(strlen(output_file_name)+55);
|
||||
@@ -32465,6 +32525,10 @@ char* output_file_name, ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex, all_ul
|
||||
sprintf(gfa_name, "%s.all.debug.ul.rinfor", output_file_name);
|
||||
write_all_ul_t(ul_r_inf, gfa_name, NULL);
|
||||
}
|
||||
if(cmk) {
|
||||
sprintf(gfa_name, "%s.all.debug.cmk", output_file_name);
|
||||
write_cmk(cmk, gfa_name, R_INF.total_reads);
|
||||
}
|
||||
free(gfa_name);
|
||||
fprintf(stderr, "debug_graph has been written.\n");
|
||||
return 1;
|
||||
@@ -32472,7 +32536,7 @@ char* output_file_name, ma_hit_t_alloc* reverse_sources, R_to_U* ruIndex, all_ul
|
||||
|
||||
|
||||
int load_debug_graph(asg_t** sg, ma_hit_t_alloc** sources, ma_sub_t** coverage_cut,
|
||||
char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf)
|
||||
char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_ul_t *ul_r_inf, uint8_t **cmk)
|
||||
{
|
||||
FILE* fp = NULL;
|
||||
char* gfa_name = (char*)malloc(strlen(output_file_name)+55);
|
||||
@@ -32494,6 +32558,11 @@ char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_u
|
||||
sprintf(gfa_name, "%s.all.debug.ul.rinfor.ul.ovlp.bin", output_file_name);
|
||||
fp = fopen(gfa_name, "r"); if(!fp) return 0; fclose(fp);
|
||||
}
|
||||
|
||||
if(cmk) {
|
||||
sprintf(gfa_name, "%s.all.debug.cmk.bin", output_file_name);
|
||||
fp = fopen(gfa_name, "r"); if(!fp) return 0; fclose(fp);
|
||||
}
|
||||
|
||||
if((sources == NULL) || (reverse_sources == NULL) || (ruIndex == NULL))
|
||||
{
|
||||
@@ -32567,6 +32636,11 @@ char* output_file_name, ma_hit_t_alloc** reverse_sources, R_to_U* ruIndex, all_u
|
||||
if(!load_all_ul_t(ul_r_inf, gfa_name, &R_INF, NULL)) return 0;
|
||||
}
|
||||
|
||||
if(cmk) {
|
||||
sprintf(gfa_name, "%s.all.debug.cmk", output_file_name);
|
||||
if(!load_cmk(cmk, gfa_name)) return 0;
|
||||
}
|
||||
|
||||
R_INF.paf = (*sources); R_INF.reverse_paf = (*reverse_sources);
|
||||
|
||||
return 1;
|
||||
@@ -39767,19 +39841,24 @@ ma_hit_t_alloc* src, uint64_t* readLen, R_to_U* ruIndex, bub_label_t *b_mask_t,
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"c85c2e91-0490-438b-977b-b7d056973996", "2");
|
||||
ma_hit_sub(min_dp, src, n_read, readLen, mini_overlap_length, cov);
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"c85c2e91-0490-438b-977b-b7d056973996", "3");
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "init_0", src, NULL, NULL, NULL);
|
||||
detect_chimeric_reads(src, n_read, readLen, *cov, asm_opt.max_ov_diff_final*2.0, ul, UL_COV_THRES);
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"c85c2e91-0490-438b-977b-b7d056973996", "4");
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "init_1", src, NULL, *cov, NULL);
|
||||
ma_hit_cut(src, n_read, readLen, mini_overlap_length, cov);
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"c85c2e91-0490-438b-977b-b7d056973996", "5");
|
||||
ma_hit_flt(src, n_read, *cov, max_hang_length, mini_overlap_length);
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"c85c2e91-0490-438b-977b-b7d056973996", "6");
|
||||
ma_hit_contained_advance(src, n_read, *cov, ruIndex, max_hang_length, mini_overlap_length);
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "init_2", src, ruIndex, *cov, NULL);
|
||||
// prt_dbg_rid_ovlp(src, -1, (char*)"7b70a587-f56c-48ac-adf8-b98a67365063_2", "7");
|
||||
|
||||
if(!ul) {
|
||||
sg = ma_sg_gen(src, n_read, *cov, max_hang_length, mini_overlap_length);
|
||||
if(asm_opt.prt_dbg_gfa) prt_dbg_gfa(sg, "raw", *cov, src, ruIndex, max_hang_length, mini_overlap_length);
|
||||
if(asm_opt.prt_dbg_gfa) prt_dbg_gfa(sg, "raw", *cov, src, ruIndex, max_hang_length, mini_overlap_length);
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "init_3", NULL, ruIndex, NULL, sg);
|
||||
asg_arc_del_trans(sg, gap_fuzz);
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "init_4", NULL, ruIndex, NULL, sg);
|
||||
} else {
|
||||
ug_opt_t uopt;
|
||||
sg = ma_sg_gen_ul(src, n_read, *cov, ruIndex, max_hang_length, mini_overlap_length, UL_COV_THRES);
|
||||
@@ -39871,6 +39950,17 @@ float min_ovlp_drop_ratio, float max_ovlp_drop_ratio, char* output_file_name,
|
||||
long long bubble_dist, int read_graph, R_to_U* ruIndex, asg_t **sg_ptr,
|
||||
ma_sub_t **coverage_cut_ptr, uint8_t *cmk, int debug_g)
|
||||
{
|
||||
// const char *dbg_ids[] = {
|
||||
// "6152e53b-a2ec-4f23-898f-462c9dee8e0f",
|
||||
// "c5bbf86e-3146-4757-9882-55a5baebf770",
|
||||
// };
|
||||
// const uint64_t dbg_ids_n[] = {
|
||||
// 3177240,
|
||||
// 2198541,
|
||||
// };
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "test_0", sources, NULL, NULL, NULL);
|
||||
|
||||
char *o_file = get_outfile_name(output_file_name);
|
||||
ma_sub_t *coverage_cut = *coverage_cut_ptr;
|
||||
asg_t *sg = *sg_ptr;
|
||||
@@ -39894,8 +39984,10 @@ ma_sub_t **coverage_cut_ptr, uint8_t *cmk, int debug_g)
|
||||
///normalize_ma_hit_t_single_side(sources, n_read);
|
||||
|
||||
normalize_ma_hit_t_single_side_advance(sources, n_read, asm_opt.is_ont, cmk);
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "test_1", sources, NULL, NULL, NULL);
|
||||
// normalize_ma_hit_t_single_side_advance_mult(sources, n_read, asm_opt.thread_num);
|
||||
normalize_ma_hit_t_single_side_advance(reverse_sources, n_read, 0, cmk);
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "test_2", sources, NULL, NULL, NULL);
|
||||
// normalize_ma_hit_t_single_side_advance_mult(reverse_sources, n_read, asm_opt.thread_num);
|
||||
|
||||
if (ha_opt_triobin(&asm_opt))
|
||||
@@ -39925,6 +40017,21 @@ ma_sub_t **coverage_cut_ptr, uint8_t *cmk, int debug_g)
|
||||
// prt_dbg_rid_ovlp(sources, 27087, NULL, "0-b");
|
||||
// prt_specific_overlap(sources, 22233, 22235, "0-c");
|
||||
// prt_specific_overlap(sources, 22235, 22233, "0-c");
|
||||
|
||||
|
||||
|
||||
///@brief debug
|
||||
if (asm_opt.flag & HA_F_VERBOSE_GFA) {
|
||||
write_debug_graph(sg, sources, coverage_cut, output_file_name, reverse_sources, ruIndex, &UL_INF, cmk);
|
||||
debug_gfa:;
|
||||
}
|
||||
|
||||
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ini_s", sources, NULL, NULL, NULL);
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ini_r", reverse_sources, NULL, NULL, NULL);
|
||||
// exit(1);
|
||||
|
||||
sg = gen_init_sg(min_dp, n_read, mini_overlap_length, max_hang_length, gap_fuzz, sources, readLen, ruIndex,
|
||||
&b_mask_t, &coverage_cut, asm_opt.ar?&UL_INF:NULL, te);
|
||||
// if(asm_opt.ar) exit(1);
|
||||
@@ -39955,24 +40062,29 @@ ma_sub_t **coverage_cut_ptr, uint8_t *cmk, int debug_g)
|
||||
free(unlean_name);
|
||||
}
|
||||
|
||||
|
||||
// if (asm_opt.flag & HA_F_VERBOSE_GFA) {
|
||||
// write_debug_graph(sg, sources, coverage_cut, output_file_name, reverse_sources, ruIndex, &UL_INF);
|
||||
// debug_gfa:;
|
||||
// }
|
||||
/**
|
||||
///@brief debug
|
||||
if (asm_opt.flag & HA_F_VERBOSE_GFA) {
|
||||
write_debug_graph(sg, sources, coverage_cut, output_file_name, reverse_sources, ruIndex, &UL_INF, cmk);
|
||||
debug_gfa:;
|
||||
}
|
||||
**/
|
||||
// hc_dbg_prt_ma_hit_t(NULL, 0, NULL, 0, "clen_z", sources, ruIndex, coverage_cut, sg);
|
||||
|
||||
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, te);
|
||||
ul_clean_gfa(&uopt, sg, sources, reverse_sources, ruIndex, clean_round, min_ovlp_drop_ratio, max_ovlp_drop_ratio,
|
||||
0.6, asm_opt.max_short_tip, gap_fuzz, &b_mask_t, !!asm_opt.ar, ha_opt_triobin(&asm_opt), UL_COV_THRES, cmk, o_file);
|
||||
/**
|
||||
///@brief debug
|
||||
if (asm_opt.flag & HA_F_VERBOSE_GFA) {
|
||||
write_debug_graph(sg, sources, coverage_cut, output_file_name, reverse_sources, ruIndex, &UL_INF);
|
||||
write_debug_graph(sg, sources, coverage_cut, output_file_name, reverse_sources, ruIndex, &UL_INF, cmk);
|
||||
debug_gfa:;
|
||||
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, te);
|
||||
// set_hom_global_coverage(&asm_opt, sg, coverage_cut, sources, reverse_sources, ruIndex, max_hang_length, mini_overlap_length);
|
||||
}
|
||||
**/
|
||||
|
||||
if(asm_opt.ar) {
|
||||
gradually_renew_g(&sources, &reverse_sources, &n_read, &readLen, &coverage_cut, ruIndex,
|
||||
@@ -40229,7 +40341,7 @@ long long bubble_dist, int read_graph, int write)
|
||||
min_thres = asm_opt.max_short_tip + 1;
|
||||
if (asm_opt.flag & HA_F_VERBOSE_GFA)
|
||||
{
|
||||
if(load_debug_graph(/**NULL**/&sg, &sources, /**NULL**/&coverage_cut, output_file_name, &reverse_sources, &ruIndex, &UL_INF))
|
||||
if(load_debug_graph(NULL/**&sg**/, &sources, NULL/**&coverage_cut**/, output_file_name, &reverse_sources, &ruIndex, &UL_INF, (asm_opt.is_ont)?(&cmk):(NULL)))
|
||||
{
|
||||
fprintf(stderr, "debug gfa has been loaded\n");
|
||||
|
||||
@@ -40249,20 +40361,39 @@ long long bubble_dist, int read_graph, int write)
|
||||
}
|
||||
///debug_info_of_specfic_read("m64011_190830_220126/31720629/ccs", sources, reverse_sources, -1, "beg");
|
||||
|
||||
// const char *dbg_ids[] = {
|
||||
// "6152e53b-a2ec-4f23-898f-462c9dee8e0f",
|
||||
// "c5bbf86e-3146-4757-9882-55a5baebf770",
|
||||
// };
|
||||
// const uint64_t dbg_ids_n[] = {
|
||||
// 3177240,
|
||||
// 2198541,
|
||||
// };
|
||||
|
||||
|
||||
if (!(asm_opt.flag & HA_F_BAN_ASSEMBLY))
|
||||
{
|
||||
// if(asm_opt.is_ont) handle_chemical_r(asm_opt.thread_num, R_INF.total_reads);
|
||||
if(asm_opt.is_ont) {
|
||||
uint64_t i = 0, j = 0;
|
||||
// uint64_t i = 0, j = 0;
|
||||
memset(ruIndex.index, -1, sizeof(uint32_t)*(ruIndex.len));
|
||||
for (i = 0; i < n_read; i++) {
|
||||
for (j = 0; j < sources[i].length; j++) sources[i].buffer[j].del = 0;
|
||||
for (j = 0; j < reverse_sources[i].length; j++) reverse_sources[i].buffer[j].del = 0;
|
||||
}
|
||||
inital_ovlap(sources, reverse_sources, n_read, asm_opt.thread_num);
|
||||
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ttst_0", sources, NULL, NULL, NULL);
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "trev_0", reverse_sources, NULL, NULL, NULL);
|
||||
|
||||
if(asm_opt.is_ont) cmk = gen_chemical_arc_rf(asm_opt.thread_num, R_INF.total_reads);
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ttst_1", sources, NULL, NULL, NULL);
|
||||
|
||||
if(asm_opt.del_hf) clean_arc_rf(asm_opt.thread_num, R_INF.total_reads);
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ttst_2", sources, NULL, NULL, NULL);
|
||||
}
|
||||
try_rescue_overlaps(sources, reverse_sources, n_read, 4, asm_opt.is_ont);
|
||||
|
||||
// hc_dbg_prt_ma_hit_t(dbg_ids, sizeof(dbg_ids)/sizeof(dbg_ids[0]), dbg_ids_n, sizeof(dbg_ids_n)/sizeof(dbg_ids_n[0]), "ttst_3", sources, NULL, NULL, NULL);
|
||||
|
||||
clean_graph(min_dp, sources, reverse_sources, n_read, readLen, mini_overlap_length,
|
||||
max_hang_length, clean_round, gap_fuzz, min_ovlp_drop_ratio, max_ovlp_drop_ratio,
|
||||
|
||||
Reference in New Issue
Block a user