xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-10-11 00:10:56 +08:00
Code Issues Packages Projects Releases Wiki Activity

change max coverage of pri-range

Browse Source
This commit is contained in:
chhylp123
2020-07-18 17:40:58 -04:00
parent 88ab801717
commit 952471657f
5 changed files with 76 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Assembly.cpp
+3
View File
@@ -476,6 +476,7 @@ static void worker_ovec(void *data, long i, int tid)
int fully_cov, abnormal; int fully_cov, abnormal;
ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.02, asm_opt.max_n_chain, 1); ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.02, asm_opt.max_n_chain, 1);
///ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.08, asm_opt.max_n_chain, 1);
clear_Cigar_record(&b->cigar1); clear_Cigar_record(&b->cigar1);
clear_Round2_alignment(&b->round2); clear_Round2_alignment(&b->round2);
@@ -514,6 +515,7 @@ static void worker_ovec_related_reads(void *data, long i, int tid)
memcmp(asm_opt.required_read_name, Get_NAME((R_INF), i), Get_NAME_LENGTH((R_INF),i)) == 0) memcmp(asm_opt.required_read_name, Get_NAME((R_INF), i), Get_NAME_LENGTH((R_INF),i)) == 0)
{ {
ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.02, asm_opt.max_n_chain, 1); ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.02, asm_opt.max_n_chain, 1);
///ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.08, asm_opt.max_n_chain, 1);
fprintf(stderr, ">%.*s\n", (int)Get_NAME_LENGTH((R_INF), i), Get_NAME((R_INF), i)); fprintf(stderr, ">%.*s\n", (int)Get_NAME_LENGTH((R_INF), i), Get_NAME((R_INF), i));
recover_UC_Read(&b->self_read, &R_INF, i); recover_UC_Read(&b->self_read, &R_INF, i);
@@ -957,6 +959,7 @@ static void worker_ov_final(void *data, long i, int tid)
//get_new_candidates(i, &g_read, &overlap_list, &array_list, &l, 0.001, 0); //get_new_candidates(i, &g_read, &overlap_list, &array_list, &l, 0.001, 0);
ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.001, asm_opt.max_n_chain, 0); ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.001, asm_opt.max_n_chain, 0);
///ha_get_new_candidates(b->ab, i, &b->self_read, &b->olist, &b->clist, 0.08, asm_opt.max_n_chain, 0);
/** /**
correct_overlap(&overlap_list, &R_INF, &g_read, &correct, &overlap_read, &POA_Graph, &DAGCon, correct_overlap(&overlap_list, &R_INF, &g_read, &correct, &overlap_read, &POA_Graph, &DAGCon,
CommandLines.cpp
+40 -39
View File
@@ -1,6 +1,7 @@
#include <zlib.h> #include <zlib.h>
#include <stdlib.h> #include <stdlib.h>
#include <stdio.h> #include <stdio.h>
#include <stdint.h>
#include <sys/time.h> #include <sys/time.h>
#include "CommandLines.h" #include "CommandLines.h"
#include "ketopt.h" #include "ketopt.h"
@@ -36,45 +37,48 @@ void Print_H(hifiasm_opt_t* asm_opt)
fprintf(stderr, "Usage: hifiasm [options] <in_1.fq> <in_2.fq> <...>\n"); fprintf(stderr, "Usage: hifiasm [options] <in_1.fq> <in_2.fq> <...>\n");
fprintf(stderr, "Options:\n"); fprintf(stderr, "Options:\n");
fprintf(stderr, " Assembly:\n"); fprintf(stderr, " Assembly:\n");
fprintf(stderr, " -o FILE prefix of output files [%s]\n", asm_opt->output_file_name); fprintf(stderr, " -o FILE prefix of output files [%s]\n", asm_opt->output_file_name);
fprintf(stderr, " -t INT number of threads [%d]\n", asm_opt->thread_num); fprintf(stderr, " -t INT number of threads [%d]\n", asm_opt->thread_num);
fprintf(stderr, " -r INT round of correction [%d]\n", asm_opt->number_of_round); fprintf(stderr, " -r INT round of correction [%d]\n", asm_opt->number_of_round);
fprintf(stderr, " -a INT round of assembly cleaning [%d]\n", asm_opt->clean_round); fprintf(stderr, " -a INT round of assembly cleaning [%d]\n", asm_opt->clean_round);
fprintf(stderr, " -k INT k-mer length (must be <64) [%d]\n", asm_opt->k_mer_length); fprintf(stderr, " -k INT k-mer length (must be <64) [%d]\n", asm_opt->k_mer_length);
fprintf(stderr, " -w INT minimizer window size [%d]\n", asm_opt->mz_win); fprintf(stderr, " -w INT minimizer window size [%d]\n", asm_opt->mz_win);
fprintf(stderr, " -f INT number of bits for bloom filter; 0 to disable [%d]\n", asm_opt->bf_shift); fprintf(stderr, " -f INT number of bits for bloom filter; 0 to disable [%d]\n", asm_opt->bf_shift);
fprintf(stderr, " -D FLOAT drop k-mers occuring >FLOAT*coverage times [%.1f]\n", asm_opt->high_factor); fprintf(stderr, " -D FLOAT drop k-mers occuring >FLOAT*coverage times [%.1f]\n", asm_opt->high_factor);
fprintf(stderr, " -N INT consider up to max(-D*coverage,-N) overlaps for each oriented read [%d]\n", asm_opt->max_n_chain); fprintf(stderr, " -N INT consider up to max(-D*coverage,-N) overlaps for each oriented read [%d]\n", asm_opt->max_n_chain);
fprintf(stderr, " -i ignore saved overlaps in *.ovlp* files\n"); fprintf(stderr, " -i ignore saved overlaps in *.ovlp* files\n");
fprintf(stderr, " -z INT length of adapters that should be removed [%d]\n", asm_opt->adapterLen); fprintf(stderr, " -z INT length of adapters that should be removed [%d]\n", asm_opt->adapterLen);
fprintf(stderr, " -m INT size of popped large bubbles for contig graph [%lld]\n", asm_opt->large_pop_bubble_size); fprintf(stderr, " -m INT size of popped large bubbles for contig graph [%lld]\n", asm_opt->large_pop_bubble_size);
fprintf(stderr, " -p INT size of popped small bubbles for haplotype-resolved unitig graph [%lld]\n", asm_opt->small_pop_bubble_size); fprintf(stderr, " -p INT size of popped small bubbles for haplotype-resolved unitig graph [%lld]\n", asm_opt->small_pop_bubble_size);
fprintf(stderr, " -n INT small removed unitig threshold [%d]\n", asm_opt->max_short_tip); fprintf(stderr, " -n INT small removed unitig threshold [%d]\n", asm_opt->max_short_tip);
fprintf(stderr, " -x FLOAT max overlap drop ratio [%.2g]\n", asm_opt->max_drop_rate); fprintf(stderr, " -x FLOAT max overlap drop ratio [%.2g]\n", asm_opt->max_drop_rate);
fprintf(stderr, " -y FLOAT min overlap drop ratio [%.2g]\n", asm_opt->min_drop_rate); fprintf(stderr, " -y FLOAT min overlap drop ratio [%.2g]\n", asm_opt->min_drop_rate);
fprintf(stderr, " --pri-range INT,INT keep contigs with coverage in this range at p_ctg.gfa. Inferred automatically in default. Set -1,-1 to disable\n"); fprintf(stderr, " --pri-range INT1[,INT2]\n\
keep contigs with coverage in this range at p_ctg.gfa. Inferred automatically in default\n\
If INT2 is not specified, it is set to infinity. Set -1 to disable\n");
fprintf(stderr, " -u disable post join contigs step which may improve N50. Don't disable in default.\n"); fprintf(stderr, " -u disable post join contigs step which may improve N50. Don't disable in default.\n");
fprintf(stderr, " --version show version number\n"); fprintf(stderr, " --version show version number\n");
fprintf(stderr, " -h show help information\n"); fprintf(stderr, " -h show help information\n");
fprintf(stderr, " Trio-partition:\n"); fprintf(stderr, " Trio-partition:\n");
fprintf(stderr, " -1 FILE hap1/paternal k-mer dump generated by \"yak count\" []\n"); fprintf(stderr, " -1 FILE hap1/paternal k-mer dump generated by \"yak count\" []\n");
fprintf(stderr, " -2 FILE hap2/maternal k-mer dump generated by \"yak count\" []\n"); fprintf(stderr, " -2 FILE hap2/maternal k-mer dump generated by \"yak count\" []\n");
fprintf(stderr, " -3 FILE list of hap1/paternal read names []\n"); fprintf(stderr, " -3 FILE list of hap1/paternal read names []\n");
fprintf(stderr, " -4 FILE list of hap2/maternal read names []\n"); fprintf(stderr, " -4 FILE list of hap2/maternal read names []\n");
fprintf(stderr, " -c INT lower bound of the binned k-mer's frequency [%d]\n", asm_opt->min_cnt); fprintf(stderr, " -c INT lower bound of the binned k-mer's frequency [%d]\n", asm_opt->min_cnt);
fprintf(stderr, " -d INT upper bound of the binned k-mer's frequency [%d]\n", asm_opt->mid_cnt); fprintf(stderr, " -d INT upper bound of the binned k-mer's frequency [%d]\n", asm_opt->mid_cnt);
fprintf(stderr, " Purge-dups:\n"); fprintf(stderr, " Purge-dups:\n");
fprintf(stderr, " -l INT level of purge-dup. In default, [%d] for non-trio; [%d] for trio (see hifiasm.1 for details)\n", fprintf(stderr, " -l INT level of purge-dup. In default, [%d] for non-trio; [%d] for trio (see hifiasm.1 for details)\n",
asm_opt->purge_level_primary, asm_opt->purge_level_trio); asm_opt->purge_level_primary, asm_opt->purge_level_trio);
fprintf(stderr, " -s FLOAT similarity threshold for duplicate haplotigs [%g]\n", fprintf(stderr, " -s FLOAT similarity threshold for duplicate haplotigs [%g]\n",
asm_opt->purge_simi_rate); asm_opt->purge_simi_rate);
fprintf(stderr, " -O INT min number of overlapped reads for duplicate haplotigs [%d]\n", fprintf(stderr, " -O INT min number of overlapped reads for duplicate haplotigs [%d]\n",
asm_opt->purge_overlap_len); asm_opt->purge_overlap_len);
fprintf(stderr, " --purge-cov INT coverage upper bound of Purge-dups, which is inferred automatically in default.\n"); fprintf(stderr, " --purge-cov INT\n\
coverage upper bound of Purge-dups, which is inferred automatically in default.\n");
fprintf(stderr, "Example: ./hifiasm -o NA12878.asm -t 32 NA12878.fq.gz\n"); fprintf(stderr, "Example: ./hifiasm -o NA12878.asm -t 32 NA12878.fq.gz\n");
@@ -124,7 +128,7 @@ void init_opt(hifiasm_opt_t* asm_opt)
asm_opt->purge_simi_rate = 0.75; asm_opt->purge_simi_rate = 0.75;
asm_opt->purge_overlap_len = 1; asm_opt->purge_overlap_len = 1;
asm_opt->recover_atg_cov_min = -1024; asm_opt->recover_atg_cov_min = -1024;
asm_opt->recover_atg_cov_max = -1024; asm_opt->recover_atg_cov_max = INT32_MAX;
asm_opt->hom_global_coverage = -1; asm_opt->hom_global_coverage = -1;
} }
@@ -315,14 +319,6 @@ int check_option(hifiasm_opt_t* asm_opt)
return 0; return 0;
} }
if(asm_opt->recover_atg_cov_min != asm_opt->recover_atg_cov_max &&
(asm_opt->recover_atg_cov_min == -1024 || asm_opt->recover_atg_cov_max == -1024))
{
fprintf(stderr, "[ERROR] primary contig coverage range should be set correctly (--primary-range)\n");
return 0;
}
if(asm_opt->fn_bin_yak[0] != NULL && check_file(asm_opt->fn_bin_yak[0], "YAK1") == 0) return 0; if(asm_opt->fn_bin_yak[0] != NULL && check_file(asm_opt->fn_bin_yak[0], "YAK1") == 0) return 0;
if(asm_opt->fn_bin_yak[1] != NULL && check_file(asm_opt->fn_bin_yak[1], "YAK2") == 0) return 0; if(asm_opt->fn_bin_yak[1] != NULL && check_file(asm_opt->fn_bin_yak[1], "YAK2") == 0) return 0;
@@ -436,6 +432,10 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
char* s = NULL; char* s = NULL;
asm_opt->recover_atg_cov_min = strtol(opt.arg, &s, 10); asm_opt->recover_atg_cov_min = strtol(opt.arg, &s, 10);
if (*s == ',') asm_opt->recover_atg_cov_max = strtol(s + 1, &s, 10); if (*s == ',') asm_opt->recover_atg_cov_max = strtol(s + 1, &s, 10);
if(asm_opt->recover_atg_cov_min == -1 || asm_opt->recover_atg_cov_max == -1)
{
asm_opt->recover_atg_cov_min = asm_opt->recover_atg_cov_max = -1;
}
} }
else if (c == 'l') else if (c == 'l')
{ ///0: disable purge_dup; 1: purge containment; 2: purge overlap { ///0: disable purge_dup; 1: purge containment; 2: purge overlap
@@ -461,6 +461,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
Print_H(asm_opt); Print_H(asm_opt);
return 0; return 0;
} }
///fprintf(stderr, "max_ov_diff_ec: %f, max_ov_diff_final: %f\n", asm_opt->max_ov_diff_ec, asm_opt->max_ov_diff_final);
get_queries(argc, argv, &opt, asm_opt); get_queries(argc, argv, &opt, asm_opt);
CommandLines.h
+1 -1
View File
@@ -3,7 +3,7 @@
#include <pthread.h> #include <pthread.h>
#define HA_VERSION "0.8-dirty-r280" #define HA_VERSION "0.9-r281"
#define VERBOSE 0 #define VERBOSE 0
Overlaps.cpp
+28 -8
View File
@@ -13503,14 +13503,25 @@ kvec_asg_arc_t_warp* new_rtg_edges)
purge_dups(*ug, read_g, coverage_cut, sources, reverse_sources, ruIndex, new_rtg_edges, purge_dups(*ug, read_g, coverage_cut, sources, reverse_sources, ruIndex, new_rtg_edges,
asm_opt.purge_simi_rate, asm_opt.purge_overlap_len, max_hang, min_ovlp, bubble_dist, asm_opt.purge_simi_rate, asm_opt.purge_overlap_len, max_hang, min_ovlp, bubble_dist,
drop_ratio, 1, 1); drop_ratio, 1, 1);
if(asm_opt.recover_atg_cov_min == -1024 || asm_opt.recover_atg_cov_max == -1024) if(asm_opt.recover_atg_cov_min == -1024)
{ {
asm_opt.recover_atg_cov_max = asm_opt.hom_global_coverage/HOM_PEAK_RATE; asm_opt.recover_atg_cov_max = asm_opt.hom_global_coverage/HOM_PEAK_RATE;
asm_opt.recover_atg_cov_min = asm_opt.recover_atg_cov_max * 0.8; asm_opt.recover_atg_cov_min = asm_opt.recover_atg_cov_max * 0.8;
asm_opt.recover_atg_cov_max = asm_opt.recover_atg_cov_max * 1.2; ///asm_opt.recover_atg_cov_max = asm_opt.recover_atg_cov_max * 1.2;
asm_opt.recover_atg_cov_max = INT32_MAX;
} }
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, %d]\n", if(asm_opt.recover_atg_cov_max != INT32_MAX)
__func__, asm_opt.recover_atg_cov_min, asm_opt.recover_atg_cov_max); {
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, %d]\n",
__func__, asm_opt.recover_atg_cov_min, asm_opt.recover_atg_cov_max);
}
else
{
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, infinity]\n",
__func__, asm_opt.recover_atg_cov_min);
}
///primary_flag = get_utg_attributes(*ug, read_g, coverage_cut, sources, ruIndex); ///primary_flag = get_utg_attributes(*ug, read_g, coverage_cut, sources, ruIndex);
@@ -22643,15 +22654,24 @@ kvec_asg_arc_t_warp* new_rtg_edges)
} }
} }
if(asm_opt.recover_atg_cov_min == -1024 || asm_opt.recover_atg_cov_max == -1024) if(asm_opt.recover_atg_cov_min == -1024)
{ {
asm_opt.recover_atg_cov_max = asm_opt.hom_global_coverage/HOM_PEAK_RATE; asm_opt.recover_atg_cov_max = asm_opt.hom_global_coverage/HOM_PEAK_RATE;
asm_opt.recover_atg_cov_min = asm_opt.recover_atg_cov_max * 0.8; asm_opt.recover_atg_cov_min = asm_opt.recover_atg_cov_max * 0.8;
asm_opt.recover_atg_cov_max = asm_opt.recover_atg_cov_max * 1.2; ///asm_opt.recover_atg_cov_max = asm_opt.recover_atg_cov_max * 1.2;
asm_opt.recover_atg_cov_max = INT32_MAX;
} }
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, %d]\n", if(asm_opt.recover_atg_cov_max != INT32_MAX)
__func__, asm_opt.recover_atg_cov_min, asm_opt.recover_atg_cov_max); {
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, %d]\n",
__func__, asm_opt.recover_atg_cov_min, asm_opt.recover_atg_cov_max);
}
else
{
fprintf(stderr, "[M::%s] primary contig coverage range: [%d, infinity]\n",
__func__, asm_opt.recover_atg_cov_min);
}
recover_utg_by_coverage(ug, read_g, coverage_cut, sources, ruIndex); recover_utg_by_coverage(ug, read_g, coverage_cut, sources, ruIndex);
hifiasm.1
+3 -2
View File
@@ -193,11 +193,12 @@ This might be helpful when users want to get an optimized assembly by multiple r
with different parameters. with different parameters.
.TP .TP
.BI --pri-range \ INT,INT .BI --pri-range \ INT1[,INT2]
Min and max coverage cutoff of primary contigs. Min and max coverage cutoff of primary contigs.
Keep contigs with coverage in this range at p_ctg.gfa. Keep contigs with coverage in this range at p_ctg.gfa.
Inferred automatically in default. Inferred automatically in default.
Set -1,-1 to disable If INT2 is not specified, it is set to infinity.
Set -1 to disable
.SS Trio-partition options .SS Trio-partition options