xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-09-15 12:47:57 +08:00
Code Issues Packages Projects Releases Wiki Activity

update -z

Browse Source
This commit is contained in:
chhylp123
2021-02-02 12:45:50 -05:00
parent 02b9a3e109
commit fda13cb0e4
5 changed files with 615 additions and 339 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
Assembly.cpp
View File

@@ -848,7 +848,8 @@ void ha_overlap_and_correct(int round)
b[i] = ha_ovec_init(0, (round == asm_opt.number_of_round - 1));
if(ha_idx) hom_cov = asm_opt.hom_cov;
if(ha_idx == NULL) ha_idx = ha_pt_gen(&asm_opt, ha_flt_tab, round == 0? 0 : 1, 0, &R_INF, &hom_cov, &het_cov); // build the index
if (round == 0 && ha_flt_tab == 0) // then asm_opt.hom_cov hasn't been updated
///debug_adapter(&asm_opt, &R_INF);
if (round == 0 && ha_flt_tab == 0) // then asm_opt.hom_cov hasn't been updated
ha_opt_update_cov(&asm_opt, hom_cov);
if (asm_opt.required_read_name)
kt_for(asm_opt.thread_num, worker_ovec_related_reads, b, R_INF.total_reads);
@@ -1600,6 +1601,8 @@ void ha_overlap_final(void)
asm_opt.het_cov = het_cov;
}
int ha_assemble(void)
{
extern void ha_extract_print_list(const All_reads *rs, int n_rounds, const char *o);

4
Correct.h
View File

@@ -21,8 +21,8 @@
///#define FLAG_THRE 0
#define MAX(x, y) ((x >= y)?(x):(y))
#define MIN(x, y) ((x <= y)?(x):(y))
#define MAX(x, y) (((x) >= (y))?(x):(y))
#define MIN(x, y) (((x) <= (y))?(x):(y))
#define DIFF(x, y) ((MAX((x), (y))) - (MIN((x), (y))))
#define OVERLAP(x_start, x_end, y_start, y_end) (MIN(x_end, y_end) - MAX(x_start, y_start) + 1)
///#define OVERLAP(x_start, x_end, y_start, y_end) MIN(x_end, y_end) - MAX(x_start, y_start) + 1

890
hic.cpp
View File

File diff suppressed because it is too large Load Diff

54
htab.cpp
View File

@@ -49,6 +49,7 @@ typedef struct {
int32_t pre;
int32_t n_thread;
int64_t chunk_size;
int adaLen;
} yak_copt_t;
void yak_copt_init(yak_copt_t *o)
@@ -598,7 +599,9 @@ static void *worker_count(void *data, int step, void *in) // callback for kt_pip
}
} else {
while ((ret = kseq_read(p->ks)) >= 0) {
int l = p->ks->seq.l;
int l = (int)(p->ks->seq.l) - (int)(p->opt->adaLen) - (int)(p->opt->adaLen);
if(l <= 0) continue;
if (p->n_seq >= 1<<28) {
fprintf(stderr, "ERROR: this implementation supports no more than %d reads\n", 1<<28);
exit(1);
@@ -612,9 +615,9 @@ static void *worker_count(void *data, int step, void *in) // callback for kt_pip
int i, n_N;
assert(l == (int)p->rs_out->read_length[p->n_seq]);
for (i = n_N = 0; i < l; ++i) // count number of ambiguous bases
if (seq_nt4_table[(uint8_t)p->ks->seq.s[i]] >= 4)
if (seq_nt4_table[(uint8_t)p->ks->seq.s[i+p->opt->adaLen]] >= 4)
++n_N;
ha_compress_base(Get_READ(*p->rs_out, p->n_seq), p->ks->seq.s, l, &p->rs_out->N_site[p->n_seq], n_N);
ha_compress_base(Get_READ(*p->rs_out, p->n_seq), p->ks->seq.s+p->opt->adaLen, l, &p->rs_out->N_site[p->n_seq], n_N);
memcpy(&p->rs_out->name[p->rs_out->name_index[p->n_seq]], p->ks->name.s, p->ks->name.l);
}
}
@@ -625,7 +628,7 @@ static void *worker_count(void *data, int step, void *in) // callback for kt_pip
REALLOC(s->seq, s->m_seq);
}
MALLOC(s->seq[s->n_seq], l);
memcpy(s->seq[s->n_seq], p->ks->seq.s, l);
memcpy(s->seq[s->n_seq], p->ks->seq.s+p->opt->adaLen, l);
s->len[s->n_seq++] = l;
++p->n_seq;
s->sum_len += l;
@@ -720,6 +723,48 @@ static void *worker_count(void *data, int step, void *in) // callback for kt_pip
return 0;
}
void debug_adapter(const hifiasm_opt_t *asm_opt, All_reads *rs)
{
int ret;
uint32_t i, m, pass, unpass;
gzFile fp = 0;
kseq_t *ks = NULL;
UC_Read ucr;
init_UC_Read(&ucr);
for (i = m = pass = unpass = 0; i < (uint32_t)asm_opt->num_reads; ++i)
{
if ((fp = gzopen(asm_opt->read_file_names[i], "r")) == 0) continue;
ks = kseq_init(fp);
while ((ret = kseq_read(ks)) >= 0)
{
int l = ks->seq.l;
if((l - asm_opt->adapterLen*2) <= 0) continue;
recover_UC_Read(&ucr, rs, m);
fprintf(stderr, "l: %d, ucr.length: %lld, asm_opt->adapterLen: %d\n",
l, ucr.length, asm_opt->adapterLen);
if(memcmp(ucr.seq, ks->seq.s+asm_opt->adapterLen, ucr.length) == 0)
{
pass++;
}
else
{
unpass++;
}
m++;
}
kseq_destroy(ks);
gzclose(fp);
ks = NULL;
fp = 0;
}
destory_UC_Read(&ucr);
fprintf(stderr, "[M::%s::# reads: %u, # pass: %u, # unpass: %u\n]", __func__, m, pass, unpass);
exit(1);
}
static ha_ct_t *yak_count(const yak_copt_t *opt, const char *fn, int flag, ha_pt_t *p0, ha_ct_t *c0, const void *flt_tab, All_reads *rs, int64_t *n_seq)
{
///for 0-th counting, flag = HAF_COUNT_ALL|HAF_RS_WRITE_LEN|HAF_CREATE_NEW
@@ -791,6 +836,7 @@ ha_ct_t *ha_count(const hifiasm_opt_t *asm_opt, int flag, ha_pt_t *p0, const voi
///for ha_pt_gen, shoud be 0
opt.bf_shift = flag & HAF_COUNT_EXACT? 0 : asm_opt->bf_shift;
opt.n_thread = asm_opt->thread_num;
opt.adaLen = asm_opt->adapterLen;
///asm_opt->num_reads is the number of fastq files
for (i = 0; i < asm_opt->num_reads; ++i)
h = yak_count(&opt, asm_opt->read_file_names[i], flag|HAF_CREATE_NEW, p0, h, flt_tab, rs, &n_seq);

1
htab.h
View File

@@ -62,6 +62,7 @@ void ha_triobin(const hifiasm_opt_t *opt);
void ha_sketch(const char *str, int len, int w, int k, uint32_t rid, int is_hpc, ha_mz1_v *p, const void *hf);
void ha_sketch_query(const char *str, int len, int w, int k, uint32_t rid, int is_hpc, ha_mz1_v *p, const void *hf, kvec_t_u8_warp* k_flag, kvec_t_u64_warp* dbg_ct);
int ha_analyze_count(int n_cnt, int start_cnt, const int64_t *cnt, int *peak_het);
void debug_adapter(const hifiasm_opt_t *asm_opt, All_reads *rs);
static inline uint64_t yak_hash64(uint64_t key, uint64_t mask) // invertible integer hash function
{