xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-10-02 23:08:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

regen_scb

Browse Source
This commit is contained in:
chhylp123
2026-05-03 03:34:23 -04:00
parent f5078f7b23
commit 7884b5ad88
15 changed files with 2177 additions and 133 deletions
Download Patch File Download Diff File Expand all files Collapse all files
anchor.cpp
+150
View File
@@ -4014,6 +4014,156 @@ void get_pi_ec_chain(ha_abuf_t *ab, uint64_t rid, uint64_t rl, uint32_t tid, cha
}
uint64_t gen_srt_chain(ha_abuf_t *ab, ha_mz1_t *rfa, uint64_t *rfi, uint64_t rfn, ha_mz1_t *qra, uint64_t *qri, uint64_t qrn, uint32_t *high_occ, uint32_t *low_occ, Candidates_list *cl, uint32_t qrid)
{
uint64_t i, k, l, max_cnt = UINT32_MAX, min_cnt = 0, ri, qi, sn, x, rz, qz, n_a0, l0, k0, i0; anchor1_t *an; k_mer_hit *p;
if(high_occ) {
max_cnt = (*high_occ);
if(max_cnt < 2) max_cnt = 2;
}
if(low_occ) {
min_cnt = (*low_occ);
if(min_cnt < 2) min_cnt = 2;
}
radix_sort_anc64(rfi, rfi + rfn);
radix_sort_anc64(qri, qri + qrn);
i = ab->n_a = 0; n_a0 = l0 = i0 = 0; k0 = 1;
for (k = 1, l = 0; k <= rfn; ++k) {
if (k == rfn || (rfi[k]>>32) != (rfi[l]>>32)) {
x = rfi[l]>>32;
for (; i < qrn && (qri[i]>>32) < x; i++);
if(ab->n_a < ab->m_a) {
n_a0 = ab->n_a; l0 = l; i0 = i; k0 = k;
}
if(i < qrn && (qri[i]>>32) == x) {
// sn = 0;
// if(ab->n_a < ab->m_a) sn = ab->seed[l].n;///ha_pt_cnt(ha_idx, ra[l].x);
for (qi = i; qi < qrn && (qri[qi]>>32) == x; qi++) {
qz = (uint32_t)qri[qi];
for (ri = l; ri < k; ri++) {
rz = (uint32_t)rfi[ri];
if(qra[qz].rev != rfa[rz].rev) continue;
if(ab->n_a < ab->m_a) {
an = &(ab->a[ab->n_a++]);
an->other_off = qra[qz].pos;
an->self_off = rfa[rz].pos;
///an->cnt: cnt<<8|span
an->cnt = ab->seed[rz].n; if(an->cnt > ((uint32_t)(0xffffffu))) an->cnt = 0xffffffu;
an->cnt <<= 8; an->cnt |= ((rfa[rz].span <= ((uint32_t)(0xffu)))?rfa[rz].span:((uint32_t)(0xffu)));
an->srt = (((uint64_t)(an->self_off))<<32)|((uint64_t)(an->other_off));
} else {
ab->n_a++;
}
}
}
i = qi;
}
l = k;
}
}
if (ab->n_a > ab->m_a) {
ab->m_a = ab->n_a; REALLOC(ab->a, ab->m_a);
k = k0; i = i0; l = l0; ab->n_a = n_a0;
for (; k <= rfn; ++k) {
if (k == rfn || (rfi[k]>>32) != (rfi[l]>>32)) {
x = rfi[l]>>32;
for (; i < qrn && (qri[i]>>32) < x; i++);
if(ab->n_a < ab->m_a) {
n_a0 = ab->n_a; l0 = l; i0 = i; k0 = k;
}
if(i < qrn && (qri[i]>>32) == x) {
// sn = 0;
// if(ab->n_a < ab->m_a) sn = ab->seed[l].n;///ha_pt_cnt(ha_idx, ra[l].x);
for (qi = i; qi < qrn && (qri[qi]>>32) == x; qi++) {
qz = (uint32_t)qri[qi];
for (ri = l; ri < k; ri++) {
rz = (uint32_t)rfi[ri];
if(qra[qz].rev != rfa[rz].rev) continue;
if(ab->n_a < ab->m_a) {
an = &(ab->a[ab->n_a++]);
an->other_off = qra[qz].pos;
an->self_off = rfa[rz].pos;
///an->cnt: cnt<<8|span
an->cnt = ab->seed[rz].n; if(an->cnt > ((uint32_t)(0xffffffu))) an->cnt = 0xffffffu;
an->cnt <<= 8; an->cnt |= ((rfa[rz].span <= ((uint32_t)(0xffu)))?rfa[rz].span:((uint32_t)(0xffu)));
an->srt = (((uint64_t)(an->self_off))<<32)|((uint64_t)(an->other_off));
} else {
ab->n_a++;
}
}
}
i = qi;
}
l = k;
}
}
}
// copy over to _cl_
sn = ab->n_a + cl->length;
if (sn > (uint64_t)cl->size) {
cl->size = sn;
REALLOC(cl->list, cl->size);
}
radix_sort_ha_an1(ab->a, ab->a + ab->n_a);
for (i = 0; i < ab->n_a; i++) {
p = &cl->list[cl->length++];
p->readID = qrid;
p->strand = 0;
p->offset = ab->a[i].other_off;
p->self_offset = ab->a[i].self_off;
if(((ab->a[i].cnt>>8) < max_cnt) && ((ab->a[i].cnt>>8) > min_cnt)){
p->cnt = 1;
} else if((ab->a[i].cnt>>8) <= min_cnt) {
p->cnt = 2;
} else{
p->cnt = 1 + (((ab->a[i].cnt>>8) + (max_cnt<<1) - 1)/(max_cnt<<1));
p->cnt = pow(p->cnt, 1.1);
}
if(p->cnt > ((uint32_t)(0xffffffu))) p->cnt = 0xffffffu;
p->cnt <<= 8; p->cnt |= (((uint32_t)(0xffu))&(ab->a[i].cnt));
}
// cl->length = ab->n_a;
return ab->n_a;
}
void gen_self_global_chain(ha_abuf_t *ab, Candidates_list *cl, uint32_t rid, uint64_t rl, uint32_t tid, char *ts, uint64_t tl, uint64_t mz_w, uint64_t mz_k, kvec_t_u8_warp* k_flag, kvec_t_u64_warp* dbg_ct, st_mt_t *sp, uint32_t *high_occ, uint32_t *low_occ, asg64_v *ix,
uint8_t is_accurate, double bw_thres, int apend_be, uint32_t gen_off, int64_t enable_mcopy, double mcopy_rate, uint32_t mcopy_khit_cut, overlap_region_alloc *ores)
{
extern void *ha_flt_tab;
uint64_t k, rn = ab->mz.n, m = cl->length;
mz1_ha_sketch(ts, tl, mz_w, mz_k, 0, !(asm_opt.flag & HA_F_NO_HPC), &ab->mz, ha_flt_tab, asm_opt.mz_sample_dist, k_flag, dbg_ct, NULL, -1, asm_opt.dp_min_len, -1, sp, asm_opt.mz_rewin, 0, NULL);
kv_resize(uint64_t, *ix, ab->mz.n);
for (k = 0; k < rn; k++) {
ix->a[k] = ab->mz.a[k].x;
ix->a[k] <<= 32; ix->a[k] |= k;
}
for (; k < ab->mz.n; k++) {
ix->a[k] = ab->mz.a[k].x;
ix->a[k] <<= 32; ix->a[k] |= (k-rn);
}
ix->n = ab->mz.n;
if(gen_srt_chain(ab, ab->mz.a, ix->a, rn, ab->mz.a + rn, ix->a + rn, ab->mz.n - rn, high_occ, low_occ, cl, tid)) {
int64_t max_skip, max_iter, max_dis, quick_check; double chn_pen_gap, chn_pen_skip;
set_lchain_dp_op(is_accurate, mz_k, &max_skip, &max_iter, &max_dis, &chn_pen_gap, &chn_pen_skip, &quick_check);
m += lchain_qdp_mcopy_fast(cl, m, cl->length - m, m, &(cl->chainDP), ores, max_skip, max_iter, max_dis, chn_pen_gap, chn_pen_skip, bw_thres,
rid, rl, tl, quick_check, apend_be, gen_off, enable_mcopy, mcopy_rate, mcopy_khit_cut, 1);
cl->length = m;
}
ab->mz.n = rn;
}
int64_t ug_map_lchain(ha_abufl_t *ab, uint32_t rid, char* rs, uint64_t rl, uint64_t mz_w, uint64_t mz_k, const ul_idx_t *uref, overlap_region_alloc *overlap_list, Candidates_list *cl, double bw_thres, double bw_thres_sec,
int max_n_chain, int apend_be, kvec_t_u8_warp* k_flag, overlap_region* f_cigar, kvec_t_u64_warp* dbg_ct, st_mt_t *sp, uint32_t *high_occ, uint32_t *low_occ, uint32_t is_accurate,
uint32_t gen_off, double mcopy_rate, uint32_t mcopy_khit_cut, uint32_t is_hpc, ha_mzl_t *res, uint64_t res_n, ha_mzl_t *idx, uint64_t idx_n, uint64_t mzl_cutoff, uint64_t chain_cutoff, kv_u_trans_t *kov)