xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-09-24 09:18:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

r878

Browse Source
This commit is contained in:
chhylp123
2026-03-28 11:02:52 -04:00
parent e53fc786bd
commit a137a6b06f
4 changed files with 259 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Correct.cpp
+186
View File
@@ -35989,6 +35989,192 @@ void gen_hc_r_alin_adv_adp_smp(gen_hc_aln_t *ez, uint32_t *a_cu, uint32_t *a_ci,
void gen_hc_r_alin_adv_adp_smp_0(gen_hc_aln_t *ez, uint8_t set_match)
{
uint64_t i, bs, k, w, ql = ez->qu->length, *wsrt = NULL, wsrt_n = 0, spn0 = 0, tot_b = 0; double err, e_max, align_gap_max, align_gap_rate, chem_drop; overlap_region t, *z; asg64_v *sp = ez->srt;
ez->ol->mapped_overlaps_length = 0; uint32_t *ocn = ez->v32->a, *osc = ez->v32->a + ez->ol->length;
if(ez->ol->length <= 0) return;
///base alignment
bs = (MAX(ez->wl[0], ez->wl[1]))+(THRESHOLD_MAX_SIZE<<1)+1;
resize_UC_Read(ez->tu, bs<<1); spn0 = sp->n;
wsrt = mmp_chn_select(ez->ol, ez->cl, sp, ez->ocw, ocn, osc, ql, &wsrt_n, set_match, ez->max_n_chain, ez->max_n_chain_f, ez->chain_cutoff, ez->ave_cov_min, 0.333333, 16, 16);
for (i = 0; i < wsrt_n; i++) {
z = &(ez->ol->list[(uint32_t)wsrt[i]]);
if(z->y_id < ez->t_cut) {
err = ez->e_rate[1]; w = ez->wl[1]; align_gap_max = ez->align_gap_max[1]; align_gap_rate = ez->align_gap_rate[1]; chem_drop = ez->chem_drop[1];
} else {
err = ez->e_rate[0]; w = ez->wl[0]; align_gap_max = ez->align_gap_max[0]; align_gap_rate = ez->align_gap_rate[0]; chem_drop = ez->chem_drop[0];
}
e_max = err * 1.5;
if((z->is_match == 0) && (!gen_hc_r_alin_flt_1_smp(z, ez->cl, ez->rref, ez->qu, ez->tu, ez->exz, ez->aux_o, err, e_max, err, w, ql, ez->rid, ez->khit, chem_drop, align_gap_rate, align_gap_max, NULL/**ez->hpz->a**/, ez->buf, 0, &tot_b))) {
continue;
}
z->is_match = 1; z->strong = z->without_large_indel = 0;
}
for (i = k = 0; i < ez->ol->length; i++) {///primary chain
z = &(ez->ol->list[i]);
if(z->is_match == 0) continue;
if(k != i) {
t = ez->ol->list[k];
ez->ol->list[k] = ez->ol->list[i];
ez->ol->list[i] = t;
}
k++;
}
ez->ol->length = k; sp->n = spn0;
}
void batch_simd_ck(gen_hc_aln_t *ez, uint64_t *wsrt, uint64_t wsrt_n, uint64_t ql, double err, uint64_t w, double align_gap_max, double align_gap_rate, double chem_drop)
{
if(!wsrt_n) return;
uint8_t rr; uint64_t fi[AVX_GS]; int32_t aux_beg[AVX_GS], aux_end[AVX_GS], t_s[AVX_GS], t_pri_l[AVX_GS];
uint64_t i, k, zk, in0, s, e, q[2], os, oe; overlap_region *z;
radix_sort_bc64(wsrt, wsrt + wsrt_n);
for (k = 1, i = 0; k <= wsrt_n; k++) {
if (k == wsrt_n || (wsrt[k]>>32) != (wsrt[i]>>32)) {
if(k - i > 1) {
for (zk = i; zk < k; zk++) {
wsrt[zk] = (((uint64_t)(ez->ol->list[(uint32_t)wsrt[zk]].x_pos_e + 1))<<32)|((uint32_t)wsrt[zk]);
}
radix_sort_bc64(wsrt + i, wsrt + k);
}
i = k;
}
}
ez->v32->n = (ez->ol->length<<1); in0 = ez->v32->n;
i = 0; s = 0; e = w; e = ((e<=ql)?e:ql); rr = 0;
for (; s < ql; ) {///[s, e)
if(rr) {
for (k = zk = in0; k < ez->v32->n; k++) {
if((wsrt[ez->v32->a[k]]>>32) == UINT32_MAX) continue;///passed
z = &(ez->ol->list[(uint32_t)wsrt[ez->v32->a[k]]]);
q[0] = z->x_pos_s;
q[1] = z->x_pos_e + 1;
///[s, e) && [q[0], q[1])
os = MAX(q[0], s); oe = MIN(q[1], e);
if(oe > os) ez->v32->a[zk++] = ez->v32->a[k];
}
ez->v32->n = zk;
}
for (; i < wsrt_n; ++i) {
z = &(ez->ol->list[(uint32_t)wsrt[i]]);
q[0] = z->x_pos_s;
q[1] = z->x_pos_e + 1;
if(q[0] >= e) break;
if(z->is_match == 0) {
os = MAX(q[0], s); oe = MIN(q[1], e);
if(oe > os) kv_push(uint32_t, *(ez->v32), i);
}
}
rr = hc_aln_simd(ez->ol->list, wsrt, ez->v32->a + in0, ez->v32->n - in0, s, e, w, err, ez->rref, ez->qu->seq, ez->tu, fi, aux_beg, aux_end, t_s, t_pri_l, ez->exz, OVERLAP_THRESHOLD_HIFI_FILTER, 0, NULL/**&tot_b**/);
s += w; e += w; e = ((e<=ql)?e:ql);
}
}
void gen_hc_r_alin_adv_adp_smp_1(gen_hc_aln_t *ez, uint8_t set_match)
{
uint64_t i, bs, k, w, ql = ez->qu->length, nol_1, *wsrt = NULL, wsrt_n = 0, spn0 = 0; double err, e_max, align_gap_max, align_gap_rate, chem_drop; overlap_region t, *z; asg64_v *sp = ez->srt;
ez->ol->mapped_overlaps_length = 0; uint32_t *ocn = ez->v32->a, *osc = ez->v32->a + ez->ol->length;
if(ez->ol->length <= 0) return;
///base alignment
bs = (MAX(ez->wl[0], ez->wl[1]))+(THRESHOLD_MAX_SIZE<<1)+1;
resize_UC_Read(ez->tu, bs<<1); spn0 = sp->n;
wsrt = mmp_chn_select(ez->ol, ez->cl, sp, ez->ocw, ocn, osc, ql, &wsrt_n, set_match, ez->max_n_chain, ez->max_n_chain_f, ez->chain_cutoff, ez->ave_cov_min, 0.333333, 16, 16);
for (i = nol_1 = 0; i < wsrt_n; i++) {
z = &(ez->ol->list[(uint32_t)wsrt[i]]);
if(z->is_match == 0) {
z->w_list.n = 0; z->align_length = 0;
}
wsrt[i] = ((((uint64_t)z->x_pos_s))<<32)|((uint32_t)wsrt[i]);
if(z->y_id < ez->t_cut) {
if(nol_1 != i) {
bs = wsrt[i]; wsrt[i] = wsrt[nol_1]; wsrt[nol_1] = bs;
}
nol_1++;
}
}
batch_simd_ck(ez, wsrt, nol_1, ql, ez->e_rate[1], ez->wl[1], ez->align_gap_max[1], ez->align_gap_rate[1], ez->chem_drop[1]);
batch_simd_ck(ez, wsrt + nol_1, wsrt_n - nol_1, ql, ez->e_rate[0], ez->wl[0], ez->align_gap_max[0], ez->align_gap_rate[0], ez->chem_drop[0]);
ocn = ez->v32->a; osc = ez->v32->a + ez->ol->length;
for (i = 0; i < wsrt_n; i++) {
if((wsrt[i]>>32) == UINT32_MAX) {
continue;
}
z = &(ez->ol->list[(uint32_t)wsrt[i]]);
if(z->y_id < ez->t_cut) {
err = ez->e_rate[1]; w = ez->wl[1]; align_gap_max = ez->align_gap_max[1]; align_gap_rate = ez->align_gap_rate[1]; chem_drop = ez->chem_drop[1];
} else {
err = ez->e_rate[0]; w = ez->wl[0]; align_gap_max = ez->align_gap_max[0]; align_gap_rate = ez->align_gap_rate[0]; chem_drop = ez->chem_drop[0];
}
e_max = err * 1.5;
if(z->is_match == 0) {
if((!pass_qovlp(z->x_pos_e+1-z->x_pos_s, z->align_length, OVERLAP_THRESHOLD_HIFI_FILTER)) || (!gen_hc_r_alin_flt_1_smp(z, ez->cl, ez->rref, ez->qu, ez->tu, ez->exz, ez->aux_o, err, e_max, err, w, ql, ez->rid, ez->khit, chem_drop,
align_gap_rate, align_gap_max, NULL/**ez->hpz->a**/, ez->buf, 1, NULL/**&tot_b**/))) {
continue;
}
}
// if((z->is_match == 0) && (!gen_hc_r_alin_flt_1_smp(z, ez->cl, ez->rref, ez->qu, ez->tu, ez->exz, ez->aux_o, err, e_max, err, w, ql, ez->rid, ez->khit, chem_drop, align_gap_rate, align_gap_max, NULL/**ez->hpz->a**/, ez->buf, 0, &tot_b))) {
// continue;
// }
z->is_match = 1; z->strong = z->without_large_indel = 0;
}
/**
// uint64_t kqs, kqe, kts, kte, kerr;
for (i = 0; i < wsrt_n; i++) {
if((wsrt[i]>>32) == UINT32_MAX) {
continue;
}
z = &(ol->list[(uint32_t)wsrt[i]]);
if(z->is_match == 0) {
if((!pass_qovlp(z->x_pos_e+1-z->x_pos_s, z->align_length, OVERLAP_THRESHOLD_HIFI_FILTER)) || (!gen_hc_r_alin_flt_1_smp(z, cl, rref, qu, tu, exz, aux_o, err, e_max,
e_rate, w.window_length, ql, rid, khit, chem_drop, align_gap_rate, align_gap_max, hpf, buf, 1, &tot_b))) {
continue;
}
}
z->is_match = 1; z->strong = z->without_large_indel = 0;
}
**/
for (i = k = 0; i < ez->ol->length; i++) {///primary chain
z = &(ez->ol->list[i]);
if(z->is_match == 0) continue;
if(k != i) {
t = ez->ol->list[k];
ez->ol->list[k] = ez->ol->list[i];
ez->ol->list[i] = t;
}
k++;
}
ez->ol->length = k; sp->n = spn0;
}
uint64_t gen_hc_r_alin_nec(overlap_region_alloc* ol, Candidates_list *cl, All_reads *rref, UC_Read* qu, UC_Read* tu, bit_extz_t *exz, overlap_region *aux_o, double e_rate, int64_t wl, int64_t rid, int64_t khit, int64_t move_gap, asg16_v* buf, uint8_t chem_drop, double align_gap_rate, int64_t align_gap_max, uint64_t sec_aln_win, uint64_t sec_aln_cov, double sec_aln_err_rate, double sec_aln_max, asg64_v *kp, uint8_t *hpf)
{