mem_opt_t *mem_opt_init()
{
mem_opt_t *o;
o = calloc(1, sizeof(mem_opt_t));
o->flag = 0;
o->a = 1; o->b = 4;
o->o_del = o->o_ins = 6;
o->e_del = o->e_ins = 1;
o->w = 100;
o->T = 30;
o->zdrop = 100;
o->pen_unpaired = 17;
o->pen_clip5 = o->pen_clip3 = 5;
o->min_seed_len = 19;
o->split_width = 10;
o->max_occ = 500;
o->max_chain_gap = 10000;
o->max_ins = 10000;
o->mask_level = 0.50;
o->drop_ratio = 0.50;
o->XA_drop_ratio = 0.80;
o->split_factor = 1.5;
o->chunk_size = 10000000;
o->n_threads = 1;
o->max_hits = 5;
o->max_matesw = 50;
o->mask_level_redun = 0.95;
o->min_chain_weight = 0;
o->max_chain_extend = 1<<30;
o->mapQ_coef_len = 50; o->mapQ_coef_fac = log(o->mapQ_coef_len);
bwa_fill_scmat(o->a, o->b, o->mat);
return o;
}
void bsw2_extend_rght(const bsw2opt_t *opt, bwtsw2_t *b, uint8_t *query, int lq, uint8_t *pac, bwtint_t l_pac, uint8_t *_mem)
{
int i;
bwtint_t k;
uint8_t *target;
int8_t mat[25];
bwa_fill_scmat(opt->a, opt->b, mat);
target = calloc(((lq + 1) / 2 * opt->a + opt->r) / opt->r + lq, 1);
for (i = 0; i < b->n; ++i) {
bsw2hit_t *p = b->hits + i;
int lt = ((lq - p->beg + 1) / 2 * opt->a + opt->r) / opt->r + lq;
int j, score, qle, tle;
if (p->l) continue;
for (k = p->k, j = 0; k < p->k + lt && k < l_pac; ++k)
target[j++] = pac[k>>2] >> (~k&3)*2 & 0x3;
lt = j;
score = ksw_extend(lq - p->beg, &query[p->beg], lt, target, 5, mat, opt->q, opt->r, opt->bw, 0, -1, 1, &qle, &tle, 0, 0, 0) - 1;
// if (score < p->G) fprintf(stderr, "[bsw2_extend_hits] %d < %d\n", score, p->G);
if (score >= p->G) {
p->G = score;
p->len = tle;
p->end = p->beg + qle;
}
}
free(target);
}
void bsw2_extend_left(const bsw2opt_t *opt, bwtsw2_t *b, uint8_t *_query, int lq, uint8_t *pac, bwtint_t l_pac, uint8_t *_mem)
{
int i;
bwtint_t k;
uint8_t *target = 0, *query;
int8_t mat[25];
bwa_fill_scmat(opt->a, opt->b, mat);
query = calloc(lq, 1);
// sort according to the descending order of query end
ks_introsort(hit, b->n, b->hits);
target = calloc(((lq + 1) / 2 * opt->a + opt->r) / opt->r + lq, 1);
// reverse _query
for (i = 0; i < lq; ++i) query[lq - i - 1] = _query[i];
// core loop
for (i = 0; i < b->n; ++i) {
bsw2hit_t *p = b->hits + i;
int lt = ((p->beg + 1) / 2 * opt->a + opt->r) / opt->r + lq;
int score, j, qle, tle;
p->n_seeds = 1;
if (p->l || p->k == 0) continue;
for (j = score = 0; j < i; ++j) {
bsw2hit_t *q = b->hits + j;
if (q->beg <= p->beg && q->k <= p->k && q->k + q->len >= p->k + p->len) {
if (q->n_seeds < (1<<13) - 2) ++q->n_seeds;
++score;
}
}
if (score) continue;
if (lt > p->k) lt = p->k;
for (k = p->k - 1, j = 0; k > 0 && j < lt; --k) // FIXME: k=0 not considered!
target[j++] = pac[k>>2] >> (~k&3)*2 & 0x3;
lt = j;
score = ksw_extend(p->beg, &query[lq - p->beg], lt, target, 5, mat, opt->q, opt->r, opt->bw, 0, -1, p->G, &qle, &tle, 0, 0, 0);
if (score > p->G) { // extensible
p->G = score;
p->k -= tle;
p->len += tle;
p->beg -= qle;
}
}
free(query); free(target);
}
/* generate CIGAR array(s) in b->cigar[] */
static void gen_cigar(const bsw2opt_t *opt, int lq, uint8_t *seq[2], int64_t l_pac, const uint8_t *pac, bwtsw2_t *b, const char *name)
{
int i;
int8_t mat[25];
bwa_fill_scmat(opt->a, opt->b, mat);
for (i = 0; i < b->n; ++i) {
bsw2hit_t *p = b->hits + i;
bsw2aux_t *q = b->aux + i;
uint8_t *query;
int beg, end, score;
if (p->l) continue;
beg = (p->flag & 0x10)? lq - p->end : p->beg;
end = (p->flag & 0x10)? lq - p->beg : p->end;
query = seq[(p->flag & 0x10)? 1 : 0] + beg;
q->cigar = bwa_gen_cigar(mat, opt->q, opt->r, opt->bw, l_pac, pac, end - beg, query, p->k, p->k + p->len, &score, &q->n_cigar, &q->nm);
#if 0
if (name && score != p->G) { // debugging only
int j, glen = 0;
for (j = 0; j < q->n_cigar; ++j)
if ((q->cigar[j]&0xf) == 1 || (q->cigar[j]&0xf) == 2)
glen += q->cigar[j]>>4;
fprintf(stderr, "[E::%s] %s - unequal score: %d != %d; (qlen, aqlen, arlen, glen, bw) = (%d, %d, %d, %d, %d)\n",
__func__, name, score, p->G, lq, end - beg, p->len, glen, opt->bw);
}
#endif
if (q->cigar && (beg != 0 || end < lq)) { // write soft clipping
q->cigar = realloc(q->cigar, 4 * (q->n_cigar + 2));
if (beg != 0) {
memmove(q->cigar + 1, q->cigar, q->n_cigar * 4);
q->cigar[0] = beg<<4 | 4;
++q->n_cigar;
}
if (end < lq) {
q->cigar[q->n_cigar] = (lq - end)<<4 | 4;
++q->n_cigar;
}
}
}