void bwa_cal_pac_pos(const char *prefix, int n_seqs, bwa_seq_t *seqs, int max_mm, float fnr)
{
int i;
char str[1024];
bwt_t *bwt;
// load forward SA
strcpy(str, prefix); strcat(str, ".bwt"); bwt = bwt_restore_bwt(str);
strcpy(str, prefix); strcat(str, ".sa"); bwt_restore_sa(str, bwt);
for (i = 0; i != n_seqs; ++i) {
bwa_seq_t *p = seqs + i;
int max_diff = fnr > 0.0? bwa_cal_maxdiff(p->len, BWA_AVG_ERR, fnr) : max_mm;
if ((p->type == BWA_TYPE_UNIQUE || p->type == BWA_TYPE_REPEAT) && p->strand) { // reverse strand only
p->pos = bwt_sa(bwt, p->sa);
p->seQ = p->mapQ = bwa_approx_mapQ(p, max_diff);
}
}
bwt_destroy(bwt);
// load reverse BWT and SA
strcpy(str, prefix); strcat(str, ".rbwt"); bwt = bwt_restore_bwt(str);
strcpy(str, prefix); strcat(str, ".rsa"); bwt_restore_sa(str, bwt);
for (i = 0; i != n_seqs; ++i) {
bwa_seq_t *p = seqs + i;
int max_diff = fnr > 0.0? bwa_cal_maxdiff(p->len, BWA_AVG_ERR, fnr) : max_mm;
if ((p->type == BWA_TYPE_UNIQUE || p->type == BWA_TYPE_REPEAT) && !p->strand) { // forward strand only
/* NB: For gapped alignment, p->pos may not be correct,
* which will be fixed in refine_gapped_core(). This
* line also determines the way "x" is calculated in
* refine_gapped_core() when (ext < 0 && is_end == 0). */
p->pos = bwt->seq_len - (bwt_sa(bwt, p->sa) + p->len);
p->seQ = p->mapQ = bwa_approx_mapQ(p, max_diff);
}
}
bwt_destroy(bwt);
}
void bwa_cal_pac_pos(const char *prefix, int n_seqs, bwa_seq_t *seqs, int max_mm, float fnr)
{
int i, j;
char str[1024];
bwt_t *bwt;
// load forward SA
strcpy(str, prefix); strcat(str, ".bwt"); bwt = bwt_restore_bwt(str);
strcpy(str, prefix); strcat(str, ".sa"); bwt_restore_sa(str, bwt);
for (i = 0; i != n_seqs; ++i) {
if (seqs[i].strand) bwa_cal_pac_pos_core(bwt, 0, &seqs[i], max_mm, fnr);
for (j = 0; j < seqs[i].n_multi; ++j) {
bwt_multi1_t *p = seqs[i].multi + j;
if (p->strand) p->pos = bwt_sa(bwt, p->pos);
}
}
bwt_destroy(bwt);
// load reverse BWT and SA
strcpy(str, prefix); strcat(str, ".rbwt"); bwt = bwt_restore_bwt(str);
strcpy(str, prefix); strcat(str, ".rsa"); bwt_restore_sa(str, bwt);
for (i = 0; i != n_seqs; ++i) {
if (!seqs[i].strand) bwa_cal_pac_pos_core(0, bwt, &seqs[i], max_mm, fnr);
for (j = 0; j < seqs[i].n_multi; ++j) {
bwt_multi1_t *p = seqs[i].multi + j;
if (!p->strand) p->pos = bwt->seq_len - (bwt_sa(bwt, p->pos) + seqs[i].len);
}
}
bwt_destroy(bwt);
}
void bwa_cal_pac_pos_core(const bwt_t *forward_bwt, const bwt_t *reverse_bwt, bwa_seq_t *seq, const int max_mm, const float fnr)
{
int max_diff;
if (seq->type != BWA_TYPE_UNIQUE && seq->type != BWA_TYPE_REPEAT) return;
max_diff = fnr > 0.0? bwa_cal_maxdiff(seq->len, BWA_AVG_ERR, fnr) : max_mm;
if (seq->strand) { // reverse strand only
seq->pos = bwt_sa(forward_bwt, seq->sa);
} else { // forward strand only
/* NB: For gapped alignment, p->pos may not be correct, which
* will be fixed in refine_gapped_core(). This line also
* determines the way "x" is calculated in
* refine_gapped_core() when (ext < 0 && is_end == 0). */
seq->pos = reverse_bwt->seq_len - (bwt_sa(reverse_bwt, seq->sa) + seq->len);
}
seq->seQ = seq->mapQ = bwa_approx_mapQ(seq, max_diff);
}
void jigsaw_cal_pac_pos (const bwt_t *bwt, bwa_seq_t *seq, int max_mm, float fnr, const int *g_log_n)
{
int j;
bwa_cal_pac_pos_core(bwt, seq, max_mm, fnr, g_log_n);
for (j = 0; j < seq->n_multi; ++j) {
bwt_multi1_t *p = seq->multi + j;
p->pos = bwt_sa(bwt, p->pos);
}
}
void bwa_cal_pac_pos_core(const bwt_t *bwt, bwa_seq_t *seq, const int max_mm, const float fnr, const int *g_log_n)
{
int max_diff;
if (seq->type != BWA_TYPE_UNIQUE && seq->type != BWA_TYPE_REPEAT) return;
max_diff = fnr > 0.0? bwa_cal_maxdiff(seq->len, BWA_AVG_ERR, fnr) : max_mm;
seq->pos = bwt_sa(bwt, seq->sa);
seq->seQ = seq->mapQ = bwa_approx_mapQ(seq, max_diff, g_log_n);
}
bwtint_t bwa_sa2pos(const bntseq_t *bns, const bwt_t *bwt, bwtint_t sapos, int len, int *strand)
{
bwtint_t pos_f;
int is_rev;
pos_f = bns_depos(bns, bwt_sa(bwt, sapos), &is_rev); // pos_f
*strand = !is_rev;
/* NB: For gapped alignment, pacpos may not be correct, which will be fixed
* in bwa_refine_gapped_core(). This line also determines the way "x" is
* calculated in bwa_refine_gapped_core() when (ext < 0 && is_end == 0). */
if (is_rev) pos_f = pos_f + 1 < len? 0 : pos_f - len + 1; // mapped to the forward strand
return pos_f; // FIXME: it is possible that pos_f < bns->anns[ref_id].offset
}
/* calculate the reference genomic positions for an SA interval (an alignment)
* the allocated memory will be released at jigsaw_free_read_seq_one
*/
void jigsaw_collect_word_hits_sai (const bwt_t *bwt, int wid, int pos_q, const bwt_aln1_t *p, list<jigsaw_word_hit_t*> *hits)
{
int j;
//jigsaw_word_hit_t *h;
int len = p->l - p->k+1;
if (len <= 0) return;
for (j = 0; j < len; ++j) {
jigsaw_word_hit_t *h = (jigsaw_word_hit_t *) calloc(1, sizeof(jigsaw_word_hit_t));
//h = hits + j;
h->wid = wid;
h->pos_q = pos_q;
h->pos_t = bwt_sa (bwt, p->k + j);
h->strand = p->a;
hits->push_back (h);
}
}
