diff --git a/align.c b/align.c index 3c015d5..6eb5919 100644 --- a/align.c +++ b/align.c @@ -29,26 +29,56 @@ static inline void mm_seq_rev(uint32_t len, uint8_t *seq) t = seq[i], seq[i] = seq[len - 1 - i], seq[len - 1 - i] = t; } +static void mm_update_extra(mm_extra_t *p, const uint8_t *qseq, const uint8_t *tseq, uint32_t n_cigar, uint32_t *cigar) +{ + uint32_t k, l, toff = 0, qoff = 0; + for (k = 0; k < n_cigar; ++k) { + uint32_t op = cigar[k]&0xf, len = cigar[k]>>4; + if (op == 0) { + for (l = 0; l < len; ++l) { + if (tseq[toff + l] > 3 || qseq[qoff + l] > 3) ++p->n_ambi; + else if (tseq[toff + l] != qseq[qoff + l]) ++p->n_diff; + } + toff += len, qoff += len, p->blen += len; + } else if (op == 1) { + int n_ambi = 0; + for (l = 0; l < len; ++l) + if (qseq[qoff + l] > 3) ++n_ambi; + qoff += len, p->blen += len; + p->n_ambi += n_ambi, p->n_diff += len - n_ambi; + } else if (op == 2) { + int n_ambi = 0; + for (l = 0; l < len; ++l) + if (tseq[toff + l] > 3) ++n_ambi; + toff += len, p->blen += len; + p->n_ambi += n_ambi, p->n_diff += len - n_ambi; + } + } +} + static void mm_append_cigar(mm_reg1_t *r, uint32_t n_cigar, uint32_t *cigar) // TODO: this calls the libc realloc() { + mm_extra_t *p; if (n_cigar == 0) return; - if (r->cigar == 0) { - uint32_t m_cigar = n_cigar + 2; - kroundup32(m_cigar); - r->cigar = (mm_cigar_t*)malloc(m_cigar * 4); - r->cigar->n = 0, r->cigar->m = m_cigar; - } else if (r->cigar->n + n_cigar > r->cigar->m - 2) { - r->cigar->m = r->cigar->n + n_cigar + 2; - kroundup32(r->cigar->m); - r->cigar = (mm_cigar_t*)realloc(r->cigar, r->cigar->m * 4); + if (r->p == 0) { + uint32_t capacity = n_cigar + sizeof(mm_extra_t); + kroundup32(capacity); + r->p = (mm_extra_t*)malloc(capacity * 4); + r->p->n_cigar = 0, r->p->capacity = capacity; + r->p->n_diff = r->p->n_ambi = 0; + } else if (r->p->n_cigar + n_cigar + sizeof(mm_extra_t) > r->p->capacity) { + r->p->capacity = r->p->n_cigar + n_cigar + sizeof(mm_extra_t); + kroundup32(r->p->capacity); + r->p = (mm_extra_t*)realloc(r->p, r->p->capacity * 4); } - if (r->cigar->n > 0 && (r->cigar->cigar[r->cigar->n-1]&0xf) == (cigar[0]&0xf)) { // same CIGAR op at the boundary - r->cigar->cigar[r->cigar->n-1] += cigar[0]>>4<<4; - if (n_cigar > 1) memcpy(r->cigar->cigar + r->cigar->n, cigar + 1, (n_cigar - 1) * 4); - r->cigar->n += n_cigar - 1; + p = r->p; + if (p->n_cigar > 0 && (p->cigar[p->n_cigar-1]&0xf) == (cigar[0]&0xf)) { // same CIGAR op at the boundary + p->cigar[p->n_cigar-1] += cigar[0]>>4<<4; + if (n_cigar > 1) memcpy(p->cigar + p->n_cigar, cigar + 1, (n_cigar - 1) * 4); + p->n_cigar += n_cigar - 1; } else { - memcpy(r->cigar->cigar + r->cigar->n, cigar, n_cigar * 4); - r->cigar->n += n_cigar; + memcpy(p->cigar + p->n_cigar, cigar, n_cigar * 4); + p->n_cigar += n_cigar; } } @@ -56,12 +86,13 @@ static void mm_align1(void *km, const mm_mapopt_t *opt, const mm_idx_t *mi, int { int32_t rid = a[r->as].x<<1>>33, rev = a[r->as].x>>63; uint8_t *tseq, *qseq; - int32_t i, k, l, rs0, re0, qs0, qe0; + int32_t i, l, bw, rs0, re0, qs0, qe0; int32_t rs, re, qs, qe; - int32_t rs1, qs1; + int32_t rs1, qs1, re1, qe1; int8_t mat[25]; ksw_gen_simple_mat(5, mat, opt->a, opt->b); + bw = (int)(opt->bw * 1.5 + 1.); rs = (int32_t)a[r->as].x + 1; // NB: this is the same as r->{rs,re} re = (int32_t)a[r->as + r->cnt - 1].x + 1; @@ -78,7 +109,7 @@ static void mm_align1(void *km, const mm_mapopt_t *opt, const mm_idx_t *mi, int } else rs0 = rs, qs0 = qs; if (qe < qlen && re < mi->seq[rid].len) { - l = qlen - re < opt->max_gap? qlen - re : opt->max_gap; + l = qlen - qe < opt->max_gap? qlen - qe : opt->max_gap; qe0 = qe + l; l += (l * opt->a - opt->q) / opt->e; l = l < opt->max_gap? l : opt->max_gap; @@ -86,7 +117,7 @@ static void mm_align1(void *km, const mm_mapopt_t *opt, const mm_idx_t *mi, int re0 = re + l; } else re0 = re, qe0 = qe; - tseq = (uint8_t*)kmalloc(km, re0 - rs0); + tseq = (uint8_t*)kmalloc(km, re0 - rs0); // TODO: we can allocate a smaller size if (qs > 0 && rs > 0) { // left extension uint32_t ql = qs - qs0, tl = rs - rs0; @@ -99,12 +130,15 @@ static void mm_align1(void *km, const mm_mapopt_t *opt, const mm_idx_t *mi, int for (k = 0; k < tl; ++k) fputc("ACGTN"[tseq[k]], stderr); fputc('\n', stderr); for (k = 0; k < ql; ++k) fputc("ACGTN"[qseq[k]], stderr); fputc('\n', stderr); #endif - ksw_extz2_sse(km, ql, qseq, tl, tseq, 5, mat, opt->q, opt->e, (int)(opt->bw * 1.5 + .499), opt->zdrop, KSW_EZ_EXTZ_ONLY|KSW_EZ_RIGHT|KSW_EZ_REV_CIGAR, ez); - mm_seq_rev(ql, qseq); + ksw_extz2_sse(km, ql, qseq, tl, tseq, 5, mat, opt->q, opt->e, bw, opt->zdrop, KSW_EZ_EXTZ_ONLY|KSW_EZ_RIGHT|KSW_EZ_REV_CIGAR, ez); mm_append_cigar(r, ez->n_cigar, ez->cigar); + mm_update_extra(r->p, qseq, tseq, ez->n_cigar, ez->cigar); + r->p->score += ez->score; rs1 = rs - (ez->max_t + 1); qs1 = qs - (ez->max_q + 1); + mm_seq_rev(ql, qseq); } else rs1 = rs, qs1 = qs; + assert(qs1 >= 0 && rs1 >= 0); for (i = 1; i < r->cnt; ++i) { // gap filling re = (int32_t)a[r->as + i].x + 1; @@ -117,21 +151,37 @@ static void mm_align1(void *km, const mm_mapopt_t *opt, const mm_idx_t *mi, int for (k = 0; k < re - rs; ++k) fputc("ACGTN"[tseq[k]], stderr); fputc('\n', stderr); for (k = 0; k < qe - qs; ++k) fputc("ACGTN"[qseq[k]], stderr); fputc('\n', stderr); #endif - ksw_extz2_sse(km, qe-qs, qseq, re-rs, tseq, 5, mat, opt->q, opt->e, (int)(opt->bw * 1.5 + .499), opt->zdrop, KSW_EZ_DYN_BAND, ez); + ksw_extz2_sse(km, qe-qs, qseq, re-rs, tseq, 5, mat, opt->q, opt->e, bw, opt->zdrop, KSW_EZ_DYN_BAND, ez); + mm_append_cigar(r, ez->n_cigar, ez->cigar); + mm_update_extra(r->p, qseq, tseq, ez->n_cigar, ez->cigar); if (ez->score == KSW_NEG_INF) { // truncated by Z-drop + r->p->score += ez->max; abort(); } else { - mm_append_cigar(r, ez->n_cigar, ez->cigar); + r->p->score += ez->score; } //for (k = 0; k < r->cigar->n; ++k) fprintf(stderr, "%d%c", r->cigar->cigar[k]>>4, "MID"[r->cigar->cigar[k]&0xf]); fputc('\n', stderr); rs = re, qs = qe; } } - if (i == r->cnt) { // right extension - } + if (i == r->cnt && qe < qe0 && re < re0) { // right extension + qseq = &qseq0[rev][qe]; + mm_idx_getseq(mi, rid, re, re0, tseq); + ksw_extz2_sse(km, qe0-qe, qseq, re0-re, tseq, 5, mat, opt->q, opt->e, bw, opt->zdrop, KSW_EZ_EXTZ_ONLY, ez); + mm_append_cigar(r, ez->n_cigar, ez->cigar); + mm_update_extra(r->p, qseq, tseq, ez->n_cigar, ez->cigar); + r->p->score += ez->score; + re1 = re + (ez->max_t + 1); + qe1 = qe + (ez->max_q + 1); + } else re1 = re, qe1 = qe; + assert(qe1 <= qlen); - for (i = 0; i < r->cigar->n; ++i) fprintf(stderr, "%d%c", r->cigar->cigar[i]>>4, "MID"[r->cigar->cigar[i]&0xf]); fputc('\n', stderr); + r->rs = rs1, r->re = re1; + if (rev) r->qs = qlen - qe1, r->qe = qlen - qs1; + else r->qs = qs1, r->qe = qe1; + +// for (i = 0; i < r->p->n_cigar; ++i) fprintf(stderr, "%d%c", r->p->cigar[i]>>4, "MID"[r->p->cigar[i]&0xf]); fputc('\n', stderr); kfree(km, tseq); } diff --git a/map.c b/map.c index 7839f86..9310ec8 100644 --- a/map.c +++ b/map.c @@ -16,7 +16,7 @@ void mm_mapopt_init(mm_mapopt_t *opt) opt->sdust_thres = 0; opt->min_score = 40; - opt->bw = 500; + opt->bw = 1000; opt->max_gap = 10000; opt->max_skip = 15; @@ -381,11 +381,14 @@ static void *worker_pipeline(void *shared, int step, void *in) printf("%s\t%d\t%d\t%d\t%c\t", t->name, t->l_seq, r->qs, r->qe, "+-"[r->rev]); if (mi->seq[r->rid].name) fputs(mi->seq[r->rid].name, stdout); else printf("%d", r->rid + 1); - printf("\t%d\t%d\t%d\t%d\t%d\t255\tcm:i:%d", mi->seq[r->rid].len, r->rs, r->re, r->score, - r->re - r->rs > r->qe - r->qs? r->re - r->rs : r->qe - r->qs, r->cnt); + printf("\t%d\t%d\t%d", mi->seq[r->rid].len, r->rs, r->re); + if (r->p) printf("\t%d\t%d\t255", r->p->blen - r->p->n_ambi - r->p->n_diff, r->p->blen); + else printf("\t%d\t%d\t255", r->score, r->re - r->rs > r->qe - r->qs? r->re - r->rs : r->qe - r->qs); + printf("\tcm:i:%d", r->cnt); if (r->parent == j) printf("\tss:i:%d", r->subsc); + if (r->p) printf("\tNM:i:%d\tAS:i:%d\tnn:i:%d", r->p->n_diff, r->p->score, r->p->n_ambi); putchar('\n'); - free(r->cigar); + free(r->p); } free(s->reg[i]); free(s->seq[i].seq); free(s->seq[i].name); diff --git a/minimap.h b/minimap.h index f28d80b..dbb0017 100644 --- a/minimap.h +++ b/minimap.h @@ -42,9 +42,13 @@ typedef struct { } mm_idx_t; typedef struct { - uint32_t n, m; + uint32_t capacity; + int32_t score; + uint32_t blen; + uint32_t n_diff, n_ambi; + uint32_t n_cigar; uint32_t cigar[]; -} mm_cigar_t; +} mm_extra_t; typedef struct { uint32_t cnt:31, rev:1; @@ -53,7 +57,7 @@ typedef struct { int32_t qs, qe, rs, re; int32_t parent, subsc; int32_t as; - mm_cigar_t *cigar; + mm_extra_t *p; } mm_reg1_t; typedef struct {