allow to read multiple files interleaved

This commit is contained in:
Heng Li
2017-09-24 14:33:05 -04:00
parent 84bbc47152
commit f0951141a1
4 changed files with 66 additions and 15 deletions
+33 -2
View File
@@ -54,11 +54,12 @@ static inline void kseq2bseq(kseq_t *ks, mm_bseq1_t *s, int with_qual)
s->l_seq = ks->seq.l;
}
mm_bseq1_t *mm_bseq_read(mm_bseq_file_t *fp, int chunk_size, int with_qual, int *n_)
mm_bseq1_t *mm_bseq_read2(mm_bseq_file_t *fp, int chunk_size, int with_qual, int check_name, int *n_)
{
int64_t size = 0;
kvec_t(mm_bseq1_t) a = {0,0,0};
kseq_t *ks = fp->ks;
*n_ = 0;
if (fp->s.seq) {
kv_resize(mm_bseq1_t, 0, a, 256);
kv_push(mm_bseq1_t, 0, a, fp->s);
@@ -73,7 +74,7 @@ mm_bseq1_t *mm_bseq_read(mm_bseq_file_t *fp, int chunk_size, int with_qual, int
kseq2bseq(ks, s, with_qual);
size += s->l_seq;
if (size >= chunk_size) {
if (a.a[a.n-1].l_seq < CHECK_PAIR_THRES) {
if (check_name && a.a[a.n-1].l_seq < CHECK_PAIR_THRES) {
while (kseq_read(ks) >= 0) {
kseq2bseq(ks, &fp->s, with_qual);
if (mm_qname_same(fp->s.name, a.a[a.n-1].name)) {
@@ -89,6 +90,36 @@ mm_bseq1_t *mm_bseq_read(mm_bseq_file_t *fp, int chunk_size, int with_qual, int
return a.a;
}
mm_bseq1_t *mm_bseq_read(mm_bseq_file_t *fp, int chunk_size, int with_qual, int *n_)
{
return mm_bseq_read2(fp, chunk_size, with_qual, 0, n_);
}
mm_bseq1_t *mm_bseq_read_multi(int n_fp, mm_bseq_file_t **fp, int chunk_size, int with_qual, int *n_)
{
int i;
int64_t size = 0;
kvec_t(mm_bseq1_t) a = {0,0,0};
*n_ = 0;
if (n_fp < 1) return 0;
while (1) {
for (i = 0; i < n_fp; ++i)
if (kseq_read(fp[i]->ks) < 0)
break;
if (i != n_fp) break; // some file reaches the end
if (a.m == 0) kv_resize(mm_bseq1_t, 0, a, 256);
for (i = 0; i < n_fp; ++i) {
mm_bseq1_t *s;
kv_pushp(mm_bseq1_t, 0, a, &s);
kseq2bseq(fp[i]->ks, s, with_qual);
size += s->l_seq;
}
if (size >= chunk_size) break;
}
*n_ = a.n;
return a.a;
}
int mm_bseq_eof(mm_bseq_file_t *fp)
{
return (ks_eof(fp->ks->f) && fp->s.seq == 0);