6 static void replace_cigar(bam1_t *b, int n, uint32_t *cigar)
8 if (n != b->core.n_cigar) {
9 int o = b->core.l_qname + b->core.n_cigar * 4;
10 if (b->data_len + (n - b->core.n_cigar) * 4 > b->m_data) {
11 b->m_data = b->data_len + (n - b->core.n_cigar) * 4;
12 kroundup32(b->m_data);
13 b->data = (uint8_t*)realloc(b->data, b->m_data);
15 memmove(b->data + b->core.l_qname + n * 4, b->data + o, b->data_len - o);
16 memcpy(b->data + b->core.l_qname, cigar, n * 4);
17 b->data_len += (n - b->core.n_cigar) * 4;
19 } else memcpy(b->data + b->core.l_qname, cigar, n * 4);
22 #define write_cigar(_c, _n, _m, _v) do { \
25 _c = (uint32_t*)realloc(_c, _m * 4); \
30 static void unpad_seq(bam1_t *b, kstring_t *s)
33 uint32_t *cigar = bam1_cigar(b);
34 uint8_t *seq = bam1_seq(b);
35 ks_resize(s, b->core.l_qseq);
36 for (k = 0, s->l = 0, j = 0; k < b->core.n_cigar; ++k) {
38 op = bam_cigar_op(cigar[k]);
39 ol = bam_cigar_oplen(cigar[k]);
40 assert(op == BAM_CMATCH || op == BAM_CDEL || op == BAM_CSOFT_CLIP);
41 if (op == BAM_CMATCH) {
42 for (i = 0; i < ol; ++i) s->s[s->l++] = bam1_seqi(seq, j);
44 } else if (op == BAM_CSOFT_CLIP) {
47 for (i = 0; i < ol; ++i) s->s[s->l++] = 0;
52 int bam_pad2unpad(bamFile in, bamFile out)
58 int n2 = 0, m2 = 0, *posmap = 0;
60 h = bam_header_read(in);
61 bam_header_write(out, h);
63 r.l = r.m = q.l = q.m = 0; r.s = q.s = 0;
64 while (bam_read1(in, b) >= 0) {
65 uint32_t *cigar = bam1_cigar(b);
67 if (b->core.pos == 0 && b->core.tid >= 0 && strcmp(bam1_qname(b), h->target_name[b->core.tid]) == 0) {
70 write_cigar(cigar2, n2, m2, bam_cigar_gen(b->core.l_qseq, BAM_CMATCH));
71 replace_cigar(b, n2, cigar2);
72 posmap = realloc(posmap, r.m * sizeof(int));
73 for (i = k = 0; i < r.l; ++i) {
74 posmap[i] = k; // note that a read should NOT start at a padding
80 if (bam_cigar_op(cigar[0]) == BAM_CSOFT_CLIP) write_cigar(cigar2, n2, m2, cigar[0]);
81 for (i = 0, k = b->core.pos; i < q.l; ++i, ++k)
82 q.s[i] = q.s[i]? (r.s[k]? BAM_CMATCH : BAM_CINS) : (r.s[k]? BAM_CDEL : BAM_CPAD);
83 for (i = k = 1, op = q.s[0]; i < q.l; ++i) {
85 write_cigar(cigar2, n2, m2, bam_cigar_gen(k, op));
89 write_cigar(cigar2, n2, m2, bam_cigar_gen(k, op));
90 if (bam_cigar_op(cigar[b->core.n_cigar-1]) == BAM_CSOFT_CLIP) write_cigar(cigar2, n2, m2, cigar[b->core.n_cigar-1]);
91 for (i = 2; i < n2; ++i)
92 if (bam_cigar_op(cigar2[i]) == BAM_CMATCH && bam_cigar_op(cigar2[i-1]) == BAM_CPAD && bam_cigar_op(cigar2[i-2]) == BAM_CMATCH)
93 cigar2[i] += cigar2[i-2], cigar2[i-2] = cigar2[i-1] = 0;
94 for (i = k = 0; i < n2; ++i)
95 if (cigar2[i]) cigar2[k++] = cigar2[i];
97 replace_cigar(b, n2, cigar2);
98 b->core.pos = posmap[b->core.pos];
102 free(r.s); free(q.s); free(posmap);
104 bam_header_destroy(h);
108 int main_pad2unpad(int argc, char *argv[])
112 fprintf(stderr, "Usage: samtools depad <in.bam>\n");
115 in = strcmp(argv[1], "-")? bam_open(argv[1], "r") : bam_dopen(fileno(stdin), "r");
116 out = bam_dopen(fileno(stdout), "w");
117 bam_pad2unpad(in, out);
118 bam_close(in); bam_close(out);