7 int bcf_hdr_read(bcf_t *b);
9 void bcf_hdr_clear(bcf_hdr_t *b)
11 free(b->name); free(b->sname); free(b->txt); free(b->ns); free(b->sns);
12 memset(b, 0, sizeof(bcf_hdr_t));
15 bcf_t *bcf_open(const char *fn, const char *mode)
18 b = calloc(1, sizeof(bcf_t));
19 if (strchr(mode, 'w')) {
20 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), "w");
22 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), "r");
24 b->fp->owned_file = 1;
25 if (strchr(mode, 'r')) bcf_hdr_read(b);
29 int bcf_close(bcf_t *b)
33 ret = bgzf_close(b->fp);
34 free(b->h.name); free(b->h.sname); free(b->h.txt); free(b->h.ns); free(b->h.sns);
39 int bcf_hdr_write(bcf_t *b)
42 if (b == 0) return -1;
44 bgzf_write(b->fp, "BCF\4", 4);
45 bgzf_write(b->fp, &h->l_nm, 4);
46 bgzf_write(b->fp, h->name, h->l_nm);
47 bgzf_write(b->fp, &h->l_smpl, 4);
48 bgzf_write(b->fp, h->sname, h->l_smpl);
49 bgzf_write(b->fp, &h->l_txt, 4);
50 bgzf_write(b->fp, h->txt, h->l_txt);
52 return 16 + h->l_nm + h->l_smpl + h->l_txt;
55 int bcf_hdr_read(bcf_t *b)
59 if (b == 0) return -1;
62 bgzf_read(b->fp, magic, 4);
63 bgzf_read(b->fp, &h->l_nm, 4);
64 h->name = malloc(h->l_nm);
65 bgzf_read(b->fp, h->name, h->l_nm);
66 bgzf_read(b->fp, &h->l_smpl, 4);
67 h->sname = malloc(h->l_smpl);
68 bgzf_read(b->fp, h->sname, h->l_smpl);
69 bgzf_read(b->fp, &h->l_txt, 4);
70 h->txt = malloc(h->l_txt);
71 bgzf_read(b->fp, h->txt, h->l_txt);
73 return 16 + h->l_nm + h->l_smpl + h->l_txt;
76 static inline char **cnt_null(int l, char *str, int *_n)
81 if (l == 0 || str == 0) return 0;
82 for (p = str; p != str + l; ++p)
85 list = calloc(n, sizeof(void*));
87 for (p = str, n = 1; p < str + l - 1; ++p)
88 if (*p == 0) list[n++] = p + 1;
92 int bcf_hdr_sync(bcf_hdr_t *b)
94 if (b == 0) return -1;
95 b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
96 b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl);
100 #define char2int(s) (((int)s[0])<<8|s[1])
102 int bcf_sync(int n_smpl, bcf1_t *b)
104 char *p, *tmp[5], *s;
106 // set ref, alt, flt, info, fmt
107 b->ref = b->alt = b->flt = b->info = b->fmt = 0;
108 for (p = b->str, n = 0; p < b->str + b->l_str; ++p)
109 if (*p == 0 && p+1 != b->str + b->l_str) tmp[n++] = p + 1;
110 if (n != 5) return -1;
111 b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4];
113 for (p = b->alt, n = 1; *p; ++p)
116 c = toupper(*b->ref);
117 if (b->alt - b->ref == 2 && (c == 'A' || c == 'C' || c == 'G' || c == 'T'))
119 // set n_gi and gi[i].fmt
120 for (p = b->fmt, n = 1; *p; ++p)
126 b->gi = realloc(b->gi, b->m_gi * sizeof(bcf_ginfo_t));
127 memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t));
130 for (p = s = b->fmt, n = 0; *p; ++p) {
131 if (*p == ':' || *(p+1) == 0) {
132 char *q = *p == ':'? p : p + 1;
133 if ((q - s) != 2) return -2;
134 b->gi[n].fmt = char2int(s);
139 for (i = 0; i < b->n_gi; ++i) {
140 if (b->gi[i].fmt == char2int("PL")) {
141 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
142 } else if (b->gi[i].fmt == char2int("DP") || b->gi[i].fmt == char2int("HQ")) {
144 } else if (b->gi[i].fmt == char2int("GQ") || b->gi[i].fmt == char2int("GT")) {
146 } else if (b->gi[i].fmt == char2int("GL")) {
149 b->gi[i].data = malloc(n_smpl * b->gi[i].len);
154 int bcf_write(bcf_t *bp, const bcf1_t *b)
158 if (b == 0) return -1;
159 bgzf_write(bp->fp, &b->tid, 4);
160 bgzf_write(bp->fp, &b->pos, 4);
161 x = b->qual<<24 | b->l_str;
162 bgzf_write(bp->fp, &x, 4);
163 bgzf_write(bp->fp, b->str, b->l_str);
165 for (i = 0; i < b->n_gi; ++i) {
166 bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * bp->h.n_smpl);
167 l += b->gi[i].len * bp->h.n_smpl;
172 int bcf_read(bcf_t *bp, bcf1_t *b)
176 if (b == 0) return -1;
177 if (bgzf_read(bp->fp, &b->tid, 4) == 0) return 0;
178 bgzf_read(bp->fp, &b->pos, 4);
179 bgzf_read(bp->fp, &x, 4);
180 b->qual = x >> 24; b->l_str = x << 8 >> 8;
181 if (b->l_str > b->m_str) {
183 kroundup32(b->m_str);
184 b->str = realloc(b->str, b->m_str);
186 bgzf_read(bp->fp, b->str, b->l_str);
188 bcf_sync(bp->h.n_smpl, b);
189 for (i = 0; i < b->n_gi; ++i) {
190 bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * bp->h.n_smpl);
191 l += b->gi[i].len * bp->h.n_smpl;
196 int bcf_destroy(bcf1_t *b)
199 if (b == 0) return -1;
201 for (i = 0; i < b->n_gi; ++i)
208 static inline void fmt_str(const char *p, kstring_t *s)
210 if (*p == 0) kputc('.', s);
214 char *bcf_fmt(bcf_t *bp, bcf1_t *b)
218 memset(&s, 0, sizeof(kstring_t));
219 kputs(bp->h.ns[b->tid], &s); kputc('\t', &s);
220 kputw(b->pos + 1, &s); kputc('\t', &s);
221 fmt_str(b->str, &s); kputc('\t', &s);
222 fmt_str(b->ref, &s); kputc('\t', &s);
223 fmt_str(b->alt, &s); kputc('\t', &s);
224 kputw(b->qual, &s); kputc('\t', &s);
225 fmt_str(b->info, &s); kputc('\t', &s);
227 x = b->n_alleles * (b->n_alleles + 1) / 2;
228 for (j = 0; j < bp->h.n_smpl; ++j) {
230 for (i = 0; i < b->n_gi; ++i) {
231 if (i) kputc(':', &s);
232 if (b->gi[i].fmt == char2int("PL")) {
233 uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
235 for (k = 0; k < x; ++k) {
236 if (k > 0) kputc(',', &s);
239 } else if (b->gi[i].fmt == char2int("DP")) {
240 kputw(((uint16_t*)b->gi[i].data)[j], &s);
241 } else if (b->gi[i].fmt == char2int("GQ")) {
242 kputw(((uint8_t*)b->gi[i].data)[j], &s);