7 void bcf_hdr_clear(bcf_hdr_t *b)
9 free(b->name); free(b->sname); free(b->txt); free(b->ns); free(b->sns);
10 memset(b, 0, sizeof(bcf_hdr_t));
13 bcf_t *bcf_open(const char *fn, const char *mode)
16 b = calloc(1, sizeof(bcf_t));
17 if (strchr(mode, 'w')) {
18 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode);
20 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
22 b->fp->owned_file = 1;
26 int bcf_close(bcf_t *b)
30 ret = bgzf_close(b->fp);
35 int bcf_hdr_write(bcf_t *b, const bcf_hdr_t *h)
37 if (b == 0 || h == 0) return -1;
38 bgzf_write(b->fp, "BCF\4", 4);
39 bgzf_write(b->fp, &h->l_nm, 4);
40 bgzf_write(b->fp, h->name, h->l_nm);
41 bgzf_write(b->fp, &h->l_smpl, 4);
42 bgzf_write(b->fp, h->sname, h->l_smpl);
43 bgzf_write(b->fp, &h->l_txt, 4);
44 bgzf_write(b->fp, h->txt, h->l_txt);
46 return 16 + h->l_nm + h->l_smpl + h->l_txt;
49 bcf_hdr_t *bcf_hdr_read(bcf_t *b)
54 h = calloc(1, sizeof(bcf_hdr_t));
55 bgzf_read(b->fp, magic, 4);
56 bgzf_read(b->fp, &h->l_nm, 4);
57 h->name = malloc(h->l_nm);
58 bgzf_read(b->fp, h->name, h->l_nm);
59 bgzf_read(b->fp, &h->l_smpl, 4);
60 h->sname = malloc(h->l_smpl);
61 bgzf_read(b->fp, h->sname, h->l_smpl);
62 bgzf_read(b->fp, &h->l_txt, 4);
63 h->txt = malloc(h->l_txt);
64 bgzf_read(b->fp, h->txt, h->l_txt);
69 void bcf_hdr_destroy(bcf_hdr_t *h)
72 free(h->name); free(h->sname); free(h->txt); free(h->ns); free(h->sns);
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 if (b->ns) free(b->ns);
96 if (b->sns) free(b->sns);
97 if (b->l_nm) b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
98 else b->ns = 0, b->n_ref = 0;
99 b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl);
103 int bcf_sync(int n_smpl, bcf1_t *b)
108 // set ref, alt, flt, info, fmt
109 b->ref = b->alt = b->flt = b->info = b->fmt = 0;
110 for (p = b->str, n = 0; p < b->str + b->l_str; ++p)
111 if (*p == 0 && p+1 != b->str + b->l_str) tmp[n++] = p + 1;
113 fprintf(stderr, "[bcf_sync] incorrect number of fields (%d != 5)\n", n);
116 b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4];
118 if (*b->alt == 0) b->n_alleles = 1;
120 for (p = b->alt, n = 1; *p; ++p)
122 b->n_alleles = n + 1;
124 // set n_gi and gi[i].fmt
125 for (p = b->fmt, n = 1; *p; ++p)
131 b->gi = realloc(b->gi, b->m_gi * sizeof(bcf_ginfo_t));
132 memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t));
135 for (p = kstrtok(b->fmt, ":", &aux), n = 0; p; p = kstrtok(0, 0, &aux))
136 b->gi[n++].fmt = bcf_str2int(p, aux.p - p);
138 for (i = 0; i < b->n_gi; ++i) {
139 if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
140 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
141 } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2)) {
143 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) {
145 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
146 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2 * 4;
148 b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len);
153 int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b)
156 if (b == 0) return -1;
157 bgzf_write(bp->fp, &b->tid, 4);
158 bgzf_write(bp->fp, &b->pos, 4);
159 bgzf_write(bp->fp, &b->qual, 4);
160 bgzf_write(bp->fp, &b->l_str, 4);
161 bgzf_write(bp->fp, b->str, b->l_str);
163 for (i = 0; i < b->n_gi; ++i) {
164 bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
165 l += b->gi[i].len * h->n_smpl;
170 int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b)
173 if (b == 0) return -1;
174 if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1;
175 bgzf_read(bp->fp, &b->pos, 4);
176 bgzf_read(bp->fp, &b->qual, 4);
177 bgzf_read(bp->fp, &b->l_str, 4);
178 if (b->l_str > b->m_str) {
180 kroundup32(b->m_str);
181 b->str = realloc(b->str, b->m_str);
183 bgzf_read(bp->fp, b->str, b->l_str);
185 bcf_sync(h->n_smpl, b);
186 for (i = 0; i < b->n_gi; ++i) {
187 bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
188 l += b->gi[i].len * h->n_smpl;
193 int bcf_destroy(bcf1_t *b)
196 if (b == 0) return -1;
198 for (i = 0; i < b->n_gi; ++i)
205 static inline void fmt_str(const char *p, kstring_t *s)
207 if (*p == 0) kputc('.', s);
211 void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
215 if (h->n_ref) kputs(h->ns[b->tid], s);
216 else kputw(b->tid, s);
218 kputw(b->pos + 1, s); kputc('\t', s);
219 fmt_str(b->str, s); kputc('\t', s);
220 fmt_str(b->ref, s); kputc('\t', s);
221 fmt_str(b->alt, s); kputc('\t', s);
222 ksprintf(s, "%.3g", b->qual); /*kputw(b->qual, s);*/ kputc('\t', s);
223 fmt_str(b->flt, s); kputc('\t', s);
229 x = b->n_alleles * (b->n_alleles + 1) / 2;
230 if (b->n_gi == 0) return;
231 for (j = 0; j < h->n_smpl; ++j) {
233 for (i = 0; i < b->n_gi; ++i) {
234 if (i) kputc(':', s);
235 if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
236 uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
238 for (k = 0; k < x; ++k) {
239 if (k > 0) kputc(',', s);
242 } else if (b->gi[i].fmt == bcf_str2int("DP", 2)) {
243 kputw(((uint16_t*)b->gi[i].data)[j], s);
244 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
245 kputw(((uint8_t*)b->gi[i].data)[j], s);
246 } else if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
247 int y = ((uint8_t*)b->gi[i].data)[j];
251 kputc('0' + (y>>3&7), s);
252 kputc("/|"[y>>6&1], s);
253 kputc('0' + (y&7), s);
255 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
256 float *d = (float*)b->gi[i].data + j * x;
258 //printf("- %lx\n", d);
259 for (k = 0; k < x; ++k) {
260 if (k > 0) kputc(',', s);
261 ksprintf(s, "%.2f", d[k]);
268 char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b)
271 s.l = s.m = 0; s.s = 0;
272 bcf_fmt_core(h, b, &s);