Mercurial > repos > siyuan > prada
comparison pyPRADA_1.2/tools/samtools-0.1.16/bcftools/bcf.c @ 0:acc2ca1a3ba4
Uploaded
author | siyuan |
---|---|
date | Thu, 20 Feb 2014 00:44:58 -0500 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:acc2ca1a3ba4 |
---|---|
1 #include <string.h> | |
2 #include <ctype.h> | |
3 #include <stdio.h> | |
4 #include "kstring.h" | |
5 #include "bcf.h" | |
6 | |
7 bcf_t *bcf_open(const char *fn, const char *mode) | |
8 { | |
9 bcf_t *b; | |
10 b = calloc(1, sizeof(bcf_t)); | |
11 if (strchr(mode, 'w')) { | |
12 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode); | |
13 } else { | |
14 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode); | |
15 } | |
16 #ifndef BCF_LITE | |
17 b->fp->owned_file = 1; | |
18 #endif | |
19 return b; | |
20 } | |
21 | |
22 int bcf_close(bcf_t *b) | |
23 { | |
24 int ret; | |
25 if (b == 0) return 0; | |
26 ret = bgzf_close(b->fp); | |
27 free(b); | |
28 return ret; | |
29 } | |
30 | |
31 int bcf_hdr_write(bcf_t *b, const bcf_hdr_t *h) | |
32 { | |
33 if (b == 0 || h == 0) return -1; | |
34 bgzf_write(b->fp, "BCF\4", 4); | |
35 bgzf_write(b->fp, &h->l_nm, 4); | |
36 bgzf_write(b->fp, h->name, h->l_nm); | |
37 bgzf_write(b->fp, &h->l_smpl, 4); | |
38 bgzf_write(b->fp, h->sname, h->l_smpl); | |
39 bgzf_write(b->fp, &h->l_txt, 4); | |
40 bgzf_write(b->fp, h->txt, h->l_txt); | |
41 bgzf_flush(b->fp); | |
42 return 16 + h->l_nm + h->l_smpl + h->l_txt; | |
43 } | |
44 | |
45 bcf_hdr_t *bcf_hdr_read(bcf_t *b) | |
46 { | |
47 uint8_t magic[4]; | |
48 bcf_hdr_t *h; | |
49 if (b == 0) return 0; | |
50 h = calloc(1, sizeof(bcf_hdr_t)); | |
51 bgzf_read(b->fp, magic, 4); | |
52 bgzf_read(b->fp, &h->l_nm, 4); | |
53 h->name = malloc(h->l_nm); | |
54 bgzf_read(b->fp, h->name, h->l_nm); | |
55 bgzf_read(b->fp, &h->l_smpl, 4); | |
56 h->sname = malloc(h->l_smpl); | |
57 bgzf_read(b->fp, h->sname, h->l_smpl); | |
58 bgzf_read(b->fp, &h->l_txt, 4); | |
59 h->txt = malloc(h->l_txt); | |
60 bgzf_read(b->fp, h->txt, h->l_txt); | |
61 bcf_hdr_sync(h); | |
62 return h; | |
63 } | |
64 | |
65 void bcf_hdr_destroy(bcf_hdr_t *h) | |
66 { | |
67 if (h == 0) return; | |
68 free(h->name); free(h->sname); free(h->txt); free(h->ns); free(h->sns); | |
69 free(h); | |
70 } | |
71 | |
72 static inline char **cnt_null(int l, char *str, int *_n) | |
73 { | |
74 int n = 0; | |
75 char *p, **list; | |
76 *_n = 0; | |
77 if (l == 0 || str == 0) return 0; | |
78 for (p = str; p != str + l; ++p) | |
79 if (*p == 0) ++n; | |
80 *_n = n; | |
81 list = calloc(n, sizeof(void*)); | |
82 list[0] = str; | |
83 for (p = str, n = 1; p < str + l - 1; ++p) | |
84 if (*p == 0) list[n++] = p + 1; | |
85 return list; | |
86 } | |
87 | |
88 int bcf_hdr_sync(bcf_hdr_t *b) | |
89 { | |
90 if (b == 0) return -1; | |
91 if (b->ns) free(b->ns); | |
92 if (b->sns) free(b->sns); | |
93 if (b->l_nm) b->ns = cnt_null(b->l_nm, b->name, &b->n_ref); | |
94 else b->ns = 0, b->n_ref = 0; | |
95 b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl); | |
96 return 0; | |
97 } | |
98 | |
99 int bcf_sync(bcf1_t *b) | |
100 { | |
101 char *p, *tmp[5]; | |
102 int i, n, n_smpl = b->n_smpl; | |
103 ks_tokaux_t aux; | |
104 // set ref, alt, flt, info, fmt | |
105 b->ref = b->alt = b->flt = b->info = b->fmt = 0; | |
106 for (p = b->str, n = 0; p < b->str + b->l_str; ++p) { | |
107 if (*p == 0 && p+1 != b->str + b->l_str) { | |
108 if (n == 5) { | |
109 ++n; | |
110 break; | |
111 } else tmp[n++] = p + 1; | |
112 } | |
113 } | |
114 if (n != 5) { | |
115 fprintf(stderr, "[%s] incorrect number of fields (%d != 5) at %d:%d\n", __func__, n, b->tid, b->pos); | |
116 return -1; | |
117 } | |
118 b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4]; | |
119 // set n_alleles | |
120 if (*b->alt == 0) b->n_alleles = 1; | |
121 else { | |
122 for (p = b->alt, n = 1; *p; ++p) | |
123 if (*p == ',') ++n; | |
124 b->n_alleles = n + 1; | |
125 } | |
126 // set n_gi and gi[i].fmt | |
127 for (p = b->fmt, n = 1; *p; ++p) | |
128 if (*p == ':') ++n; | |
129 if (n > b->m_gi) { | |
130 int old_m = b->m_gi; | |
131 b->m_gi = n; | |
132 kroundup32(b->m_gi); | |
133 b->gi = realloc(b->gi, b->m_gi * sizeof(bcf_ginfo_t)); | |
134 memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t)); | |
135 } | |
136 b->n_gi = n; | |
137 for (p = kstrtok(b->fmt, ":", &aux), n = 0; p; p = kstrtok(0, 0, &aux)) | |
138 b->gi[n++].fmt = bcf_str2int(p, aux.p - p); | |
139 // set gi[i].len | |
140 for (i = 0; i < b->n_gi; ++i) { | |
141 if (b->gi[i].fmt == bcf_str2int("PL", 2)) { | |
142 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2; | |
143 } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2)) { | |
144 b->gi[i].len = 2; | |
145 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) { | |
146 b->gi[i].len = 1; | |
147 } else if (b->gi[i].fmt == bcf_str2int("SP", 2)) { | |
148 b->gi[i].len = 4; | |
149 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) { | |
150 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2 * 4; | |
151 } | |
152 b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len); | |
153 } | |
154 return 0; | |
155 } | |
156 | |
157 int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b) | |
158 { | |
159 int i, l = 0; | |
160 if (b == 0) return -1; | |
161 bgzf_write(bp->fp, &b->tid, 4); | |
162 bgzf_write(bp->fp, &b->pos, 4); | |
163 bgzf_write(bp->fp, &b->qual, 4); | |
164 bgzf_write(bp->fp, &b->l_str, 4); | |
165 bgzf_write(bp->fp, b->str, b->l_str); | |
166 l = 12 + b->l_str; | |
167 for (i = 0; i < b->n_gi; ++i) { | |
168 bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl); | |
169 l += b->gi[i].len * h->n_smpl; | |
170 } | |
171 return l; | |
172 } | |
173 | |
174 int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b) | |
175 { | |
176 int i, l = 0; | |
177 if (b == 0) return -1; | |
178 if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1; | |
179 b->n_smpl = h->n_smpl; | |
180 bgzf_read(bp->fp, &b->pos, 4); | |
181 bgzf_read(bp->fp, &b->qual, 4); | |
182 bgzf_read(bp->fp, &b->l_str, 4); | |
183 if (b->l_str > b->m_str) { | |
184 b->m_str = b->l_str; | |
185 kroundup32(b->m_str); | |
186 b->str = realloc(b->str, b->m_str); | |
187 } | |
188 bgzf_read(bp->fp, b->str, b->l_str); | |
189 l = 12 + b->l_str; | |
190 if (bcf_sync(b) < 0) return -2; | |
191 for (i = 0; i < b->n_gi; ++i) { | |
192 bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl); | |
193 l += b->gi[i].len * h->n_smpl; | |
194 } | |
195 return l; | |
196 } | |
197 | |
198 int bcf_destroy(bcf1_t *b) | |
199 { | |
200 int i; | |
201 if (b == 0) return -1; | |
202 free(b->str); | |
203 for (i = 0; i < b->m_gi; ++i) | |
204 free(b->gi[i].data); | |
205 free(b->gi); | |
206 free(b); | |
207 return 0; | |
208 } | |
209 | |
210 static inline void fmt_str(const char *p, kstring_t *s) | |
211 { | |
212 if (*p == 0) kputc('.', s); | |
213 else kputs(p, s); | |
214 } | |
215 | |
216 void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s) | |
217 { | |
218 int i, j, x; | |
219 s->l = 0; | |
220 if (h->n_ref) kputs(h->ns[b->tid], s); | |
221 else kputw(b->tid, s); | |
222 kputc('\t', s); | |
223 kputw(b->pos + 1, s); kputc('\t', s); | |
224 fmt_str(b->str, s); kputc('\t', s); | |
225 fmt_str(b->ref, s); kputc('\t', s); | |
226 fmt_str(b->alt, s); kputc('\t', s); | |
227 ksprintf(s, "%.3g", b->qual); kputc('\t', s); | |
228 fmt_str(b->flt, s); kputc('\t', s); | |
229 fmt_str(b->info, s); | |
230 if (b->fmt[0]) { | |
231 kputc('\t', s); | |
232 fmt_str(b->fmt, s); | |
233 } | |
234 x = b->n_alleles * (b->n_alleles + 1) / 2; | |
235 if (b->n_gi == 0) return; | |
236 for (j = 0; j < h->n_smpl; ++j) { | |
237 kputc('\t', s); | |
238 for (i = 0; i < b->n_gi; ++i) { | |
239 if (i) kputc(':', s); | |
240 if (b->gi[i].fmt == bcf_str2int("PL", 2)) { | |
241 uint8_t *d = (uint8_t*)b->gi[i].data + j * x; | |
242 int k; | |
243 for (k = 0; k < x; ++k) { | |
244 if (k > 0) kputc(',', s); | |
245 kputw(d[k], s); | |
246 } | |
247 } else if (b->gi[i].fmt == bcf_str2int("DP", 2)) { | |
248 kputw(((uint16_t*)b->gi[i].data)[j], s); | |
249 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) { | |
250 kputw(((uint8_t*)b->gi[i].data)[j], s); | |
251 } else if (b->gi[i].fmt == bcf_str2int("SP", 2)) { | |
252 kputw(((int32_t*)b->gi[i].data)[j], s); | |
253 } else if (b->gi[i].fmt == bcf_str2int("GT", 2)) { | |
254 int y = ((uint8_t*)b->gi[i].data)[j]; | |
255 if (y>>7&1) { | |
256 kputsn("./.", 3, s); | |
257 } else { | |
258 kputc('0' + (y>>3&7), s); | |
259 kputc("/|"[y>>6&1], s); | |
260 kputc('0' + (y&7), s); | |
261 } | |
262 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) { | |
263 float *d = (float*)b->gi[i].data + j * x; | |
264 int k; | |
265 //printf("- %lx\n", d); | |
266 for (k = 0; k < x; ++k) { | |
267 if (k > 0) kputc(',', s); | |
268 ksprintf(s, "%.2f", d[k]); | |
269 } | |
270 } else kputc('.', s); // custom fields | |
271 } | |
272 } | |
273 } | |
274 | |
275 char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b) | |
276 { | |
277 kstring_t s; | |
278 s.l = s.m = 0; s.s = 0; | |
279 bcf_fmt_core(h, b, &s); | |
280 return s.s; | |
281 } | |
282 | |
283 int bcf_append_info(bcf1_t *b, const char *info, int l) | |
284 { | |
285 int shift = b->fmt - b->str; | |
286 int l_fmt = b->l_str - shift; | |
287 char *ori = b->str; | |
288 if (b->l_str + l > b->m_str) { // enlarge if necessary | |
289 b->m_str = b->l_str + l; | |
290 kroundup32(b->m_str); | |
291 b->str = realloc(b->str, b->m_str); | |
292 } | |
293 memmove(b->str + shift + l, b->str + shift, l_fmt); // move the FORMAT field | |
294 memcpy(b->str + shift - 1, info, l); // append to the INFO field | |
295 b->str[shift + l - 1] = '\0'; | |
296 b->fmt = b->str + shift + l; | |
297 b->l_str += l; | |
298 if (ori != b->str) bcf_sync(b); // synchronize when realloc changes the pointer | |
299 return 0; | |
300 } | |
301 | |
302 int bcf_cpy(bcf1_t *r, const bcf1_t *b) | |
303 { | |
304 char *t1 = r->str; | |
305 bcf_ginfo_t *t2 = r->gi; | |
306 int i, t3 = r->m_str, t4 = r->m_gi; | |
307 *r = *b; | |
308 r->str = t1; r->gi = t2; r->m_str = t3; r->m_gi = t4; | |
309 if (r->m_str < b->m_str) { | |
310 r->m_str = b->m_str; | |
311 r->str = realloc(r->str, r->m_str); | |
312 } | |
313 memcpy(r->str, b->str, r->m_str); | |
314 bcf_sync(r); // calling bcf_sync() is simple but inefficient | |
315 for (i = 0; i < r->n_gi; ++i) | |
316 memcpy(r->gi[i].data, b->gi[i].data, r->n_smpl * r->gi[i].len); | |
317 return 0; | |
318 } | |
319 | |
320 int bcf_is_indel(const bcf1_t *b) | |
321 { | |
322 char *p; | |
323 if (strlen(b->ref) > 1) return 1; | |
324 for (p = b->alt; *p; ++p) | |
325 if (*p != ',' && p[1] != ',' && p[1] != '\0') | |
326 return 1; | |
327 return 0; | |
328 } |