11 KHASH_SET_INIT_INT64(set64)
14 KSTREAM_INIT(gzFile, gzread, 16384)
24 #define VC_KEEPALT 256
25 #define VC_ACGT_ONLY 512
27 #define VC_CALL_GT 2048
30 int flag, prior_type, n1;
31 char *fn_list, *prior_file;
35 khash_t(set64) *bcf_load_pos(const char *fn, bcf_hdr_t *_h)
40 int ret, dret, lineno = 1;
42 khash_t(set64) *hash = 0;
44 hash = kh_init(set64);
45 str2id = bcf_build_refhash(_h);
46 str = calloc(1, sizeof(kstring_t));
47 fp = strcmp(fn, "-")? gzopen(fn, "r") : gzdopen(fileno(stdin), "r");
49 while (ks_getuntil(ks, 0, str, &dret) >= 0) {
50 int tid = bcf_str2id(str2id, str->s);
51 if (tid >= 0 && dret != '\n') {
52 if (ks_getuntil(ks, 0, str, &dret) >= 0) {
53 uint64_t x = (uint64_t)tid<<32 | (atoi(str->s) - 1);
54 kh_put(set64, hash, x, &ret);
56 } else fprintf(stderr, "[%s] %s is not a reference name (line %d).\n", __func__, str->s, lineno);
57 if (dret != '\n') while ((dret = ks_getc(ks)) > 0 && dret != '\n');
61 bcf_str2id_destroy(str2id);
64 free(str->s); free(str);
68 static double test_hwe(const double g[3])
70 extern double kf_gammaq(double p, double x);
71 double fexp, chi2, f[3], n;
73 n = g[0] + g[1] + g[2];
74 fexp = (2. * g[2] + g[1]) / (2. * n);
75 if (fexp > 1. - 1e-10) fexp = 1. - 1e-10;
76 if (fexp < 1e-10) fexp = 1e-10;
77 f[0] = n * (1. - fexp) * (1. - fexp);
78 f[1] = n * 2. * fexp * (1. - fexp);
79 f[2] = n * fexp * fexp;
80 for (i = 0, chi2 = 0.; i < 3; ++i)
81 chi2 += (g[i] - f[i]) * (g[i] - f[i]) / f[i];
82 return kf_gammaq(.5, chi2 / 2.);
87 int mq, depth, is_tested, d[4];
90 static double ttest(int n1, int n2, int a[4])
92 extern double kf_betai(double a, double b, double x);
94 if (n1 == 0 || n2 == 0 || n1 + n2 < 3) return 1.0;
95 u1 = (double)a[0] / n1; u2 = (double)a[2] / n2;
96 if (u1 <= u2) return 1.;
97 t = (u1 - u2) / sqrt(((a[1] - n1 * u1 * u1) + (a[3] - n2 * u2 * u2)) / (n1 + n2 - 2) * (1./n1 + 1./n2));
99 // printf("%d,%d,%d,%d,%lf,%lf,%lf\n", a[0], a[1], a[2], a[3], t, u1, u2);
100 return t < 0.? 1. : .5 * kf_betai(.5*v, .5, v/(v+t*t));
103 static int test16_core(int anno[16], anno16_t *a)
105 extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
108 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
109 memcpy(a->d, anno, 4 * sizeof(int));
110 a->depth = anno[0] + anno[1] + anno[2] + anno[3];
111 a->is_tested = (anno[0] + anno[1] > 0 && anno[2] + anno[3] > 0);
112 if (a->depth == 0) return -1;
113 a->mq = (int)(sqrt((anno[9] + anno[11]) / a->depth) + .499);
114 kt_fisher_exact(anno[0], anno[1], anno[2], anno[3], &left, &right, &a->p[0]);
115 for (i = 1; i < 4; ++i)
116 a->p[i] = ttest(anno[0] + anno[1], anno[2] + anno[3], anno+4*i);
120 static int test16(bcf1_t *b, anno16_t *a)
124 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
125 a->d[0] = a->d[1] = a->d[2] = a->d[3] = 0.;
126 a->mq = a->depth = a->is_tested = 0;
127 if ((p = strstr(b->info, "I16=")) == 0) return -1;
129 for (i = 0; i < 16; ++i) {
130 anno[i] = strtol(p, &p, 10);
131 if (anno[i] == 0 && (errno == EINVAL || errno == ERANGE)) return -2;
134 return test16_core(anno, a);
137 static void rm_info(bcf1_t *b, const char *key)
140 if ((p = strstr(b->info, key)) == 0) return;
141 for (q = p; *q && *q != ';'; ++q);
142 if (p > b->info && *(p-1) == ';') --p;
143 memmove(p, q, b->l_str - (q - b->str));
148 static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag)
151 int is_var = (pr->p_ref < pref);
152 double p_hwe, r = is_var? pr->p_ref : 1. - pr->p_ref;
155 p_hwe = pr->g[0] >= 0.? test_hwe(pr->g) : 1.0; // only do HWE g[] is calculated
159 memset(&s, 0, sizeof(kstring_t));
160 kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
161 kputs(b->alt, &s); kputc('\0', &s); kputc('\0', &s);
163 if (b->info[0]) kputc(';', &s);
164 ksprintf(&s, "AF1=%.3lf;AFE=%.3lf", 1.-pr->f_em, 1.-pr->f_exp);
165 ksprintf(&s, ";DP4=%d,%d,%d,%d;MQ=%d", a.d[0], a.d[1], a.d[2], a.d[3], a.mq);
167 if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%lg,%lg,%lg,%lg", pr->pc[0], pr->pc[1], pr->pc[2], pr->pc[3]);
168 ksprintf(&s, ";PV4=%.2lg,%.2lg,%.2lg,%.2lg", a.p[0], a.p[1], a.p[2], a.p[3]);
170 if (pr->g[0] >= 0. && p_hwe <= .2)
171 ksprintf(&s, ";GC=%.2lf,%.2lf,%.2lf;HWE=%.3lf", pr->g[2], pr->g[1], pr->g[0], p_hwe);
173 kputs(b->fmt, &s); kputc('\0', &s);
175 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
176 b->qual = r < 1e-100? 99 : -4.343 * log(r);
177 if (b->qual > 99) b->qual = 99;
179 if (!is_var) bcf_shrink_alt(b, 1);
180 else if (!(flag&VC_KEEPALT))
181 bcf_shrink_alt(b, pr->rank0 < 2? 2 : pr->rank0+1);
182 if (is_var && (flag&VC_CALL_GT)) { // call individual genotype
183 int i, x, old_n_gi = b->n_gi;
184 s.m = b->m_str; s.l = b->l_str - 1; s.s = b->str;
185 kputs(":GT:GQ", &s); kputc('\0', &s);
186 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
188 for (i = 0; i < b->n_smpl; ++i) {
189 x = bcf_p1_call_gt(pa, pr->f_em, i);
190 ((uint8_t*)b->gi[old_n_gi].data)[i] = (x&3) == 0? 1<<3|1 : (x&3) == 1? 1 : 0;
191 ((uint8_t*)b->gi[old_n_gi+1].data)[i] = x>>2;
197 int bcfview(int argc, char *argv[])
199 extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b);
200 bcf_t *bp, *bout = 0;
203 uint64_t n_processed = 0;
208 char moder[4], modew[4];
209 khash_t(set64) *hash = 0;
211 tid = begin = end = -1;
212 memset(&vc, 0, sizeof(viewconf_t));
213 vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5;
214 while ((c = getopt(argc, argv, "N1:l:cHAGvLbSuP:t:p:Qg")) >= 0) {
216 case '1': vc.n1 = atoi(optarg); break;
217 case 'l': vc.fn_list = strdup(optarg); break;
218 case 'N': vc.flag |= VC_ACGT_ONLY; break;
219 case 'G': vc.flag |= VC_NO_GENO; break;
220 case 'L': vc.flag |= VC_NO_PL; break;
221 case 'A': vc.flag |= VC_KEEPALT; break;
222 case 'b': vc.flag |= VC_BCFOUT; break;
223 case 'S': vc.flag |= VC_VCFIN; break;
224 case 'c': vc.flag |= VC_CALL; break;
225 case 'v': vc.flag |= VC_VARONLY; break;
226 case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
227 case 'H': vc.flag |= VC_HWE; break;
228 case 'g': vc.flag |= VC_CALL_GT; break;
229 case 't': vc.theta = atof(optarg); break;
230 case 'p': vc.pref = atof(optarg); break;
231 case 'Q': vc.flag |= VC_QCALL; break;
233 if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
234 else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
235 else if (strcmp(optarg, "flat") == 0) vc.prior_type = MC_PTYPE_FLAT;
236 else vc.prior_file = strdup(optarg);
240 if (argc == optind) {
241 fprintf(stderr, "\n");
242 fprintf(stderr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
243 fprintf(stderr, "Options: -c SNP calling\n");
244 fprintf(stderr, " -b output BCF instead of VCF\n");
245 fprintf(stderr, " -u uncompressed BCF output\n");
246 fprintf(stderr, " -S input is VCF\n");
247 fprintf(stderr, " -A keep all possible alternate alleles at variant sites\n");
248 fprintf(stderr, " -G suppress all individual genotype information\n");
249 fprintf(stderr, " -g call genotypes for variant sites\n");
250 fprintf(stderr, " -L discard the PL genotype field\n");
251 fprintf(stderr, " -H perform Hardy-Weinberg test (slower)\n");
252 fprintf(stderr, " -v output potential variant sites only\n");
253 fprintf(stderr, " -N skip sites where REF is not A/C/G/T\n");
254 fprintf(stderr, " -Q output the QCALL likelihood format\n");
255 fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
256 fprintf(stderr, " -l FILE list of sites to output [all sites]\n");
257 fprintf(stderr, " -t FLOAT scaled mutation rate [%.4lg]\n", vc.theta);
258 fprintf(stderr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3lg]\n", vc.pref);
259 fprintf(stderr, " -P STR type of prior: full, cond2, flat [full]\n");
260 fprintf(stderr, "\n");
264 b = calloc(1, sizeof(bcf1_t));
266 if (!(vc.flag & VC_VCFIN)) strcat(moder, "b");
268 if (vc.flag & VC_BCFOUT) strcat(modew, "b");
269 if (vc.flag & VC_UNCOMP) strcat(modew, "u");
270 bp = vcf_open(argv[optind], moder);
271 h = vcf_hdr_read(bp);
272 bout = vcf_open("-", modew);
273 if (!(vc.flag & VC_QCALL)) vcf_hdr_write(bout, h);
274 if (vc.flag & VC_CALL) {
275 p1 = bcf_p1_init(h->n_smpl);
277 if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
278 fprintf(stderr, "[%s] fail to read the prior AFS.\n", __func__);
281 } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
283 bcf_p1_set_n1(p1, vc.n1);
284 bcf_p1_init_subprior(p1, vc.prior_type, vc.theta);
287 if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h);
288 if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
289 void *str2id = bcf_build_refhash(h);
290 if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
292 idx = bcf_idx_load(argv[optind]);
295 off = bcf_idx_query(idx, tid, begin);
297 fprintf(stderr, "[%s] no records in the query region.\n", __func__);
298 return 1; // FIXME: a lot of memory leaks...
300 bgzf_seek(bp->fp, off, SEEK_SET);
301 bcf_idx_destroy(idx);
305 while (vcf_read(bp, h, b) > 0) {
306 if (vc.flag & VC_ACGT_ONLY) {
308 if (b->ref[0] == 0 || b->ref[1] != 0) continue;
309 x = toupper(b->ref[0]);
310 if (x != 'A' && x != 'C' && x != 'G' && x != 'T') continue;
313 uint64_t x = (uint64_t)b->tid<<32 | b->pos;
314 khint_t k = kh_get(set64, hash, x);
315 if (kh_size(hash) == 0) break;
316 if (k == kh_end(hash)) continue;
317 kh_del(set64, hash, k);
320 int l = strlen(b->ref);
321 l = b->pos + (l > 0? l : 1);
322 if (b->tid != tid || b->pos >= end) break;
323 if (!(l > begin && end > b->pos)) continue;
326 if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
330 if (vc.flag & VC_CALL) { // call variants
333 bcf_p1_cal(b, p1, &pr); // pr.g[3] is not calculated here
334 if (vc.flag&VC_HWE) bcf_p1_cal_g3(p1, pr.g);
335 if (n_processed % 100000 == 0) {
336 fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
339 if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
340 update_bcf1(h->n_smpl, b, p1, &pr, vc.pref, vc.flag);
342 if (vc.flag & VC_NO_GENO) { // do not output GENO fields
346 vcf_write(bout, h, b);
348 if (vc.prior_file) free(vc.prior_file);
349 if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
352 vcf_close(bp); vcf_close(bout);
353 if (hash) kh_destroy(set64, hash);
354 if (vc.fn_list) free(vc.fn_list);
355 if (p1) bcf_p1_destroy(p1);