11 KHASH_SET_INIT_INT64(set64)
14 KSTREAM_INIT(gzFile, gzread, 16384)
26 char *fn_list, *prior_file;
30 khash_t(set64) *bcf_load_pos(const char *fn, bcf_hdr_t *_h)
35 int ret, dret, lineno = 1;
37 khash_t(set64) *hash = 0;
39 hash = kh_init(set64);
40 str2id = bcf_build_refhash(_h);
41 str = calloc(1, sizeof(kstring_t));
42 fp = strcmp(fn, "-")? gzopen(fn, "r") : gzdopen(fileno(stdin), "r");
44 while (ks_getuntil(ks, 0, str, &dret) >= 0) {
45 int tid = bcf_str2id(str2id, str->s);
46 if (tid >= 0 && dret != '\n') {
47 if (ks_getuntil(ks, 0, str, &dret) >= 0) {
48 uint64_t x = (uint64_t)tid<<32 | (atoi(str->s) - 1);
49 kh_put(set64, hash, x, &ret);
51 } else fprintf(stderr, "[%s] %s is not a reference name (line %d).\n", __func__, str->s, lineno);
52 if (dret != '\n') while ((dret = ks_getc(ks)) > 0 && dret != '\n');
56 bcf_str2id_destroy(str2id);
59 free(str->s); free(str);
63 static double test_hwe(const double g[3])
65 extern double kf_gammaq(double p, double x);
66 double fexp, chi2, f[3], n;
68 n = g[0] + g[1] + g[2];
69 fexp = (2. * g[2] + g[1]) / (2. * n);
70 if (fexp > 1. - 1e-10) fexp = 1. - 1e-10;
71 if (fexp < 1e-10) fexp = 1e-10;
72 f[0] = n * (1. - fexp) * (1. - fexp);
73 f[1] = n * 2. * fexp * (1. - fexp);
74 f[2] = n * fexp * fexp;
75 for (i = 0, chi2 = 0.; i < 3; ++i)
76 chi2 += (g[i] - f[i]) * (g[i] - f[i]) / f[i];
77 return kf_gammaq(.5, chi2 / 2.);
82 int mq, depth, is_tested, d[4];
85 static double ttest(int n1, int n2, int a[4])
87 extern double kf_betai(double a, double b, double x);
89 if (n1 == 0 || n2 == 0 || n1 + n2 < 3) return 1.0;
90 t = (a[0] / n1 - a[2] / n2) / sqrt((a[1] + a[3]) / (n1 + n2 - 2) * (1./n1 + 1./n2));
92 // printf("%d,%d,%d,%d,%lf\n", a[0], a[1], a[2], a[3], t);
93 return t < 0.? 1. : .5 * kf_betai(.5*v, .5, v/(v+t*t));
96 static int test16_core(int anno[16], anno16_t *a)
98 extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
101 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
102 memcpy(a->d, anno, 4 * sizeof(int));
103 a->depth = anno[0] + anno[1] + anno[2] + anno[3];
104 a->is_tested = (anno[0] + anno[1] > 0 && anno[2] + anno[3] > 0);
105 if (a->depth == 0) return -1;
106 a->mq = (int)(sqrt((anno[9] + anno[11]) / a->depth) + .499);
107 kt_fisher_exact(anno[0], anno[1], anno[2], anno[3], &left, &right, &a->p[0]);
108 for (i = 1; i < 4; ++i)
109 a->p[i] = ttest(anno[0] + anno[1], anno[2] + anno[3], anno+4*i);
113 static int test16(bcf1_t *b, anno16_t *a)
117 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
118 a->d[0] = a->d[1] = a->d[2] = a->d[3] = 0.;
119 a->mq = a->depth = a->is_tested = 0;
120 if ((p = strstr(b->info, "I16=")) == 0) return -1;
122 for (i = 0; i < 16; ++i) {
123 anno[i] = strtol(p, &p, 10);
124 if (anno[i] == 0 && (errno == EINVAL || errno == ERANGE)) return -2;
127 return test16_core(anno, a);
130 static void rm_info(int n_smpl, bcf1_t *b, const char *key)
133 if ((p = strstr(b->info, key)) == 0) return;
134 for (q = p; *q && *q != ';'; ++q);
135 if (p > b->info && *(p-1) == ';') --p;
136 memmove(p, q, b->l_str - (q - b->str));
141 static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag)
144 int is_var = (pr->p_ref < pref);
145 double p_hwe, r = is_var? pr->p_ref : 1. - pr->p_ref;
148 p_hwe = test_hwe(pr->g);
150 rm_info(n_smpl, b, "I16=");
152 memset(&s, 0, sizeof(kstring_t));
153 kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
157 kputc('\0', &s); kputc('\0', &s);
159 if (b->info[0]) kputc(';', &s);
160 ksprintf(&s, "AF1=%.3lf;AFE=%.3lf", 1.-pr->f_em, 1.-pr->f_exp);
161 ksprintf(&s, ";DP4=%d,%d,%d,%d;MQ=%d", a.d[0], a.d[1], a.d[2], a.d[3], a.mq);
162 if (a.is_tested) ksprintf(&s, ";PV4=%.2lg,%.2lg,%.2lg,%.2lg", a.p[0], a.p[1], a.p[2], a.p[3]);
163 if (p_hwe <= .2) ksprintf(&s, ";GC=%.2lf,%.2lf,%.2lf;HWE=%.3lf", pr->g[2], pr->g[1], pr->g[0], p_hwe);
165 kputs(b->fmt, &s); kputc('\0', &s);
167 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
168 b->qual = r < 1e-100? 99 : -3.434 * log(r);
169 if (b->qual > 99) b->qual = 99;
174 int bcfview(int argc, char *argv[])
176 bcf_t *bp, *bout = 0;
179 uint64_t n_processed = 0;
184 char moder[4], modew[4];
185 khash_t(set64) *hash = 0;
187 tid = begin = end = -1;
188 memset(&vc, 0, sizeof(viewconf_t));
189 vc.prior_type = -1; vc.theta = 1e-3; vc.pref = 0.9;
190 while ((c = getopt(argc, argv, "l:cGvLbSuP:t:p:")) >= 0) {
192 case 'l': vc.fn_list = strdup(optarg); break;
193 case 'G': vc.flag |= VC_NO_GENO; break;
194 case 'L': vc.flag |= VC_NO_PL; break;
195 case 'b': vc.flag |= VC_BCFOUT; break;
196 case 'S': vc.flag |= VC_VCFIN; break;
197 case 'c': vc.flag |= VC_CALL; break;
198 case 'v': vc.flag |= VC_VARONLY; break;
199 case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
200 case 't': vc.theta = atof(optarg); break;
201 case 'p': vc.pref = atof(optarg); break;
203 if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
204 else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
205 else if (strcmp(optarg, "flat") == 0) vc.prior_type = MC_PTYPE_FLAT;
206 else vc.prior_file = strdup(optarg);
210 if (argc == optind) {
211 fprintf(stderr, "\n");
212 fprintf(stderr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
213 fprintf(stderr, "Options: -c SNP calling\n");
214 fprintf(stderr, " -b output BCF instead of VCF\n");
215 fprintf(stderr, " -u uncompressed BCF output\n");
216 fprintf(stderr, " -S input is VCF\n");
217 fprintf(stderr, " -G suppress all individual genotype information\n");
218 fprintf(stderr, " -L discard the PL genotype field\n");
219 fprintf(stderr, " -v output potential variant sites only\n");
220 fprintf(stderr, " -l FILE list of sites to output [all sites]\n");
221 fprintf(stderr, " -t FLOAT scaled mutation rate [%.4lg]\n", vc.theta);
222 fprintf(stderr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3lg]\n", vc.pref);
223 fprintf(stderr, " -P STR type of prior: full, cond2, flat [full]\n");
224 fprintf(stderr, "\n");
228 b = calloc(1, sizeof(bcf1_t));
230 if (!(vc.flag & VC_VCFIN)) strcat(moder, "b");
232 if (vc.flag & VC_BCFOUT) strcat(modew, "b");
233 if (vc.flag & VC_UNCOMP) strcat(modew, "u");
234 bp = vcf_open(argv[optind], moder);
235 h = vcf_hdr_read(bp);
236 bout = vcf_open("-", modew);
237 vcf_hdr_write(bout, h);
238 if (vc.flag & VC_CALL) {
239 p1 = bcf_p1_init(h->n_smpl);
241 if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
242 fprintf(stderr, "[%s] fail to read the prior AFS.\n", __func__);
245 } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
247 if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h);
248 if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
249 void *str2id = bcf_build_refhash(h);
250 if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
252 idx = bcf_idx_load(argv[optind]);
255 off = bcf_idx_query(idx, tid, begin, end);
256 bgzf_seek(bp->fp, off, SEEK_SET);
257 bcf_idx_destroy(idx);
261 while (vcf_read(bp, h, b) > 0) {
263 uint64_t x = (uint64_t)b->tid<<32 | b->pos;
264 khint_t k = kh_get(set64, hash, x);
265 if (k == kh_end(hash)) continue;
268 int l = strlen(b->ref);
269 l = b->pos + (l > 0? l : 1);
270 if (b->tid != tid || b->pos >= end) break;
271 if (!(l > begin && end > b->pos)) continue;
273 if (vc.flag & VC_CALL) {
275 bcf_p1_cal(b, p1, &pr);
276 if ((n_processed + 1) % 50000 == 0) bcf_p1_dump_afs(p1);
277 if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
278 update_bcf1(h->n_smpl, b, p1, &pr, vc.pref, vc.flag);
280 if (vc.flag & VC_NO_GENO) {
284 vcf_write(bout, h, b);
287 if (vc.prior_file) free(vc.prior_file);
288 if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
291 vcf_close(bp); vcf_close(bout);
292 if (hash) kh_destroy(set64, hash);
293 if (vc.fn_list) free(vc.fn_list);
294 if (p1) bcf_p1_destroy(p1);