12 KSTREAM_INIT(gzFile, gzread, 16384)
20 #define VC_KEEPALT 256
21 #define VC_ACGT_ONLY 512
23 #define VC_CALL_GT 2048
25 #define VC_NO_INDEL 8192
26 #define VC_ANNO_MAX 16384
27 #define VC_FIX_PL 32768
29 #define VC_PAIRCALL 0x20000
30 #define VC_QCNT 0x40000
33 int flag, prior_type, n1, n_sub, *sublist, n_perm;
35 char *prior_file, **subsam, *fn_dict;
37 double theta, pref, indel_frac, min_perm_p, min_smpl_frac, min_lrt;
41 void *bed_read(const char *fn);
42 void bed_destroy(void *_h);
43 int bed_overlap(const void *_h, const char *chr, int beg, int end);
47 int mq, depth, is_tested, d[4];
50 static double ttest(int n1, int n2, int a[4])
52 extern double kf_betai(double a, double b, double x);
54 if (n1 == 0 || n2 == 0 || n1 + n2 < 3) return 1.0;
55 u1 = (double)a[0] / n1; u2 = (double)a[2] / n2;
56 if (u1 <= u2) return 1.;
57 t = (u1 - u2) / sqrt(((a[1] - n1 * u1 * u1) + (a[3] - n2 * u2 * u2)) / (n1 + n2 - 2) * (1./n1 + 1./n2));
59 // printf("%d,%d,%d,%d,%lf,%lf,%lf\n", a[0], a[1], a[2], a[3], t, u1, u2);
60 return t < 0.? 1. : .5 * kf_betai(.5*v, .5, v/(v+t*t));
63 static int test16_core(int anno[16], anno16_t *a)
65 extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
68 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
69 memcpy(a->d, anno, 4 * sizeof(int));
70 a->depth = anno[0] + anno[1] + anno[2] + anno[3];
71 a->is_tested = (anno[0] + anno[1] > 0 && anno[2] + anno[3] > 0);
72 if (a->depth == 0) return -1;
73 a->mq = (int)(sqrt((anno[9] + anno[11]) / a->depth) + .499);
74 kt_fisher_exact(anno[0], anno[1], anno[2], anno[3], &left, &right, &a->p[0]);
75 for (i = 1; i < 4; ++i)
76 a->p[i] = ttest(anno[0] + anno[1], anno[2] + anno[3], anno+4*i);
80 static int test16(bcf1_t *b, anno16_t *a)
84 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
85 a->d[0] = a->d[1] = a->d[2] = a->d[3] = 0.;
86 a->mq = a->depth = a->is_tested = 0;
87 if ((p = strstr(b->info, "I16=")) == 0) return -1;
89 for (i = 0; i < 16; ++i) {
90 errno = 0; anno[i] = strtol(p, &p, 10);
91 if (anno[i] == 0 && (errno == EINVAL || errno == ERANGE)) return -2;
94 return test16_core(anno, a);
97 static void rm_info(bcf1_t *b, const char *key)
100 if ((p = strstr(b->info, key)) == 0) return;
101 for (q = p; *q && *q != ';'; ++q);
102 if (p > b->info && *(p-1) == ';') --p;
103 memmove(p, q, b->l_str - (q - b->str));
108 static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag, double em[10], int cons_llr, int64_t cons_gt)
115 has_I16 = test16(b, &a) >= 0? 1 : 0;
116 rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
118 memset(&s, 0, sizeof(kstring_t));
119 kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
120 kputs(b->alt, &s); kputc('\0', &s); kputc('\0', &s);
122 if (b->info[0]) kputc(';', &s);
124 if (em[0] >= 0) ksprintf(&s, "AF1=%.4g", 1 - em[0]);
125 if (em[4] >= 0 && em[4] <= 0.05) ksprintf(&s, ";G3=%.4g,%.4g,%.4g;HWE=%.3g", em[3], em[2], em[1], em[4]);
126 if (em[5] >= 0 && em[6] >= 0) ksprintf(&s, ";AF2=%.4g,%.4g", 1 - em[5], 1 - em[6]);
127 if (em[7] >= 0) ksprintf(&s, ";LRT=%.3g", em[7]);
128 if (em[8] >= 0) ksprintf(&s, ";LRT2=%.3g", em[8]);
129 //if (em[9] >= 0) ksprintf(&s, ";LRT1=%.3g", em[9]);
132 ksprintf(&s, ";CLR=%d", cons_llr);
134 ksprintf(&s, ";UGT=%c%c%c;CGT=%c%c%c", cons_gt&0xff, cons_gt>>8&0xff, cons_gt>>16&0xff,
135 cons_gt>>32&0xff, cons_gt>>40&0xff, cons_gt>>48&0xff);
137 if (pr == 0) { // if pr is unset, return
138 kputc('\0', &s); kputs(b->fmt, &s); kputc('\0', &s);
140 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
145 is_var = (pr->p_ref < pref);
146 r = is_var? pr->p_ref : pr->p_var;
148 // ksprintf(&s, ";CI95=%.4g,%.4g", pr->cil, pr->cih); // FIXME: when EM is not used, ";" should be omitted!
149 ksprintf(&s, ";AC1=%d", pr->ac);
150 if (has_I16) ksprintf(&s, ";DP4=%d,%d,%d,%d;MQ=%d", a.d[0], a.d[1], a.d[2], a.d[3], a.mq);
151 fq = pr->p_ref_folded < 0.5? -4.343 * log(pr->p_ref_folded) : 4.343 * log(pr->p_var_folded);
152 if (fq < -999) fq = -999;
153 if (fq > 999) fq = 999;
154 ksprintf(&s, ";FQ=%.3g", fq);
155 if (pr->cmp[0] >= 0.) { // two sample groups
157 for (i = 1; i < 3; ++i) {
158 double x = pr->cmp[i] + pr->cmp[0]/2.;
159 q[i] = x == 0? 255 : (int)(-4.343 * log(x) + .499);
160 if (q[i] > 255) q[i] = 255;
162 if (pr->perm_rank >= 0) ksprintf(&s, ";PR=%d", pr->perm_rank);
163 // ksprintf(&s, ";LRT3=%.3g", pr->lrt);
164 ksprintf(&s, ";PCHI2=%.3g;PC2=%d,%d", q[1], q[2], pr->p_chi2);
166 if (has_I16 && a.is_tested) ksprintf(&s, ";PV4=%.2g,%.2g,%.2g,%.2g", a.p[0], a.p[1], a.p[2], a.p[3]);
168 kputs(b->fmt, &s); kputc('\0', &s);
170 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
171 b->qual = r < 1e-100? 999 : -4.343 * log(r);
172 if (b->qual > 999) b->qual = 999;
174 if (!is_var) bcf_shrink_alt(b, 1);
175 else if (!(flag&VC_KEEPALT))
176 bcf_shrink_alt(b, pr->rank0 < 2? 2 : pr->rank0+1);
177 if (is_var && (flag&VC_CALL_GT)) { // call individual genotype
178 int i, x, old_n_gi = b->n_gi;
179 s.m = b->m_str; s.l = b->l_str - 1; s.s = b->str;
180 kputs(":GT:GQ", &s); kputc('\0', &s);
181 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
183 for (i = 0; i < b->n_smpl; ++i) {
184 x = bcf_p1_call_gt(pa, pr->f_exp, i);
185 ((uint8_t*)b->gi[old_n_gi].data)[i] = (x&3) == 0? 1<<3|1 : (x&3) == 1? 1 : 0;
186 ((uint8_t*)b->gi[old_n_gi+1].data)[i] = x>>2;
192 static char **read_samples(const char *fn, int *_n)
197 int dret, n = 0, max = 0;
200 s.l = s.m = 0; s.s = 0;
201 fp = gzopen(fn, "r");
202 if (fp == 0) return 0; // fail to open file
204 while (ks_getuntil(ks, 0, &s, &dret) >= 0) {
207 max = max? max<<1 : 4;
208 sam = realloc(sam, sizeof(void*)*max);
211 sam[n] = malloc(s.l + 2);
213 sam[n][l+1] = 2; // by default, diploid
215 if (ks_getuntil(ks, 0, &s, &dret) >= 0) { // read ploidy, 1 or 2
216 int x = (int)s.s[0] - '0';
217 if (x == 1 || x == 2) sam[n][l+1] = x;
218 else fprintf(stderr, "(%s) ploidy can only be 1 or 2; assume diploid\n", __func__);
220 if (dret != '\n') ks_getuntil(ks, '\n', &s, &dret);
231 static void write_header(bcf_hdr_t *h)
234 str.l = h->l_txt? h->l_txt - 1 : 0;
235 str.m = str.l + 1; str.s = h->txt;
236 if (!strstr(str.s, "##INFO=<ID=DP,"))
237 kputs("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">\n", &str);
238 if (!strstr(str.s, "##INFO=<ID=DP4,"))
239 kputs("##INFO=<ID=DP4,Number=4,Type=Integer,Description=\"# high-quality ref-forward bases, ref-reverse, alt-forward and alt-reverse bases\">\n", &str);
240 if (!strstr(str.s, "##INFO=<ID=MQ,"))
241 kputs("##INFO=<ID=MQ,Number=1,Type=Integer,Description=\"Root-mean-square mapping quality of covering reads\">\n", &str);
242 if (!strstr(str.s, "##INFO=<ID=FQ,"))
243 kputs("##INFO=<ID=FQ,Number=1,Type=Float,Description=\"Phred probability of all samples being the same\">\n", &str);
244 if (!strstr(str.s, "##INFO=<ID=AF1,"))
245 kputs("##INFO=<ID=AF1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele frequency (assuming HWE)\">\n", &str);
246 if (!strstr(str.s, "##INFO=<ID=AC1,"))
247 kputs("##INFO=<ID=AC1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele count (no HWE assumption)\">\n", &str);
248 if (!strstr(str.s, "##INFO=<ID=G3,"))
249 kputs("##INFO=<ID=G3,Number=3,Type=Float,Description=\"ML estimate of genotype frequencies\">\n", &str);
250 if (!strstr(str.s, "##INFO=<ID=HWE,"))
251 kputs("##INFO=<ID=HWE,Number=1,Type=Float,Description=\"Chi^2 based HWE test P-value based on G3\">\n", &str);
252 if (!strstr(str.s, "##INFO=<ID=CLR,"))
253 kputs("##INFO=<ID=CLR,Number=1,Type=Integer,Description=\"Log ratio of genotype likelihoods with and without the constraint\">\n", &str);
254 if (!strstr(str.s, "##INFO=<ID=UGT,"))
255 kputs("##INFO=<ID=UGT,Number=1,Type=String,Description=\"The most probable unconstrained genotype configuration in the trio\">\n", &str);
256 if (!strstr(str.s, "##INFO=<ID=CGT,"))
257 kputs("##INFO=<ID=CGT,Number=1,Type=String,Description=\"The most probable constrained genotype configuration in the trio\">\n", &str);
258 // if (!strstr(str.s, "##INFO=<ID=CI95,"))
259 // kputs("##INFO=<ID=CI95,Number=2,Type=Float,Description=\"Equal-tail Bayesian credible interval of the site allele frequency at the 95% level\">\n", &str);
260 if (!strstr(str.s, "##INFO=<ID=PV4,"))
261 kputs("##INFO=<ID=PV4,Number=4,Type=Float,Description=\"P-values for strand bias, baseQ bias, mapQ bias and tail distance bias\">\n", &str);
262 if (!strstr(str.s, "##INFO=<ID=INDEL,"))
263 kputs("##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">\n", &str);
264 if (!strstr(str.s, "##INFO=<ID=PC2,"))
265 kputs("##INFO=<ID=PC2,Number=2,Type=Integer,Description=\"Phred probability of the nonRef allele frequency in group1 samples being larger (,smaller) than in group2.\">\n", &str);
266 if (!strstr(str.s, "##INFO=<ID=PCHI2,"))
267 kputs("##INFO=<ID=PCHI2,Number=1,Type=Float,Description=\"Posterior weighted chi^2 P-value for testing the association between group1 and group2 samples.\">\n", &str);
268 if (!strstr(str.s, "##INFO=<ID=QCHI2,"))
269 kputs("##INFO=<ID=QCHI2,Number=1,Type=Integer,Description=\"Phred scaled PCHI2.\">\n", &str);
270 if (!strstr(str.s, "##INFO=<ID=RP,"))
271 kputs("##INFO=<ID=PR,Number=1,Type=Integer,Description=\"# permutations yielding a smaller PCHI2.\">\n", &str);
272 if (!strstr(str.s, "##FORMAT=<ID=GT,"))
273 kputs("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">\n", &str);
274 if (!strstr(str.s, "##FORMAT=<ID=GQ,"))
275 kputs("##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">\n", &str);
276 if (!strstr(str.s, "##FORMAT=<ID=GL,"))
277 kputs("##FORMAT=<ID=GL,Number=3,Type=Float,Description=\"Likelihoods for RR,RA,AA genotypes (R=ref,A=alt)\">\n", &str);
278 if (!strstr(str.s, "##FORMAT=<ID=DP,"))
279 kputs("##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"# high-quality bases\">\n", &str);
280 if (!strstr(str.s, "##FORMAT=<ID=SP,"))
281 kputs("##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">\n", &str);
282 if (!strstr(str.s, "##FORMAT=<ID=PL,"))
283 kputs("##FORMAT=<ID=PL,Number=-1,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods, number of values is (#ALT+1)*(#ALT+2)/2\">\n", &str);
284 h->l_txt = str.l + 1; h->txt = str.s;
287 double bcf_pair_freq(const bcf1_t *b0, const bcf1_t *b1, double f[4]);
289 int bcfview(int argc, char *argv[])
291 extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b);
292 extern void bcf_p1_indel_prior(bcf_p1aux_t *ma, double x);
293 extern int bcf_fix_gt(bcf1_t *b);
294 extern int bcf_anno_max(bcf1_t *b);
295 extern int bcf_shuffle(bcf1_t *b, int seed);
296 extern uint32_t *bcf_trio_prep(int is_x, int is_son);
297 extern int bcf_trio_call(uint32_t *prep, const bcf1_t *b, int *llr, int64_t *gt);
298 extern int bcf_pair_call(const bcf1_t *b);
299 extern int bcf_min_diff(const bcf1_t *b);
301 bcf_t *bp, *bout = 0;
304 uint64_t n_processed = 0, qcnt[256];
307 bcf_hdr_t *hin, *hout;
309 char moder[4], modew[4];
311 tid = begin = end = -1;
312 memset(&vc, 0, sizeof(viewconf_t));
313 vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5; vc.indel_frac = -1.; vc.n_perm = 0; vc.min_perm_p = 0.01; vc.min_smpl_frac = 0; vc.min_lrt = 1;
314 memset(qcnt, 0, 8 * 256);
315 while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:T:Y")) >= 0) {
317 case '1': vc.n1 = atoi(optarg); break;
318 case 'l': vc.bed = bed_read(optarg); break;
319 case 'D': vc.fn_dict = strdup(optarg); break;
320 case 'F': vc.flag |= VC_FIX_PL; break;
321 case 'N': vc.flag |= VC_ACGT_ONLY; break;
322 case 'G': vc.flag |= VC_NO_GENO; break;
323 case 'A': vc.flag |= VC_KEEPALT; break;
324 case 'b': vc.flag |= VC_BCFOUT; break;
325 case 'S': vc.flag |= VC_VCFIN; break;
326 case 'c': vc.flag |= VC_CALL; break;
327 case 'e': vc.flag |= VC_EM; break;
328 case 'v': vc.flag |= VC_VARONLY | VC_CALL; break;
329 case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
330 case 'g': vc.flag |= VC_CALL_GT | VC_CALL; break;
331 case 'I': vc.flag |= VC_NO_INDEL; break;
332 case 'M': vc.flag |= VC_ANNO_MAX; break;
333 case 'Y': vc.flag |= VC_QCNT; break;
334 case 't': vc.theta = atof(optarg); break;
335 case 'p': vc.pref = atof(optarg); break;
336 case 'i': vc.indel_frac = atof(optarg); break;
337 case 'Q': vc.flag |= VC_QCALL; break;
338 case 'L': vc.flag |= VC_ADJLD; break;
339 case 'U': vc.n_perm = atoi(optarg); break;
340 case 'C': vc.min_lrt = atof(optarg); break;
341 case 'X': vc.min_perm_p = atof(optarg); break;
342 case 'd': vc.min_smpl_frac = atof(optarg); break;
343 case 's': vc.subsam = read_samples(optarg, &vc.n_sub);
344 vc.ploidy = calloc(vc.n_sub + 1, 1);
345 for (tid = 0; tid < vc.n_sub; ++tid) vc.ploidy[tid] = vc.subsam[tid][strlen(vc.subsam[tid]) + 1];
349 if (strcmp(optarg, "trioauto") == 0) vc.trio_aux = bcf_trio_prep(0, 0);
350 else if (strcmp(optarg, "trioxd") == 0) vc.trio_aux = bcf_trio_prep(1, 0);
351 else if (strcmp(optarg, "trioxs") == 0) vc.trio_aux = bcf_trio_prep(1, 1);
352 else if (strcmp(optarg, "pair") == 0) vc.flag |= VC_PAIRCALL;
354 fprintf(stderr, "[%s] Option '-T' can only take value trioauto, trioxd or trioxs.\n", __func__);
359 if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
360 else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
361 else if (strcmp(optarg, "flat") == 0) vc.prior_type = MC_PTYPE_FLAT;
362 else vc.prior_file = strdup(optarg);
366 if (argc == optind) {
367 fprintf(stderr, "\n");
368 fprintf(stderr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
369 fprintf(stderr, "Input/output options:\n\n");
370 fprintf(stderr, " -A keep all possible alternate alleles at variant sites\n");
371 fprintf(stderr, " -b output BCF instead of VCF\n");
372 fprintf(stderr, " -D FILE sequence dictionary for VCF->BCF conversion [null]\n");
373 fprintf(stderr, " -F PL generated by r921 or before (which generate old ordering)\n");
374 fprintf(stderr, " -G suppress all individual genotype information\n");
375 fprintf(stderr, " -l FILE list of sites (chr pos) or regions (BED) to output [all sites]\n");
376 fprintf(stderr, " -L calculate LD for adjacent sites\n");
377 fprintf(stderr, " -N skip sites where REF is not A/C/G/T\n");
378 fprintf(stderr, " -Q output the QCALL likelihood format\n");
379 fprintf(stderr, " -s FILE list of samples to use [all samples]\n");
380 fprintf(stderr, " -S input is VCF\n");
381 fprintf(stderr, " -u uncompressed BCF output (force -b)\n");
382 fprintf(stderr, "\nConsensus/variant calling options:\n\n");
383 fprintf(stderr, " -c SNP calling (force -e)\n");
384 fprintf(stderr, " -d FLOAT skip loci where less than FLOAT fraction of samples covered [0]\n");
385 fprintf(stderr, " -e likelihood based analyses\n");
386 fprintf(stderr, " -g call genotypes at variant sites (force -c)\n");
387 fprintf(stderr, " -i FLOAT indel-to-substitution ratio [%.4g]\n", vc.indel_frac);
388 fprintf(stderr, " -I skip indels\n");
389 fprintf(stderr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3g]\n", vc.pref);
390 fprintf(stderr, " -P STR type of prior: full, cond2, flat [full]\n");
391 fprintf(stderr, " -t FLOAT scaled substitution mutation rate [%.4g]\n", vc.theta);
392 fprintf(stderr, " -T STR constrained calling; STR can be: pair, trioauto, trioxd and trioxs (see manual) [null]\n");
393 fprintf(stderr, " -v output potential variant sites only (force -c)\n");
394 fprintf(stderr, "\nContrast calling and association test options:\n\n");
395 fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
396 fprintf(stderr, " -C FLOAT posterior constrast for LRT<FLOAT and P(ref|D)<0.5 [%g]\n", vc.min_lrt);
397 fprintf(stderr, " -U INT number of permutations for association testing (effective with -1) [0]\n");
398 fprintf(stderr, " -X FLOAT only perform permutations for P(chi^2)<FLOAT [%g]\n", vc.min_perm_p);
399 fprintf(stderr, "\n");
403 if (vc.flag & VC_CALL) vc.flag |= VC_EM;
404 if ((vc.flag & VC_VCFIN) && (vc.flag & VC_BCFOUT) && vc.fn_dict == 0) {
405 fprintf(stderr, "[%s] For VCF->BCF conversion please specify the sequence dictionary with -D\n", __func__);
408 if (vc.n1 <= 0) vc.n_perm = 0; // TODO: give a warning here!
410 seeds = malloc(vc.n_perm * sizeof(int));
412 for (c = 0; c < vc.n_perm; ++c) seeds[c] = lrand48();
414 b = calloc(1, sizeof(bcf1_t));
415 blast = calloc(1, sizeof(bcf1_t));
417 if (!(vc.flag & VC_VCFIN)) strcat(moder, "b");
419 if (vc.flag & VC_BCFOUT) strcat(modew, "b");
420 if (vc.flag & VC_UNCOMP) strcat(modew, "u");
421 bp = vcf_open(argv[optind], moder);
422 hin = hout = vcf_hdr_read(bp);
423 if (vc.fn_dict && (vc.flag & VC_VCFIN))
424 vcf_dictread(bp, hin, vc.fn_dict);
425 bout = vcf_open("-", modew);
426 if (!(vc.flag & VC_QCALL)) {
428 vc.sublist = calloc(vc.n_sub, sizeof(int));
429 hout = bcf_hdr_subsam(hin, vc.n_sub, vc.subsam, vc.sublist);
431 if (vc.flag & VC_CALL) write_header(hout);
432 vcf_hdr_write(bout, hout);
434 if (vc.flag & VC_CALL) {
435 p1 = bcf_p1_init(hout->n_smpl, vc.ploidy);
437 if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
438 fprintf(stderr, "[%s] fail to read the prior AFS.\n", __func__);
441 } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
442 if (vc.n1 > 0 && vc.min_lrt > 0.) { // set n1
443 bcf_p1_set_n1(p1, vc.n1);
444 bcf_p1_init_subprior(p1, vc.prior_type, vc.theta);
446 if (vc.indel_frac > 0.) bcf_p1_indel_prior(p1, vc.indel_frac); // otherwise use the default indel_frac
448 if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
449 void *str2id = bcf_build_refhash(hout);
450 if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
452 idx = bcf_idx_load(argv[optind]);
455 off = bcf_idx_query(idx, tid, begin);
457 fprintf(stderr, "[%s] no records in the query region.\n", __func__);
458 return 1; // FIXME: a lot of memory leaks...
460 bgzf_seek(bp->fp, off, SEEK_SET);
461 bcf_idx_destroy(idx);
465 while (vcf_read(bp, hin, b) > 0) {
466 int is_indel, cons_llr = -1;
467 int64_t cons_gt = -1;
469 if ((vc.flag & VC_VARONLY) && strcmp(b->alt, "X") == 0) continue;
470 if ((vc.flag & VC_VARONLY) && vc.min_smpl_frac > 0.) {
471 extern int bcf_smpl_covered(const bcf1_t *b);
472 int n = bcf_smpl_covered(b);
473 if ((double)n / b->n_smpl < vc.min_smpl_frac) continue;
475 if (vc.n_sub) bcf_subsam(vc.n_sub, vc.sublist, b);
476 if (vc.flag & VC_FIX_PL) bcf_fix_pl(b);
477 is_indel = bcf_is_indel(b);
478 if ((vc.flag & VC_NO_INDEL) && is_indel) continue;
479 if ((vc.flag & VC_ACGT_ONLY) && !is_indel) {
481 if (b->ref[0] == 0 || b->ref[1] != 0) continue;
482 x = toupper(b->ref[0]);
483 if (x != 'A' && x != 'C' && x != 'G' && x != 'T') continue;
485 if (vc.bed && !bed_overlap(vc.bed, hin->ns[b->tid], b->pos, b->pos + strlen(b->ref))) continue;
487 int l = strlen(b->ref);
488 l = b->pos + (l > 0? l : 1);
489 if (b->tid != tid || b->pos >= end) break;
490 if (!(l > begin && end > b->pos)) continue;
493 if (vc.flag & VC_QCNT) { // summarize the difference
494 int x = bcf_min_diff(b);
495 if (x > 255) x = 255;
496 if (x >= 0) ++qcnt[x];
498 if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
502 if (vc.trio_aux) // do trio calling
503 bcf_trio_call(vc.trio_aux, b, &cons_llr, &cons_gt);
504 else if (vc.flag & VC_PAIRCALL)
505 cons_llr = bcf_pair_call(b);
506 if (vc.flag & (VC_CALL|VC_ADJLD|VC_EM)) bcf_gl2pl(b);
507 if (vc.flag & VC_EM) bcf_em1(b, vc.n1, 0x1ff, em);
510 for (i = 0; i < 9; ++i) em[i] = -1.;
512 if (vc.flag & VC_CALL) { // call variants
514 int calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
515 if (n_processed % 100000 == 0) {
516 fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
519 if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
520 if (vc.n_perm && vc.n1 > 0 && pr.p_chi2 < vc.min_perm_p) { // permutation test
523 for (i = 0; i < vc.n_perm; ++i) {
524 #ifdef BCF_PERM_LRT // LRT based permutation is much faster but less robust to artifacts
526 bcf_shuffle(b, seeds[i]);
527 bcf_em1(b, vc.n1, 1<<7, x);
528 if (x[7] < em[7]) ++n;
530 bcf_shuffle(b, seeds[i]);
531 bcf_p1_cal(b, 1, p1, &r);
532 if (pr.p_chi2 >= r.p_chi2) ++n;
537 if (calret >= 0) update_bcf1(b, p1, &pr, vc.pref, vc.flag, em, cons_llr, cons_gt);
538 } else if (vc.flag & VC_EM) update_bcf1(b, 0, 0, 0, vc.flag, em, cons_llr, cons_gt);
539 if (vc.flag & VC_ADJLD) { // compute LD
541 if ((r2 = bcf_pair_freq(blast, b, f)) >= 0) {
543 s.m = s.l = 0; s.s = 0;
544 if (*b->info) kputc(';', &s);
545 ksprintf(&s, "NEIR=%.3f;NEIF4=%.3f,%.3f,%.3f,%.3f", r2, f[0], f[1], f[2], f[3]);
546 bcf_append_info(b, s.s, s.l);
551 if (vc.flag & VC_ANNO_MAX) bcf_anno_max(b);
552 if (vc.flag & VC_NO_GENO) { // do not output GENO fields
555 b->l_str = b->fmt - b->str + 1;
556 } else bcf_fix_gt(b);
557 vcf_write(bout, hout, b);
559 if (vc.prior_file) free(vc.prior_file);
560 if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
561 if (hin != hout) bcf_hdr_destroy(hout);
562 bcf_hdr_destroy(hin);
563 bcf_destroy(b); bcf_destroy(blast);
564 vcf_close(bp); vcf_close(bout);
565 if (vc.fn_dict) free(vc.fn_dict);
566 if (vc.ploidy) free(vc.ploidy);
567 if (vc.trio_aux) free(vc.trio_aux);
570 for (i = 0; i < vc.n_sub; ++i) free(vc.subsam[i]);
571 free(vc.subsam); free(vc.sublist);
573 if (vc.bed) bed_destroy(vc.bed);
574 if (vc.flag & VC_QCNT)
575 for (c = 0; c < 256; ++c)
576 fprintf(stderr, "QT\t%d\t%lld\n", c, (long long)qcnt[c]);
577 if (seeds) free(seeds);
578 if (p1) bcf_p1_destroy(p1);