14 #define srand48(x) srand(x)
15 #define lrand48() rand()
19 KSTREAM_INIT(gzFile, gzread, 16384)
27 #define VC_KEEPALT 256
28 #define VC_ACGT_ONLY 512
30 #define VC_CALL_GT 2048
32 #define VC_NO_INDEL 8192
33 #define VC_ANNO_MAX 16384
34 #define VC_FIX_PL 32768
36 #define VC_PAIRCALL 0x20000
37 #define VC_QCNT 0x40000
40 int flag, prior_type, n1, n_sub, *sublist, n_perm;
42 char *prior_file, **subsam, *fn_dict;
44 double theta, pref, indel_frac, min_perm_p, min_smpl_frac, min_lrt;
48 void *bed_read(const char *fn);
49 void bed_destroy(void *_h);
50 int bed_overlap(const void *_h, const char *chr, int beg, int end);
54 int mq, depth, is_tested, d[4];
57 static double ttest(int n1, int n2, int a[4])
59 extern double kf_betai(double a, double b, double x);
61 if (n1 == 0 || n2 == 0 || n1 + n2 < 3) return 1.0;
62 u1 = (double)a[0] / n1; u2 = (double)a[2] / n2;
63 if (u1 <= u2) return 1.;
64 t = (u1 - u2) / sqrt(((a[1] - n1 * u1 * u1) + (a[3] - n2 * u2 * u2)) / (n1 + n2 - 2) * (1./n1 + 1./n2));
66 // printf("%d,%d,%d,%d,%lf,%lf,%lf\n", a[0], a[1], a[2], a[3], t, u1, u2);
67 return t < 0.? 1. : .5 * kf_betai(.5*v, .5, v/(v+t*t));
70 static int test16_core(int anno[16], anno16_t *a)
72 extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
75 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
76 memcpy(a->d, anno, 4 * sizeof(int));
77 a->depth = anno[0] + anno[1] + anno[2] + anno[3];
78 a->is_tested = (anno[0] + anno[1] > 0 && anno[2] + anno[3] > 0);
79 if (a->depth == 0) return -1;
80 a->mq = (int)(sqrt((anno[9] + anno[11]) / a->depth) + .499);
81 kt_fisher_exact(anno[0], anno[1], anno[2], anno[3], &left, &right, &a->p[0]);
82 for (i = 1; i < 4; ++i)
83 a->p[i] = ttest(anno[0] + anno[1], anno[2] + anno[3], anno+4*i);
87 static int test16(bcf1_t *b, anno16_t *a)
91 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
92 a->d[0] = a->d[1] = a->d[2] = a->d[3] = 0.;
93 a->mq = a->depth = a->is_tested = 0;
94 if ((p = strstr(b->info, "I16=")) == 0) return -1;
96 for (i = 0; i < 16; ++i) {
97 errno = 0; anno[i] = strtol(p, &p, 10);
98 if (anno[i] == 0 && (errno == EINVAL || errno == ERANGE)) return -2;
101 return test16_core(anno, a);
104 static void rm_info(bcf1_t *b, const char *key)
107 if ((p = strstr(b->info, key)) == 0) return;
108 for (q = p; *q && *q != ';'; ++q);
109 if (p > b->info && *(p-1) == ';') --p;
110 memmove(p, q, b->l_str - (q - b->str));
115 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)
122 has_I16 = test16(b, &a) >= 0? 1 : 0;
123 rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
125 memset(&s, 0, sizeof(kstring_t));
126 kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
127 kputs(b->alt, &s); kputc('\0', &s); kputc('\0', &s);
129 if (b->info[0]) kputc(';', &s);
131 if (em[0] >= 0) ksprintf(&s, "AF1=%.4g", 1 - em[0]);
132 if (em[4] >= 0 && em[4] <= 0.05) ksprintf(&s, ";G3=%.4g,%.4g,%.4g;HWE=%.3g", em[3], em[2], em[1], em[4]);
133 if (em[5] >= 0 && em[6] >= 0) ksprintf(&s, ";AF2=%.4g,%.4g", 1 - em[5], 1 - em[6]);
134 if (em[7] >= 0) ksprintf(&s, ";LRT=%.3g", em[7]);
135 if (em[8] >= 0) ksprintf(&s, ";LRT2=%.3g", em[8]);
138 ksprintf(&s, ";CLR=%d", cons_llr);
140 ksprintf(&s, ";UGT=%c%c%c;CGT=%c%c%c", cons_gt&0xff, cons_gt>>8&0xff, cons_gt>>16&0xff,
141 cons_gt>>32&0xff, cons_gt>>40&0xff, cons_gt>>48&0xff);
143 if (pr == 0) { // if pr is unset, return
144 kputc('\0', &s); kputs(b->fmt, &s); kputc('\0', &s);
146 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
151 is_var = (pr->p_ref < pref);
152 r = is_var? pr->p_ref : pr->p_var;
154 // ksprintf(&s, ";CI95=%.4g,%.4g", pr->cil, pr->cih); // FIXME: when EM is not used, ";" should be omitted!
155 ksprintf(&s, ";AC1=%d", pr->ac);
156 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);
157 fq = pr->p_ref_folded < 0.5? -4.343 * log(pr->p_ref_folded) : 4.343 * log(pr->p_var_folded);
158 if (fq < -999) fq = -999;
159 if (fq > 999) fq = 999;
160 ksprintf(&s, ";FQ=%.3g", fq);
161 if (pr->cmp[0] >= 0.) { // two sample groups
163 for (i = 1; i < 3; ++i) {
164 double x = pr->cmp[i] + pr->cmp[0]/2.;
165 q[i] = x == 0? 255 : (int)(-4.343 * log(x) + .499);
166 if (q[i] > 255) q[i] = 255;
168 if (pr->perm_rank >= 0) ksprintf(&s, ";PR=%d", pr->perm_rank);
169 // ksprintf(&s, ";LRT3=%.3g", pr->lrt);
170 ksprintf(&s, ";PCHI2=%.3g;PC2=%d,%d", q[1], q[2], pr->p_chi2);
172 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]);
174 kputs(b->fmt, &s); kputc('\0', &s);
176 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
177 b->qual = r < 1e-100? 999 : -4.343 * log(r);
178 if (b->qual > 999) b->qual = 999;
180 if (!is_var) bcf_shrink_alt(b, 1);
181 else if (!(flag&VC_KEEPALT))
182 bcf_shrink_alt(b, pr->rank0 < 2? 2 : pr->rank0+1);
183 if (is_var && (flag&VC_CALL_GT)) { // call individual genotype
184 int i, x, old_n_gi = b->n_gi;
185 s.m = b->m_str; s.l = b->l_str - 1; s.s = b->str;
186 kputs(":GT:GQ", &s); kputc('\0', &s);
187 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
189 for (i = 0; i < b->n_smpl; ++i) {
190 x = bcf_p1_call_gt(pa, pr->f_exp, i);
191 ((uint8_t*)b->gi[old_n_gi].data)[i] = (x&3) == 0? 1<<3|1 : (x&3) == 1? 1 : 0;
192 ((uint8_t*)b->gi[old_n_gi+1].data)[i] = x>>2;
198 static char **read_samples(const char *fn, int *_n)
203 int dret, n = 0, max = 0;
206 s.l = s.m = 0; s.s = 0;
207 fp = gzopen(fn, "r");
208 if (fp == 0) return 0; // fail to open file
210 while (ks_getuntil(ks, 0, &s, &dret) >= 0) {
213 max = max? max<<1 : 4;
214 sam = realloc(sam, sizeof(void*)*max);
217 sam[n] = malloc(s.l + 2);
219 sam[n][l+1] = 2; // by default, diploid
221 if (ks_getuntil(ks, 0, &s, &dret) >= 0) { // read ploidy, 1 or 2
222 int x = (int)s.s[0] - '0';
223 if (x == 1 || x == 2) sam[n][l+1] = x;
224 else fprintf(pysamerr, "(%s) ploidy can only be 1 or 2; assume diploid\n", __func__);
226 if (dret != '\n') ks_getuntil(ks, '\n', &s, &dret);
237 static void write_header(bcf_hdr_t *h)
240 str.l = h->l_txt? h->l_txt - 1 : 0;
241 str.m = str.l + 1; str.s = h->txt;
242 if (!strstr(str.s, "##INFO=<ID=DP,"))
243 kputs("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">\n", &str);
244 if (!strstr(str.s, "##INFO=<ID=DP4,"))
245 kputs("##INFO=<ID=DP4,Number=4,Type=Integer,Description=\"# high-quality ref-forward bases, ref-reverse, alt-forward and alt-reverse bases\">\n", &str);
246 if (!strstr(str.s, "##INFO=<ID=MQ,"))
247 kputs("##INFO=<ID=MQ,Number=1,Type=Integer,Description=\"Root-mean-square mapping quality of covering reads\">\n", &str);
248 if (!strstr(str.s, "##INFO=<ID=FQ,"))
249 kputs("##INFO=<ID=FQ,Number=1,Type=Float,Description=\"Phred probability of all samples being the same\">\n", &str);
250 if (!strstr(str.s, "##INFO=<ID=AF1,"))
251 kputs("##INFO=<ID=AF1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele frequency (assuming HWE)\">\n", &str);
252 if (!strstr(str.s, "##INFO=<ID=AC1,"))
253 kputs("##INFO=<ID=AC1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele count (no HWE assumption)\">\n", &str);
254 if (!strstr(str.s, "##INFO=<ID=G3,"))
255 kputs("##INFO=<ID=G3,Number=3,Type=Float,Description=\"ML estimate of genotype frequencies\">\n", &str);
256 if (!strstr(str.s, "##INFO=<ID=HWE,"))
257 kputs("##INFO=<ID=HWE,Number=1,Type=Float,Description=\"Chi^2 based HWE test P-value based on G3\">\n", &str);
258 if (!strstr(str.s, "##INFO=<ID=CLR,"))
259 kputs("##INFO=<ID=CLR,Number=1,Type=Integer,Description=\"Log ratio of genotype likelihoods with and without the constraint\">\n", &str);
260 if (!strstr(str.s, "##INFO=<ID=UGT,"))
261 kputs("##INFO=<ID=UGT,Number=1,Type=String,Description=\"The most probable unconstrained genotype configuration in the trio\">\n", &str);
262 if (!strstr(str.s, "##INFO=<ID=CGT,"))
263 kputs("##INFO=<ID=CGT,Number=1,Type=String,Description=\"The most probable constrained genotype configuration in the trio\">\n", &str);
264 // if (!strstr(str.s, "##INFO=<ID=CI95,"))
265 // 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);
266 if (!strstr(str.s, "##INFO=<ID=PV4,"))
267 kputs("##INFO=<ID=PV4,Number=4,Type=Float,Description=\"P-values for strand bias, baseQ bias, mapQ bias and tail distance bias\">\n", &str);
268 if (!strstr(str.s, "##INFO=<ID=INDEL,"))
269 kputs("##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">\n", &str);
270 if (!strstr(str.s, "##INFO=<ID=PC2,"))
271 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);
272 if (!strstr(str.s, "##INFO=<ID=PCHI2,"))
273 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);
274 if (!strstr(str.s, "##INFO=<ID=QCHI2,"))
275 kputs("##INFO=<ID=QCHI2,Number=1,Type=Integer,Description=\"Phred scaled PCHI2.\">\n", &str);
276 if (!strstr(str.s, "##INFO=<ID=RP,"))
277 kputs("##INFO=<ID=PR,Number=1,Type=Integer,Description=\"# permutations yielding a smaller PCHI2.\">\n", &str);
278 if (!strstr(str.s, "##INFO=<ID=VDB,"))
279 kputs("##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias\">\n", &str);
280 if (!strstr(str.s, "##FORMAT=<ID=GT,"))
281 kputs("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">\n", &str);
282 if (!strstr(str.s, "##FORMAT=<ID=GQ,"))
283 kputs("##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">\n", &str);
284 if (!strstr(str.s, "##FORMAT=<ID=GL,"))
285 kputs("##FORMAT=<ID=GL,Number=3,Type=Float,Description=\"Likelihoods for RR,RA,AA genotypes (R=ref,A=alt)\">\n", &str);
286 if (!strstr(str.s, "##FORMAT=<ID=DP,"))
287 kputs("##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"# high-quality bases\">\n", &str);
288 if (!strstr(str.s, "##FORMAT=<ID=SP,"))
289 kputs("##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">\n", &str);
290 if (!strstr(str.s, "##FORMAT=<ID=PL,"))
291 kputs("##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">\n", &str);
292 h->l_txt = str.l + 1; h->txt = str.s;
295 double bcf_pair_freq(const bcf1_t *b0, const bcf1_t *b1, double f[4]);
297 int bcfview(int argc, char *argv[])
299 extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b);
300 extern void bcf_p1_indel_prior(bcf_p1aux_t *ma, double x);
301 extern int bcf_fix_gt(bcf1_t *b);
302 extern int bcf_anno_max(bcf1_t *b);
303 extern int bcf_shuffle(bcf1_t *b, int seed);
304 extern uint32_t *bcf_trio_prep(int is_x, int is_son);
305 extern int bcf_trio_call(uint32_t *prep, const bcf1_t *b, int *llr, int64_t *gt);
306 extern int bcf_pair_call(const bcf1_t *b);
307 extern int bcf_min_diff(const bcf1_t *b);
309 bcf_t *bp, *bout = 0;
312 uint64_t n_processed = 0, qcnt[256];
315 bcf_hdr_t *hin, *hout;
317 char moder[4], modew[4];
319 tid = begin = end = -1;
320 memset(&vc, 0, sizeof(viewconf_t));
321 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;
322 memset(qcnt, 0, 8 * 256);
323 while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:T:Y")) >= 0) {
325 case '1': vc.n1 = atoi(optarg); break;
326 case 'l': vc.bed = bed_read(optarg); break;
327 case 'D': vc.fn_dict = strdup(optarg); break;
328 case 'F': vc.flag |= VC_FIX_PL; break;
329 case 'N': vc.flag |= VC_ACGT_ONLY; break;
330 case 'G': vc.flag |= VC_NO_GENO; break;
331 case 'A': vc.flag |= VC_KEEPALT; break;
332 case 'b': vc.flag |= VC_BCFOUT; break;
333 case 'S': vc.flag |= VC_VCFIN; break;
334 case 'c': vc.flag |= VC_CALL; break;
335 case 'e': vc.flag |= VC_EM; break;
336 case 'v': vc.flag |= VC_VARONLY | VC_CALL; break;
337 case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
338 case 'g': vc.flag |= VC_CALL_GT | VC_CALL; break;
339 case 'I': vc.flag |= VC_NO_INDEL; break;
340 case 'M': vc.flag |= VC_ANNO_MAX; break;
341 case 'Y': vc.flag |= VC_QCNT; break;
342 case 't': vc.theta = atof(optarg); break;
343 case 'p': vc.pref = atof(optarg); break;
344 case 'i': vc.indel_frac = atof(optarg); break;
345 case 'Q': vc.flag |= VC_QCALL; break;
346 case 'L': vc.flag |= VC_ADJLD; break;
347 case 'U': vc.n_perm = atoi(optarg); break;
348 case 'C': vc.min_lrt = atof(optarg); break;
349 case 'X': vc.min_perm_p = atof(optarg); break;
350 case 'd': vc.min_smpl_frac = atof(optarg); break;
351 case 's': vc.subsam = read_samples(optarg, &vc.n_sub);
352 vc.ploidy = calloc(vc.n_sub + 1, 1);
353 for (tid = 0; tid < vc.n_sub; ++tid) vc.ploidy[tid] = vc.subsam[tid][strlen(vc.subsam[tid]) + 1];
357 if (strcmp(optarg, "trioauto") == 0) vc.trio_aux = bcf_trio_prep(0, 0);
358 else if (strcmp(optarg, "trioxd") == 0) vc.trio_aux = bcf_trio_prep(1, 0);
359 else if (strcmp(optarg, "trioxs") == 0) vc.trio_aux = bcf_trio_prep(1, 1);
360 else if (strcmp(optarg, "pair") == 0) vc.flag |= VC_PAIRCALL;
362 fprintf(pysamerr, "[%s] Option '-T' can only take value trioauto, trioxd or trioxs.\n", __func__);
367 if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
368 else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
369 else if (strcmp(optarg, "flat") == 0) vc.prior_type = MC_PTYPE_FLAT;
370 else vc.prior_file = strdup(optarg);
374 if (argc == optind) {
375 fprintf(pysamerr, "\n");
376 fprintf(pysamerr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
377 fprintf(pysamerr, "Input/output options:\n\n");
378 fprintf(pysamerr, " -A keep all possible alternate alleles at variant sites\n");
379 fprintf(pysamerr, " -b output BCF instead of VCF\n");
380 fprintf(pysamerr, " -D FILE sequence dictionary for VCF->BCF conversion [null]\n");
381 fprintf(pysamerr, " -F PL generated by r921 or before (which generate old ordering)\n");
382 fprintf(pysamerr, " -G suppress all individual genotype information\n");
383 fprintf(pysamerr, " -l FILE list of sites (chr pos) or regions (BED) to output [all sites]\n");
384 fprintf(pysamerr, " -L calculate LD for adjacent sites\n");
385 fprintf(pysamerr, " -N skip sites where REF is not A/C/G/T\n");
386 fprintf(pysamerr, " -Q output the QCALL likelihood format\n");
387 fprintf(pysamerr, " -s FILE list of samples to use [all samples]\n");
388 fprintf(pysamerr, " -S input is VCF\n");
389 fprintf(pysamerr, " -u uncompressed BCF output (force -b)\n");
390 fprintf(pysamerr, "\nConsensus/variant calling options:\n\n");
391 fprintf(pysamerr, " -c SNP calling (force -e)\n");
392 fprintf(pysamerr, " -d FLOAT skip loci where less than FLOAT fraction of samples covered [0]\n");
393 fprintf(pysamerr, " -e likelihood based analyses\n");
394 fprintf(pysamerr, " -g call genotypes at variant sites (force -c)\n");
395 fprintf(pysamerr, " -i FLOAT indel-to-substitution ratio [%.4g]\n", vc.indel_frac);
396 fprintf(pysamerr, " -I skip indels\n");
397 fprintf(pysamerr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3g]\n", vc.pref);
398 fprintf(pysamerr, " -P STR type of prior: full, cond2, flat [full]\n");
399 fprintf(pysamerr, " -t FLOAT scaled substitution mutation rate [%.4g]\n", vc.theta);
400 fprintf(pysamerr, " -T STR constrained calling; STR can be: pair, trioauto, trioxd and trioxs (see manual) [null]\n");
401 fprintf(pysamerr, " -v output potential variant sites only (force -c)\n");
402 fprintf(pysamerr, "\nContrast calling and association test options:\n\n");
403 fprintf(pysamerr, " -1 INT number of group-1 samples [0]\n");
404 fprintf(pysamerr, " -C FLOAT posterior constrast for LRT<FLOAT and P(ref|D)<0.5 [%g]\n", vc.min_lrt);
405 fprintf(pysamerr, " -U INT number of permutations for association testing (effective with -1) [0]\n");
406 fprintf(pysamerr, " -X FLOAT only perform permutations for P(chi^2)<FLOAT [%g]\n", vc.min_perm_p);
407 fprintf(pysamerr, "\n");
411 if (vc.flag & VC_CALL) vc.flag |= VC_EM;
412 if ((vc.flag & VC_VCFIN) && (vc.flag & VC_BCFOUT) && vc.fn_dict == 0) {
413 fprintf(pysamerr, "[%s] For VCF->BCF conversion please specify the sequence dictionary with -D\n", __func__);
416 if (vc.n1 <= 0) vc.n_perm = 0; // TODO: give a warning here!
418 seeds = malloc(vc.n_perm * sizeof(int));
420 for (c = 0; c < vc.n_perm; ++c) seeds[c] = lrand48();
422 b = calloc(1, sizeof(bcf1_t));
423 blast = calloc(1, sizeof(bcf1_t));
425 if (!(vc.flag & VC_VCFIN)) strcat(moder, "b");
427 if (vc.flag & VC_BCFOUT) strcat(modew, "b");
428 if (vc.flag & VC_UNCOMP) strcat(modew, "u");
429 bp = vcf_open(argv[optind], moder);
430 hin = hout = vcf_hdr_read(bp);
431 if (vc.fn_dict && (vc.flag & VC_VCFIN))
432 vcf_dictread(bp, hin, vc.fn_dict);
433 bout = vcf_open("-", modew);
434 if (!(vc.flag & VC_QCALL)) {
436 vc.sublist = calloc(vc.n_sub, sizeof(int));
437 hout = bcf_hdr_subsam(hin, vc.n_sub, vc.subsam, vc.sublist);
439 if (vc.flag & VC_CALL) write_header(hout);
440 vcf_hdr_write(bout, hout);
442 if (vc.flag & VC_CALL) {
443 p1 = bcf_p1_init(hout->n_smpl, vc.ploidy);
445 if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
446 fprintf(pysamerr, "[%s] fail to read the prior AFS.\n", __func__);
449 } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
450 if (vc.n1 > 0 && vc.min_lrt > 0.) { // set n1
451 bcf_p1_set_n1(p1, vc.n1);
452 bcf_p1_init_subprior(p1, vc.prior_type, vc.theta);
454 if (vc.indel_frac > 0.) bcf_p1_indel_prior(p1, vc.indel_frac); // otherwise use the default indel_frac
456 if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
457 void *str2id = bcf_build_refhash(hout);
458 if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
460 idx = bcf_idx_load(argv[optind]);
463 off = bcf_idx_query(idx, tid, begin);
465 fprintf(pysamerr, "[%s] no records in the query region.\n", __func__);
466 return 1; // FIXME: a lot of memory leaks...
468 bgzf_seek(bp->fp, off, SEEK_SET);
469 bcf_idx_destroy(idx);
473 while (vcf_read(bp, hin, b) > 0) {
474 int is_indel, cons_llr = -1;
475 int64_t cons_gt = -1;
477 if ((vc.flag & VC_VARONLY) && strcmp(b->alt, "X") == 0) continue;
478 if ((vc.flag & VC_VARONLY) && vc.min_smpl_frac > 0.) {
479 extern int bcf_smpl_covered(const bcf1_t *b);
480 int n = bcf_smpl_covered(b);
481 if ((double)n / b->n_smpl < vc.min_smpl_frac) continue;
483 if (vc.n_sub) bcf_subsam(vc.n_sub, vc.sublist, b);
484 if (vc.flag & VC_FIX_PL) bcf_fix_pl(b);
485 is_indel = bcf_is_indel(b);
486 if ((vc.flag & VC_NO_INDEL) && is_indel) continue;
487 if ((vc.flag & VC_ACGT_ONLY) && !is_indel) {
489 if (b->ref[0] == 0 || b->ref[1] != 0) continue;
490 x = toupper(b->ref[0]);
491 if (x != 'A' && x != 'C' && x != 'G' && x != 'T') continue;
493 if (vc.bed && !bed_overlap(vc.bed, hin->ns[b->tid], b->pos, b->pos + strlen(b->ref))) continue;
495 int l = strlen(b->ref);
496 l = b->pos + (l > 0? l : 1);
497 if (b->tid != tid || b->pos >= end) break;
498 if (!(l > begin && end > b->pos)) continue;
501 if ((vc.flag & VC_QCNT) && !is_indel) { // summarize the difference
502 int x = bcf_min_diff(b);
503 if (x > 255) x = 255;
504 if (x >= 0) ++qcnt[x];
506 if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
510 if (vc.trio_aux) // do trio calling
511 bcf_trio_call(vc.trio_aux, b, &cons_llr, &cons_gt);
512 else if (vc.flag & VC_PAIRCALL)
513 cons_llr = bcf_pair_call(b);
514 if (vc.flag & (VC_CALL|VC_ADJLD|VC_EM)) bcf_gl2pl(b);
515 if (vc.flag & VC_EM) bcf_em1(b, vc.n1, 0x1ff, em);
518 for (i = 0; i < 9; ++i) em[i] = -1.;
520 if (vc.flag & VC_CALL) { // call variants
522 int calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
523 if (n_processed % 100000 == 0) {
524 fprintf(pysamerr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
527 if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
528 if (vc.n_perm && vc.n1 > 0 && pr.p_chi2 < vc.min_perm_p) { // permutation test
531 for (i = 0; i < vc.n_perm; ++i) {
532 #ifdef BCF_PERM_LRT // LRT based permutation is much faster but less robust to artifacts
534 bcf_shuffle(b, seeds[i]);
535 bcf_em1(b, vc.n1, 1<<7, x);
536 if (x[7] < em[7]) ++n;
538 bcf_shuffle(b, seeds[i]);
539 bcf_p1_cal(b, 1, p1, &r);
540 if (pr.p_chi2 >= r.p_chi2) ++n;
545 if (calret >= 0) update_bcf1(b, p1, &pr, vc.pref, vc.flag, em, cons_llr, cons_gt);
546 } else if (vc.flag & VC_EM) update_bcf1(b, 0, 0, 0, vc.flag, em, cons_llr, cons_gt);
547 if (vc.flag & VC_ADJLD) { // compute LD
549 if ((r2 = bcf_pair_freq(blast, b, f)) >= 0) {
551 s.m = s.l = 0; s.s = 0;
552 if (*b->info) kputc(';', &s);
553 ksprintf(&s, "NEIR=%.3f;NEIF4=%.3f,%.3f,%.3f,%.3f", r2, f[0], f[1], f[2], f[3]);
554 bcf_append_info(b, s.s, s.l);
559 if (vc.flag & VC_ANNO_MAX) bcf_anno_max(b);
560 if (vc.flag & VC_NO_GENO) { // do not output GENO fields
563 b->l_str = b->fmt - b->str + 1;
564 } else bcf_fix_gt(b);
565 vcf_write(bout, hout, b);
567 if (vc.prior_file) free(vc.prior_file);
568 if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
569 if (hin != hout) bcf_hdr_destroy(hout);
570 bcf_hdr_destroy(hin);
571 bcf_destroy(b); bcf_destroy(blast);
572 vcf_close(bp); vcf_close(bout);
573 if (vc.fn_dict) free(vc.fn_dict);
574 if (vc.ploidy) free(vc.ploidy);
575 if (vc.trio_aux) free(vc.trio_aux);
578 for (i = 0; i < vc.n_sub; ++i) free(vc.subsam[i]);
579 free(vc.subsam); free(vc.sublist);
581 if (vc.bed) bed_destroy(vc.bed);
582 if (vc.flag & VC_QCNT)
583 for (c = 0; c < 256; ++c)
584 fprintf(pysamerr, "QT\t%d\t%lld\n", c, (long long)qcnt[c]);
585 if (seeds) free(seeds);
586 if (p1) bcf_p1_destroy(p1);