#define VC_VARONLY 16
#define VC_VCFIN 32
#define VC_UNCOMP 64
-#define VC_HWE 128
#define VC_KEEPALT 256
#define VC_ACGT_ONLY 512
#define VC_QCALL 1024
#define VC_ANNO_MAX 16384
typedef struct {
- int flag, prior_type, n1;
- char *fn_list, *prior_file;
+ int flag, prior_type, n1, n_sub, *sublist;
+ char *fn_list, *prior_file, **subsam, *fn_dict;
+ uint8_t *ploidy;
double theta, pref, indel_frac;
} viewconf_t;
return hash;
}
-static double test_hwe(const double g[3])
-{
- extern double kf_gammaq(double p, double x);
- double fexp, chi2, f[3], n;
- int i;
- n = g[0] + g[1] + g[2];
- fexp = (2. * g[2] + g[1]) / (2. * n);
- if (fexp > 1. - 1e-10) fexp = 1. - 1e-10;
- if (fexp < 1e-10) fexp = 1e-10;
- f[0] = n * (1. - fexp) * (1. - fexp);
- f[1] = n * 2. * fexp * (1. - fexp);
- f[2] = n * fexp * fexp;
- for (i = 0, chi2 = 0.; i < 3; ++i)
- chi2 += (g[i] - f[i]) * (g[i] - f[i]) / f[i];
- return kf_gammaq(.5, chi2 / 2.);
-}
-
typedef struct {
double p[4];
int mq, depth, is_tested, d[4];
{
kstring_t s;
int is_var = (pr->p_ref < pref);
- double p_hwe, r = is_var? pr->p_ref : pr->p_var, fq;
+ double r = is_var? pr->p_ref : pr->p_var, fq;
anno16_t a;
- p_hwe = pr->g[0] >= 0.? test_hwe(pr->g) : 1.0; // only do HWE g[] is calculated
test16(b, &a);
rm_info(b, "I16=");
if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%g,%g,%g,%g", pr->pc[0], pr->pc[1], pr->pc[2], pr->pc[3]);
ksprintf(&s, ";PV4=%.2g,%.2g,%.2g,%.2g", a.p[0], a.p[1], a.p[2], a.p[3]);
}
- if (pr->g[0] >= 0. && p_hwe <= .2)
- ksprintf(&s, ";GC=%.2f,%.2f,%.2f;HWE=%.3f", pr->g[2], pr->g[1], pr->g[0], p_hwe);
kputc('\0', &s);
kputs(b->fmt, &s); kputc('\0', &s);
free(b->str);
return is_var;
}
+static char **read_samples(const char *fn, int *_n)
+{
+ gzFile fp;
+ kstream_t *ks;
+ kstring_t s;
+ int dret, n = 0, max = 0;
+ char **sam = 0;
+ *_n = 0;
+ s.l = s.m = 0; s.s = 0;
+ fp = gzopen(fn, "r");
+ if (fp == 0) return 0; // fail to open file
+ ks = ks_init(fp);
+ while (ks_getuntil(ks, 0, &s, &dret) >= 0) {
+ int l;
+ if (max == n) {
+ max = max? max<<1 : 4;
+ sam = realloc(sam, sizeof(void*)*max);
+ }
+ l = s.l;
+ sam[n] = malloc(s.l + 2);
+ strcpy(sam[n], s.s);
+ sam[n][l+1] = 2; // by default, diploid
+ if (dret != '\n') {
+ if (ks_getuntil(ks, 0, &s, &dret) >= 0) { // read ploidy, 1 or 2
+ int x = (int)s.s[0] - '0';
+ if (x == 1 || x == 2) sam[n][l+1] = x;
+ else fprintf(stderr, "(%s) ploidy can only be 1 or 2; assume diploid\n", __func__);
+ }
+ if (dret != '\n') ks_getuntil(ks, '\n', &s, &dret);
+ }
+ ++n;
+ }
+ ks_destroy(ks);
+ gzclose(fp);
+ free(s.s);
+ *_n = n;
+ return sam;
+}
+
static void write_header(bcf_hdr_t *h)
{
kstring_t str;
if (!strstr(str.s, "##INFO=<ID=PV4,"))
kputs("##INFO=<ID=PV4,Number=4,Type=Float,Description=\"P-values for strand bias, baseQ bias, mapQ bias and tail distance bias\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=INDEL,"))
- kputs("##INFO=<ID=INDEL,Number=0,Type=Flag,Descriptin=\"Indicates that the variant is an INDEL.\">\n", &str);
+ kputs("##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=GT,"))
kputs("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=GQ,"))
uint64_t n_processed = 0;
viewconf_t vc;
bcf_p1aux_t *p1 = 0;
- bcf_hdr_t *h;
+ bcf_hdr_t *hin, *hout;
int tid, begin, end;
char moder[4], modew[4];
khash_t(set64) *hash = 0;
tid = begin = end = -1;
memset(&vc, 0, sizeof(viewconf_t));
vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5; vc.indel_frac = -1.;
- while ((c = getopt(argc, argv, "N1:l:cHAGvbSuP:t:p:QgLi:IM")) >= 0) {
+ while ((c = getopt(argc, argv, "N1:l:cHAGvbSuP:t:p:QgLi:IMs:D:")) >= 0) {
switch (c) {
case '1': vc.n1 = atoi(optarg); break;
case 'l': vc.fn_list = strdup(optarg); break;
+ case 'D': vc.fn_dict = strdup(optarg); break;
case 'N': vc.flag |= VC_ACGT_ONLY; break;
case 'G': vc.flag |= VC_NO_GENO; break;
case 'A': vc.flag |= VC_KEEPALT; break;
case 'c': vc.flag |= VC_CALL; break;
case 'v': vc.flag |= VC_VARONLY | VC_CALL; break;
case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
- case 'H': vc.flag |= VC_HWE; break;
case 'g': vc.flag |= VC_CALL_GT | VC_CALL; break;
case 'I': vc.flag |= VC_NO_INDEL; break;
case 'M': vc.flag |= VC_ANNO_MAX; break;
case 'i': vc.indel_frac = atof(optarg); break;
case 'Q': vc.flag |= VC_QCALL; break;
case 'L': vc.flag |= VC_ADJLD; break;
+ case 's': vc.subsam = read_samples(optarg, &vc.n_sub);
+ vc.ploidy = calloc(vc.n_sub + 1, 1);
+ for (tid = 0; tid < vc.n_sub; ++tid) vc.ploidy[tid] = vc.subsam[tid][strlen(vc.subsam[tid]) + 1];
+ tid = -1;
+ break;
case 'P':
if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
fprintf(stderr, " -Q output the QCALL likelihood format\n");
fprintf(stderr, " -L calculate LD for adjacent sites\n");
fprintf(stderr, " -I skip indels\n");
+ fprintf(stderr, " -D FILE sequence dictionary for VCF->BCF conversion [null]\n");
fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
fprintf(stderr, " -l FILE list of sites to output [all sites]\n");
fprintf(stderr, " -t FLOAT scaled substitution mutation rate [%.4g]\n", vc.theta);
fprintf(stderr, " -i FLOAT indel-to-substitution ratio [%.4g]\n", vc.indel_frac);
fprintf(stderr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3g]\n", vc.pref);
fprintf(stderr, " -P STR type of prior: full, cond2, flat [full]\n");
+ fprintf(stderr, " -s FILE list of samples to use [all samples]\n");
fprintf(stderr, "\n");
return 1;
}
+ if ((vc.flag & VC_VCFIN) && (vc.flag & VC_BCFOUT) && vc.fn_dict == 0) {
+ fprintf(stderr, "[%s] For VCF->BCF conversion please specify the sequence dictionary with -D\n", __func__);
+ return 1;
+ }
b = calloc(1, sizeof(bcf1_t));
blast = calloc(1, sizeof(bcf1_t));
strcpy(moder, "r");
if (vc.flag & VC_BCFOUT) strcat(modew, "b");
if (vc.flag & VC_UNCOMP) strcat(modew, "u");
bp = vcf_open(argv[optind], moder);
- h = vcf_hdr_read(bp);
+ hin = hout = vcf_hdr_read(bp);
+ if (vc.fn_dict && (vc.flag & VC_VCFIN))
+ vcf_dictread(bp, hin, vc.fn_dict);
bout = vcf_open("-", modew);
if (!(vc.flag & VC_QCALL)) {
- if (vc.flag & VC_CALL) write_header(h);
- vcf_hdr_write(bout, h);
+ if (vc.n_sub) {
+ vc.sublist = calloc(vc.n_sub, sizeof(int));
+ hout = bcf_hdr_subsam(hin, vc.n_sub, vc.subsam, vc.sublist);
+ }
+ if (vc.flag & VC_CALL) write_header(hout);
+ vcf_hdr_write(bout, hout);
}
if (vc.flag & VC_CALL) {
- p1 = bcf_p1_init(h->n_smpl);
+ p1 = bcf_p1_init(hout->n_smpl, vc.ploidy);
if (vc.prior_file) {
if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
fprintf(stderr, "[%s] fail to read the prior AFS.\n", __func__);
}
if (vc.indel_frac > 0.) bcf_p1_indel_prior(p1, vc.indel_frac); // otherwise use the default indel_frac
}
- if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h);
+ if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, hin);
if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
- void *str2id = bcf_build_refhash(h);
+ void *str2id = bcf_build_refhash(hout);
if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
bcf_idx_t *idx;
idx = bcf_idx_load(argv[optind]);
}
}
}
- while (vcf_read(bp, h, b) > 0) {
- int is_indel = bcf_is_indel(b);
+ while (vcf_read(bp, hin, b) > 0) {
+ int is_indel;
+ if (vc.n_sub) bcf_subsam(vc.n_sub, vc.sublist, b);
+ is_indel = bcf_is_indel(b);
if ((vc.flag & VC_NO_INDEL) && is_indel) continue;
if ((vc.flag & VC_ACGT_ONLY) && !is_indel) {
int x;
}
++n_processed;
if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
- bcf_2qcall(h, b);
+ bcf_2qcall(hout, b);
continue;
}
if (vc.flag & (VC_CALL|VC_ADJLD)) bcf_gl2pl(b);
if (vc.flag & VC_CALL) { // call variants
bcf_p1rst_t pr;
bcf_p1_cal(b, p1, &pr); // pr.g[3] is not calculated here
- if (vc.flag&VC_HWE) bcf_p1_cal_g3(p1, pr.g);
if (n_processed % 100000 == 0) {
fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
bcf_p1_dump_afs(p1);
}
if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
- update_bcf1(h->n_smpl, b, p1, &pr, vc.pref, vc.flag);
+ update_bcf1(hout->n_smpl, b, p1, &pr, vc.pref, vc.flag);
}
if (vc.flag & VC_ADJLD) { // compute LD
double f[4], r2;
b->fmt[0] = '\0';
b->l_str = b->fmt - b->str + 1;
} else bcf_fix_gt(b);
- vcf_write(bout, h, b);
+ vcf_write(bout, hout, b);
}
if (vc.prior_file) free(vc.prior_file);
if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
- bcf_hdr_destroy(h);
+ if (hin != hout) bcf_hdr_destroy(hout);
+ bcf_hdr_destroy(hin);
bcf_destroy(b); bcf_destroy(blast);
vcf_close(bp); vcf_close(bout);
if (hash) kh_destroy(set64, hash);
if (vc.fn_list) free(vc.fn_list);
+ if (vc.fn_dict) free(vc.fn_dict);
+ if (vc.ploidy) free(vc.ploidy);
+ if (vc.n_sub) {
+ int i;
+ for (i = 0; i < vc.n_sub; ++i) free(vc.subsam[i]);
+ free(vc.subsam); free(vc.sublist);
+ }
if (p1) bcf_p1_destroy(p1);
return 0;
}
projects / samtools.git / blobdiff