#include "kseq.h"
KSTREAM_INIT(gzFile, gzread, 16384)
-#define VC_NO_PL 1
#define VC_NO_GENO 2
#define VC_BCFOUT 4
#define VC_CALL 8
#define VC_KEEPALT 256
#define VC_ACGT_ONLY 512
#define VC_QCALL 1024
+#define VC_CALL_GT 2048
typedef struct {
int flag, prior_type, n1;
return test16_core(anno, a);
}
-static void rm_info(int n_smpl, bcf1_t *b, const char *key)
+static void rm_info(bcf1_t *b, const char *key)
{
char *p, *q;
if ((p = strstr(b->info, key)) == 0) return;
if (p > b->info && *(p-1) == ';') --p;
memmove(p, q, b->l_str - (q - b->str));
b->l_str -= q - p;
- bcf_sync(n_smpl, b);
+ bcf_sync(b);
}
static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag)
p_hwe = pr->g[0] >= 0.? test_hwe(pr->g) : 1.0; // only do HWE g[] is calculated
test16(b, &a);
- rm_info(n_smpl, b, "I16=");
+ rm_info(b, "I16=");
memset(&s, 0, sizeof(kstring_t));
kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
b->m_str = s.m; b->l_str = s.l; b->str = s.s;
b->qual = r < 1e-100? 99 : -4.343 * log(r);
if (b->qual > 99) b->qual = 99;
- bcf_sync(n_smpl, b);
- if (!is_var) bcf_shrink_alt(n_smpl, b, 1);
+ bcf_sync(b);
+ if (!is_var) bcf_shrink_alt(b, 1);
else if (!(flag&VC_KEEPALT))
- bcf_shrink_alt(n_smpl, b, pr->rank0 < 2? 2 : pr->rank0+1);
+ bcf_shrink_alt(b, pr->rank0 < 2? 2 : pr->rank0+1);
+ if (is_var && (flag&VC_CALL_GT)) { // call individual genotype
+ int i, x, old_n_gi = b->n_gi;
+ s.m = b->m_str; s.l = b->l_str - 1; s.s = b->str;
+ kputs(":GT:GQ", &s); kputc('\0', &s);
+ b->m_str = s.m; b->l_str = s.l; b->str = s.s;
+ bcf_sync(b);
+ for (i = 0; i < b->n_smpl; ++i) {
+ x = bcf_p1_call_gt(pa, pr->f_em, i);
+ ((uint8_t*)b->gi[old_n_gi].data)[i] = (x&3) == 0? 1<<3|1 : (x&3) == 1? 1 : 0;
+ ((uint8_t*)b->gi[old_n_gi+1].data)[i] = x>>2;
+ }
+ }
return is_var;
}
tid = begin = end = -1;
memset(&vc, 0, sizeof(viewconf_t));
vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5;
- while ((c = getopt(argc, argv, "N1:l:cHAGvLbSuP:t:p:Q")) >= 0) {
+ while ((c = getopt(argc, argv, "N1:l:cHAGvbSuP:t:p:Qg")) >= 0) {
switch (c) {
case '1': vc.n1 = atoi(optarg); break;
case 'l': vc.fn_list = strdup(optarg); break;
case 'N': vc.flag |= VC_ACGT_ONLY; break;
case 'G': vc.flag |= VC_NO_GENO; break;
- case 'L': vc.flag |= VC_NO_PL; break;
case 'A': vc.flag |= VC_KEEPALT; break;
case 'b': vc.flag |= VC_BCFOUT; break;
case 'S': vc.flag |= VC_VCFIN; break;
case 'c': vc.flag |= VC_CALL; break;
- case 'v': vc.flag |= VC_VARONLY; 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 't': vc.theta = atof(optarg); break;
case 'p': vc.pref = atof(optarg); break;
case 'Q': vc.flag |= VC_QCALL; break;
fprintf(stderr, "\n");
fprintf(stderr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
fprintf(stderr, "Options: -c SNP calling\n");
+ fprintf(stderr, " -v output potential variant sites only (force -c)\n");
+ fprintf(stderr, " -g call genotypes at variant sites (force -c)\n");
fprintf(stderr, " -b output BCF instead of VCF\n");
- fprintf(stderr, " -u uncompressed BCF output\n");
+ fprintf(stderr, " -u uncompressed BCF output (force -b)\n");
fprintf(stderr, " -S input is VCF\n");
fprintf(stderr, " -A keep all possible alternate alleles at variant sites\n");
fprintf(stderr, " -G suppress all individual genotype information\n");
- fprintf(stderr, " -L discard the PL genotype field\n");
fprintf(stderr, " -H perform Hardy-Weinberg test (slower)\n");
- fprintf(stderr, " -v output potential variant sites only\n");
fprintf(stderr, " -N skip sites where REF is not A/C/G/T\n");
fprintf(stderr, " -Q output the QCALL likelihood format\n");
fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
idx = bcf_idx_load(argv[optind]);
if (idx) {
uint64_t off;
- off = bcf_idx_query(idx, tid, begin, end);
+ off = bcf_idx_query(idx, tid, begin);
+ if (off == 0) {
+ fprintf(stderr, "[%s] no records in the query region.\n", __func__);
+ return 1; // FIXME: a lot of memory leaks...
+ }
bgzf_seek(bp->fp, off, SEEK_SET);
bcf_idx_destroy(idx);
}
}
if (vc.flag & VC_CALL) { // call variants
bcf_p1rst_t pr;
- bcf_gl2pl(h->n_smpl, b);
+ bcf_gl2pl(b);
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) {
projects / samtools.git / blobdiff