X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;ds=sidebyside;f=bcftools%2Fcall1.c;h=5e4fc9fbc267928260c678ef3ce62f942fcebc7b;hb=7943d9b7b1c159da840cbcfeb9cb277bb58474f2;hp=1eacd4f731f1e35866a024d024ffe24cf4d8b49b;hpb=dda27af89b70c52e075f6531d56c0bbccbc7246d;p=samtools.git diff --git a/bcftools/call1.c b/bcftools/call1.c index 1eacd4f..5e4fc9f 100644 --- a/bcftools/call1.c +++ b/bcftools/call1.c @@ -24,11 +24,14 @@ KSTREAM_INIT(gzFile, gzread, 16384) #define VC_ACGT_ONLY 512 #define VC_QCALL 1024 #define VC_CALL_GT 2048 +#define VC_ADJLD 4096 +#define VC_NO_INDEL 8192 +#define VC_FOLDED 16384 typedef struct { int flag, prior_type, n1; char *fn_list, *prior_file; - double theta, pref; + double theta, pref, indel_frac; } viewconf_t; khash_t(set64) *bcf_load_pos(const char *fn, bcf_hdr_t *_h) @@ -160,7 +163,8 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p kputs(b->alt, &s); kputc('\0', &s); kputc('\0', &s); kputs(b->info, &s); if (b->info[0]) kputc(';', &s); - ksprintf(&s, "AF1=%.3lf;AFE=%.3lf", 1.-pr->f_em, 1.-pr->f_exp); +// ksprintf(&s, "AF1=%.4lg;AFE=%.4lg;CI95=%.4lg,%.4lg", 1.-pr->f_em, 1.-pr->f_exp, pr->cil, pr->cih); + ksprintf(&s, "AF1=%.4lg;CI95=%.4lg,%.4lg", 1.-pr->f_em, pr->cil, pr->cih); ksprintf(&s, ";DP4=%d,%d,%d,%d;MQ=%d", a.d[0], a.d[1], a.d[2], a.d[3], a.mq); if (a.is_tested) { if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%lg,%lg,%lg,%lg", pr->pc[0], pr->pc[1], pr->pc[2], pr->pc[3]); @@ -193,11 +197,14 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p return is_var; } +double bcf_ld_freq(const bcf1_t *b0, const bcf1_t *b1, double f[4]); + int bcfview(int argc, char *argv[]) { extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b); + extern void bcf_p1_indel_prior(bcf_p1aux_t *ma, double x); bcf_t *bp, *bout = 0; - bcf1_t *b; + bcf1_t *b, *blast; int c; uint64_t n_processed = 0; viewconf_t vc; @@ -209,9 +216,10 @@ int bcfview(int argc, char *argv[]) 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:cHAGvbSuP:t:p:Qg")) >= 0) { + vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5; vc.indel_frac = -1.; + while ((c = getopt(argc, argv, "fN1:l:cHAGvbSuP:t:p:QgLi:I")) >= 0) { switch (c) { + case 'f': vc.flag |= VC_FOLDED; break; case '1': vc.n1 = atoi(optarg); break; case 'l': vc.fn_list = strdup(optarg); break; case 'N': vc.flag |= VC_ACGT_ONLY; break; @@ -224,9 +232,12 @@ int bcfview(int argc, char *argv[]) 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 't': vc.theta = atof(optarg); break; case 'p': vc.pref = atof(optarg); break; + case 'i': vc.indel_frac = atof(optarg); break; case 'Q': vc.flag |= VC_QCALL; break; + case 'L': vc.flag |= VC_ADJLD; 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; @@ -249,9 +260,13 @@ int bcfview(int argc, char *argv[]) fprintf(stderr, " -H perform Hardy-Weinberg test (slower)\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, " -L calculate LD for adjacent sites\n"); + fprintf(stderr, " -I skip indels\n"); + fprintf(stderr, " -f reference-free variant calling\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 mutation rate [%.4lg]\n", vc.theta); + fprintf(stderr, " -t FLOAT scaled substitution mutation rate [%.4lg]\n", vc.theta); + fprintf(stderr, " -i FLOAT indel-to-substitution ratio [%.4lg]\n", vc.indel_frac); fprintf(stderr, " -p FLOAT variant if P(ref|D) 0.) bcf_p1_indel_prior(p1, vc.indel_frac); // otherwise use the default indel_frac + if (vc.flag & VC_FOLDED) bcf_p1_set_folded(p1); } if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h); if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) { @@ -300,7 +318,9 @@ int bcfview(int argc, char *argv[]) } } while (vcf_read(bp, h, b) > 0) { - if (vc.flag & VC_ACGT_ONLY) { + int 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; if (b->ref[0] == 0 || b->ref[1] != 0) continue; x = toupper(b->ref[0]); @@ -324,9 +344,9 @@ int bcfview(int argc, char *argv[]) bcf_2qcall(h, b); continue; } + if (vc.flag & (VC_CALL|VC_ADJLD)) bcf_gl2pl(b); if (vc.flag & VC_CALL) { // call variants bcf_p1rst_t pr; - 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) { @@ -336,6 +356,18 @@ int bcfview(int argc, char *argv[]) if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue; update_bcf1(h->n_smpl, b, p1, &pr, vc.pref, vc.flag); } + if (vc.flag & VC_ADJLD) { // compute LD + double f[4], r2; + if ((r2 = bcf_ld_freq(blast, b, f)) >= 0) { + kstring_t s; + s.m = s.l = 0; s.s = 0; + if (*b->info) kputc(';', &s); + ksprintf(&s, "NEIR=%.3lf;NEIF=%.3lf,%.3lf", r2, f[0]+f[2], f[0]+f[1]); + bcf_append_info(b, s.s, s.l); + free(s.s); + } + bcf_cpy(blast, b); + } if (vc.flag & VC_NO_GENO) { // do not output GENO fields b->n_gi = 0; b->fmt[0] = '\0'; @@ -345,11 +377,10 @@ int bcfview(int argc, char *argv[]) if (vc.prior_file) free(vc.prior_file); if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1); bcf_hdr_destroy(h); - bcf_destroy(b); + 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 (p1) bcf_p1_destroy(p1); return 0; } -