#include "kstring.h"
#include "time.h"
+#ifdef _WIN32
+#define srand48(x) srand(x)
+#define lrand48() rand()
+#endif
+
#include "kseq.h"
KSTREAM_INIT(gzFile, gzread, 16384)
#define VC_ANNO_MAX 16384
#define VC_FIX_PL 32768
#define VC_EM 0x10000
+#define VC_PAIRCALL 0x20000
+#define VC_QCNT 0x40000
+#define VC_INDEL_ONLY 0x80000
typedef struct {
int flag, prior_type, n1, n_sub, *sublist, n_perm;
+ uint32_t *trio_aux;
char *prior_file, **subsam, *fn_dict;
uint8_t *ploidy;
- double theta, pref, indel_frac, min_perm_p, min_smpl_frac, min_lrt;
+ double theta, pref, indel_frac, min_perm_p, min_smpl_frac, min_lrt, min_ma_lrt;
void *bed;
} viewconf_t;
void bed_destroy(void *_h);
int bed_overlap(const void *_h, const char *chr, int beg, int end);
-typedef struct {
- double p[4];
- int mq, depth, is_tested, d[4];
-} anno16_t;
-
static double ttest(int n1, int n2, int a[4])
{
extern double kf_betai(double a, double b, double x);
return 0;
}
-static int test16(bcf1_t *b, anno16_t *a)
+int test16(bcf1_t *b, anno16_t *a)
{
char *p;
int i, anno[16];
return test16_core(anno, a);
}
-static void rm_info(bcf1_t *b, const char *key)
-{
- char *p, *q;
- if ((p = strstr(b->info, key)) == 0) return;
- for (q = p; *q && *q != ';'; ++q);
- if (p > b->info && *(p-1) == ';') --p;
- memmove(p, q, b->l_str - (q - b->str));
- b->l_str -= q - p;
- bcf_sync(b);
-}
-
-static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag, double em[9])
+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)
{
kstring_t s;
int has_I16, is_var;
anno16_t a;
has_I16 = test16(b, &a) >= 0? 1 : 0;
- rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
+ //rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
memset(&s, 0, sizeof(kstring_t));
kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
if (em[4] >= 0 && em[4] <= 0.05) ksprintf(&s, ";G3=%.4g,%.4g,%.4g;HWE=%.3g", em[3], em[2], em[1], em[4]);
if (em[5] >= 0 && em[6] >= 0) ksprintf(&s, ";AF2=%.4g,%.4g", 1 - em[5], 1 - em[6]);
if (em[7] >= 0) ksprintf(&s, ";LRT=%.3g", em[7]);
+ if (em[8] >= 0) ksprintf(&s, ";LRT2=%.3g", em[8]);
+ }
+ if (cons_llr > 0) {
+ ksprintf(&s, ";CLR=%d", cons_llr);
+ if (cons_gt > 0)
+ ksprintf(&s, ";UGT=%c%c%c;CGT=%c%c%c", cons_gt&0xff, cons_gt>>8&0xff, cons_gt>>16&0xff,
+ cons_gt>>32&0xff, cons_gt>>40&0xff, cons_gt>>48&0xff);
}
if (pr == 0) { // if pr is unset, return
kputc('\0', &s); kputs(b->fmt, &s); kputc('\0', &s);
is_var = (pr->p_ref < pref);
r = is_var? pr->p_ref : pr->p_var;
- ksprintf(&s, ";CI95=%.4g,%.4g", pr->cil, pr->cih); // FIXME: when EM is not used, ";" should be omitted!
+// ksprintf(&s, ";CI95=%.4g,%.4g", pr->cil, pr->cih); // FIXME: when EM is not used, ";" should be omitted!
+ ksprintf(&s, ";AC1=%d", pr->ac);
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);
fq = pr->p_ref_folded < 0.5? -4.343 * log(pr->p_ref_folded) : 4.343 * log(pr->p_var_folded);
if (fq < -999) fq = -999;
if (q[i] > 255) q[i] = 255;
}
if (pr->perm_rank >= 0) ksprintf(&s, ";PR=%d", pr->perm_rank);
+ // ksprintf(&s, ";LRT3=%.3g", pr->lrt);
ksprintf(&s, ";PCHI2=%.3g;PC2=%d,%d", q[1], q[2], pr->p_chi2);
}
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]);
kputc('\0', &s);
+ rm_info(&s, "QS=");
+ rm_info(&s, "I16=");
kputs(b->fmt, &s); kputc('\0', &s);
free(b->str);
b->m_str = s.m; b->l_str = s.l; b->str = s.s;
if (!strstr(str.s, "##INFO=<ID=FQ,"))
kputs("##INFO=<ID=FQ,Number=1,Type=Float,Description=\"Phred probability of all samples being the same\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=AF1,"))
- kputs("##INFO=<ID=AF1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the site allele frequency of the first ALT allele\">\n", &str);
+ kputs("##INFO=<ID=AF1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele frequency (assuming HWE)\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=AC1,"))
+ kputs("##INFO=<ID=AC1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the first ALT allele count (no HWE assumption)\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=AN,"))
+ kputs("##INFO=<ID=AN,Number=1,Type=Integer,Description=\"Total number of alleles in called genotypes\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=IS,"))
+ kputs("##INFO=<ID=IS,Number=2,Type=Float,Description=\"Maximum number of reads supporting an indel and fraction of indel reads\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=AC,"))
+ kputs("##INFO=<ID=AC,Number=A,Type=Integer,Description=\"Allele count in genotypes for each ALT allele, in the same order as listed\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=G3,"))
kputs("##INFO=<ID=G3,Number=3,Type=Float,Description=\"ML estimate of genotype frequencies\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=HWE,"))
kputs("##INFO=<ID=HWE,Number=1,Type=Float,Description=\"Chi^2 based HWE test P-value based on G3\">\n", &str);
- if (!strstr(str.s, "##INFO=<ID=CI95,"))
- 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);
+ if (!strstr(str.s, "##INFO=<ID=CLR,"))
+ kputs("##INFO=<ID=CLR,Number=1,Type=Integer,Description=\"Log ratio of genotype likelihoods with and without the constraint\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=UGT,"))
+ kputs("##INFO=<ID=UGT,Number=1,Type=String,Description=\"The most probable unconstrained genotype configuration in the trio\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=CGT,"))
+ kputs("##INFO=<ID=CGT,Number=1,Type=String,Description=\"The most probable constrained genotype configuration in the trio\">\n", &str);
+// if (!strstr(str.s, "##INFO=<ID=CI95,"))
+// 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);
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=QCHI2,Number=1,Type=Integer,Description=\"Phred scaled PCHI2.\">\n", &str);
if (!strstr(str.s, "##INFO=<ID=RP,"))
kputs("##INFO=<ID=PR,Number=1,Type=Integer,Description=\"# permutations yielding a smaller PCHI2.\">\n", &str);
+ if (!strstr(str.s, "##INFO=<ID=VDB,"))
+ kputs("##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=GT,"))
kputs("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=GQ,"))
kputs("##FORMAT=<ID=GL,Number=3,Type=Float,Description=\"Likelihoods for RR,RA,AA genotypes (R=ref,A=alt)\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=DP,"))
kputs("##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"# high-quality bases\">\n", &str);
+ if (!strstr(str.s, "##FORMAT=<ID=DV,"))
+ kputs("##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"# high-quality non-reference bases\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=SP,"))
kputs("##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">\n", &str);
if (!strstr(str.s, "##FORMAT=<ID=PL,"))
- 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);
+ kputs("##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">\n", &str);
h->l_txt = str.l + 1; h->txt = str.s;
}
extern int bcf_fix_gt(bcf1_t *b);
extern int bcf_anno_max(bcf1_t *b);
extern int bcf_shuffle(bcf1_t *b, int seed);
+ extern uint32_t *bcf_trio_prep(int is_x, int is_son);
+ extern int bcf_trio_call(uint32_t *prep, const bcf1_t *b, int *llr, int64_t *gt);
+ extern int bcf_pair_call(const bcf1_t *b);
+ extern int bcf_min_diff(const bcf1_t *b);
+ extern int bcf_p1_get_M(bcf_p1aux_t *b);
+
+ extern gzFile bcf_p1_fp_lk;
+
bcf_t *bp, *bout = 0;
bcf1_t *b, *blast;
int c, *seeds = 0;
- uint64_t n_processed = 0;
+ uint64_t n_processed = 0, qcnt[256];
viewconf_t vc;
bcf_p1aux_t *p1 = 0;
bcf_hdr_t *hin, *hout;
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.; vc.n_perm = 0; vc.min_perm_p = 0.01; vc.min_smpl_frac = 0; vc.min_lrt = 1;
- while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:")) >= 0) {
+ 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; vc.min_ma_lrt = -1;
+ memset(qcnt, 0, 8 * 256);
+ while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:T:Ywm:K:")) >= 0) {
switch (c) {
case '1': vc.n1 = atoi(optarg); break;
- case 'l': vc.bed = bed_read(optarg); break;
+ case 'l': vc.bed = bed_read(optarg); if (!vc.bed) { fprintf(stderr,"Could not read \"%s\"\n", optarg); return 1; } break;
case 'D': vc.fn_dict = strdup(optarg); break;
case 'F': vc.flag |= VC_FIX_PL; break;
case 'N': vc.flag |= VC_ACGT_ONLY; break;
case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
case 'g': vc.flag |= VC_CALL_GT | VC_CALL; break;
case 'I': vc.flag |= VC_NO_INDEL; break;
+ case 'w': vc.flag |= VC_INDEL_ONLY; break;
case 'M': vc.flag |= VC_ANNO_MAX; break;
+ case 'Y': vc.flag |= VC_QCNT; break;
+ case 'm': vc.min_ma_lrt = atof(optarg); break;
case 't': vc.theta = atof(optarg); break;
case 'p': vc.pref = atof(optarg); break;
case 'i': vc.indel_frac = atof(optarg); break;
case 'C': vc.min_lrt = atof(optarg); break;
case 'X': vc.min_perm_p = atof(optarg); break;
case 'd': vc.min_smpl_frac = atof(optarg); break;
+ case 'K': bcf_p1_fp_lk = gzopen(optarg, "w"); 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 'T':
+ if (strcmp(optarg, "trioauto") == 0) vc.trio_aux = bcf_trio_prep(0, 0);
+ else if (strcmp(optarg, "trioxd") == 0) vc.trio_aux = bcf_trio_prep(1, 0);
+ else if (strcmp(optarg, "trioxs") == 0) vc.trio_aux = bcf_trio_prep(1, 1);
+ else if (strcmp(optarg, "pair") == 0) vc.flag |= VC_PAIRCALL;
+ else {
+ fprintf(stderr, "[%s] Option '-T' can only take value trioauto, trioxd or trioxs.\n", __func__);
+ return 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, " -g call genotypes at variant sites (force -c)\n");
fprintf(stderr, " -i FLOAT indel-to-substitution ratio [%.4g]\n", vc.indel_frac);
fprintf(stderr, " -I skip indels\n");
+ fprintf(stderr, " -m FLOAT alternative model for multiallelic and rare-variant calling, include if P(chi^2)>=FLOAT\n");
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, " -t FLOAT scaled substitution mutation rate [%.4g]\n", vc.theta);
+ fprintf(stderr, " -T STR constrained calling; STR can be: pair, trioauto, trioxd and trioxs (see manual) [null]\n");
fprintf(stderr, " -v output potential variant sites only (force -c)\n");
fprintf(stderr, "\nContrast calling and association test options:\n\n");
fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
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);
+ write_header(hout); // always print the header
vcf_hdr_write(bout, hout);
}
if (vc.flag & VC_CALL) {
}
}
}
+ if (bcf_p1_fp_lk && p1) {
+ int32_t M = bcf_p1_get_M(p1);
+ gzwrite(bcf_p1_fp_lk, &M, 4);
+ }
while (vcf_read(bp, hin, b) > 0) {
- int is_indel;
- double em[9];
+ int is_indel, cons_llr = -1;
+ int64_t cons_gt = -1;
+ double em[10];
if ((vc.flag & VC_VARONLY) && strcmp(b->alt, "X") == 0) continue;
if ((vc.flag & VC_VARONLY) && vc.min_smpl_frac > 0.) {
extern int bcf_smpl_covered(const bcf1_t *b);
if (vc.flag & VC_FIX_PL) bcf_fix_pl(b);
is_indel = bcf_is_indel(b);
if ((vc.flag & VC_NO_INDEL) && is_indel) continue;
+ if ((vc.flag & VC_INDEL_ONLY) && !is_indel) continue;
if ((vc.flag & VC_ACGT_ONLY) && !is_indel) {
int x;
if (b->ref[0] == 0 || b->ref[1] != 0) continue;
if (!(l > begin && end > b->pos)) continue;
}
++n_processed;
+ if ((vc.flag & VC_QCNT) && !is_indel) { // summarize the difference
+ int x = bcf_min_diff(b);
+ if (x > 255) x = 255;
+ if (x >= 0) ++qcnt[x];
+ }
if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
bcf_2qcall(hout, b);
continue;
}
- if (vc.flag & VC_EM) bcf_em1(b, vc.n1, 0xff, em);
+ if (vc.trio_aux) // do trio calling
+ bcf_trio_call(vc.trio_aux, b, &cons_llr, &cons_gt);
+ else if (vc.flag & VC_PAIRCALL)
+ cons_llr = bcf_pair_call(b);
+ if (vc.flag & (VC_CALL|VC_ADJLD|VC_EM)) bcf_gl2pl(b);
+ if (vc.flag & VC_EM) bcf_em1(b, vc.n1, 0x1ff, em);
else {
int i;
for (i = 0; i < 9; ++i) em[i] = -1.;
}
- if (vc.flag & (VC_CALL|VC_ADJLD)) bcf_gl2pl(b);
- if (vc.flag & VC_CALL) { // call variants
+ if ( !(vc.flag&VC_KEEPALT) && (vc.flag&VC_CALL) && vc.min_ma_lrt>=0 )
+ {
+ bcf_p1_set_ploidy(b, p1); // could be improved: do this per site to allow pseudo-autosomal regions
+ int gts = call_multiallelic_gt(b,p1,vc.min_ma_lrt);
+ if ( gts<=1 && vc.flag & VC_VARONLY ) continue;
+ }
+ else if (vc.flag & VC_CALL) { // call variants
bcf_p1rst_t pr;
- int calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
+ int calret;
+ gzwrite(bcf_p1_fp_lk, &b->tid, 4);
+ gzwrite(bcf_p1_fp_lk, &b->pos, 4);
+ gzwrite(bcf_p1_fp_lk, &em[0], sizeof(double));
+ calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
if (n_processed % 100000 == 0) {
fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
bcf_p1_dump_afs(p1);
int i, n = 0;
for (i = 0; i < vc.n_perm; ++i) {
#ifdef BCF_PERM_LRT // LRT based permutation is much faster but less robust to artifacts
- double x[9];
+ double x[10];
bcf_shuffle(b, seeds[i]);
bcf_em1(b, vc.n1, 1<<7, x);
if (x[7] < em[7]) ++n;
}
pr.perm_rank = n;
}
- if (calret >= 0) update_bcf1(b, p1, &pr, vc.pref, vc.flag, em);
- } else if (vc.flag & VC_EM) update_bcf1(b, 0, 0, 0, vc.flag, em);
+ if (calret >= 0) update_bcf1(b, p1, &pr, vc.pref, vc.flag, em, cons_llr, cons_gt);
+ } else if (vc.flag & VC_EM) update_bcf1(b, 0, 0, 0, vc.flag, em, cons_llr, cons_gt);
if (vc.flag & VC_ADJLD) { // compute LD
double f[4], r2;
if ((r2 = bcf_pair_freq(blast, b, f)) >= 0) {
} else bcf_fix_gt(b);
vcf_write(bout, hout, b);
}
+
+ if (bcf_p1_fp_lk) gzclose(bcf_p1_fp_lk);
if (vc.prior_file) free(vc.prior_file);
if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
if (hin != hout) bcf_hdr_destroy(hout);
vcf_close(bp); vcf_close(bout);
if (vc.fn_dict) free(vc.fn_dict);
if (vc.ploidy) free(vc.ploidy);
+ if (vc.trio_aux) free(vc.trio_aux);
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 (vc.bed) bed_destroy(vc.bed);
+ if (vc.flag & VC_QCNT)
+ for (c = 0; c < 256; ++c)
+ fprintf(stderr, "QT\t%d\t%lld\n", c, (long long)qcnt[c]);
if (seeds) free(seeds);
if (p1) bcf_p1_destroy(p1);
return 0;
projects / samtools.git / blobdiff