* changed 3.434 to 4.343 (typo!)

[samtools.git] / bcftools / call1.c

diff --git a/bcftools/call1.c b/bcftools/call1.c
index 75483e921a9ee10c45bfe6e4acf8e2ad3261f42c..c5d466e12b164d3df39b093b1cd07377207e2285 100644 (file)
--- a/bcftools/call1.c
+++ b/bcftools/call1.c
@@ -20,9 +20,11 @@ KSTREAM_INIT(gzFile, gzread, 16384)
 #define VC_VARONLY 16
 #define VC_VCFIN   32
 #define VC_UNCOMP  64
+#define VC_HWE     128
+#define VC_KEEPALT 256
 
 typedef struct {
-       int flag, prior_type;
+       int flag, prior_type, n1;
        char *fn_list, *prior_file;
        double theta, pref;
 } viewconf_t;
@@ -147,29 +149,33 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p
        double p_hwe, r = is_var? pr->p_ref : 1. - pr->p_ref;
        anno16_t a;
 
-       p_hwe = test_hwe(pr->g);
+       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=");
 
        memset(&s, 0, sizeof(kstring_t));
        kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
-       if (is_var) {
-               kputs(b->alt, &s);
-       }
-       kputc('\0', &s); kputc('\0', &s);
+       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, ";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) ksprintf(&s, ";PV4=%.2lg,%.2lg,%.2lg,%.2lg", a.p[0], a.p[1], a.p[2], a.p[3]);
-       if (p_hwe <= .2) ksprintf(&s, ";GC=%.2lf,%.2lf,%.2lf;HWE=%.3lf", pr->g[2], pr->g[1], pr->g[0], p_hwe);
+       if (a.is_tested) {
+               if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%.4lf,%.4lf,%.2lg,%.2lg", pr->pc[0], pr->pc[1], pr->pc[2], pr->pc[3]);
+               ksprintf(&s, ";PV4=%.2lg,%.2lg,%.2lg,%.2lg", a.p[0], a.p[1], a.p[2], a.p[3]);
+       }
+       if (pr->g[0] >= 0. && p_hwe <= .2)
+               ksprintf(&s, ";GC=%.2lf,%.2lf,%.2lf;HWE=%.3lf", pr->g[2], pr->g[1], pr->g[0], p_hwe);
        kputc('\0', &s);
        kputs(b->fmt, &s); kputc('\0', &s);
        free(b->str);
        b->m_str = s.m; b->l_str = s.l; b->str = s.s;
-       b->qual = r < 1e-100? 99 : -3.434 * log(r);
+       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);
+       else if (!(flag&VC_KEEPALT))
+               bcf_shrink_alt(n_smpl, b, pr->rank0 < 2? 2 : pr->rank0+1);
        return is_var;
 }
 
@@ -188,17 +194,20 @@ int bcfview(int argc, char *argv[])
 
        tid = begin = end = -1;
        memset(&vc, 0, sizeof(viewconf_t));
-       vc.prior_type = -1; vc.theta = 1e-3; vc.pref = 0.9;
-       while ((c = getopt(argc, argv, "l:cGvLbSuP:t:p:")) >= 0) {
+       vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5;
+       while ((c = getopt(argc, argv, "1:l:cHAGvLbSuP:t:p:")) >= 0) {
                switch (c) {
+               case '1': vc.n1 = atoi(optarg); break;
                case 'l': vc.fn_list = strdup(optarg); 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 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
+               case 'H': vc.flag |= VC_HWE; break;
                case 't': vc.theta = atof(optarg); break;
                case 'p': vc.pref = atof(optarg); break;
                case 'P':
@@ -216,9 +225,12 @@ int bcfview(int argc, char *argv[])
                fprintf(stderr, "         -b        output BCF instead of VCF\n");
                fprintf(stderr, "         -u        uncompressed BCF output\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, "         -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, "         -p FLOAT  variant if P(ref|D)<FLOAT [%.3lg]\n", vc.pref);
@@ -245,6 +257,7 @@ int bcfview(int argc, char *argv[])
                                return 1;
                        }
                } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
+               if (vc.n1 > 0) bcf_p1_set_n1(p1, vc.n1);
        }
        if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h);
        if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
@@ -272,10 +285,16 @@ int bcfview(int argc, char *argv[])
                        if (b->tid != tid || b->pos >= end) break;
                        if (!(l > begin && end > b->pos)) continue;
                }
+               ++n_processed;
                if (vc.flag & VC_CALL) {
                        bcf_p1rst_t pr;
-                       bcf_p1_cal(b, p1, &pr);
-                       if ((n_processed + 1) % 50000 == 0) bcf_p1_dump_afs(p1);
+                       bcf_gl2pl(h->n_smpl, 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) {
+                               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);
                }
@@ -284,7 +303,6 @@ int bcfview(int argc, char *argv[])
                        b->fmt[0] = '\0';
                }
                vcf_write(bout, h, b);
-               ++n_processed;
        }
        if (vc.prior_file) free(vc.prior_file);
        if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);