* convert BCF to QCALL input

[samtools.git] / bcftools / call1.c

diff --git a/bcftools/call1.c b/bcftools/call1.c
index 995d500eb69ad028a1b0ed2a75da5bde88d92bee..08d759d453e92fc907e03431ccb3120a012f7ebf 100644 (file)
--- a/bcftools/call1.c
+++ b/bcftools/call1.c
@@ -22,6 +22,8 @@ KSTREAM_INIT(gzFile, gzread, 16384)
 #define VC_UNCOMP  64
 #define VC_HWE     128
 #define VC_KEEPALT 256
+#define VC_ACGT_ONLY 512
+#define VC_QCALL   1024
 
 typedef struct {
        int flag, prior_type, n1;
@@ -161,7 +163,7 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p
        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) {
-               if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%.2lg,%.2lg,%.2lg,%.2lg", pr->pc[0], pr->pc[1], pr->pc[2], pr->pc[3]);
+               if (pr->pc[0] >= 0.) ksprintf(&s, ";PC4=%lg,%lg,%lg,%lg", 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)
@@ -170,7 +172,7 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p
        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);
@@ -181,6 +183,7 @@ static int update_bcf1(int n_smpl, bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p
 
 int bcfview(int argc, char *argv[])
 {
+       extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b);
        bcf_t *bp, *bout = 0;
        bcf1_t *b;
        int c;
@@ -194,11 +197,12 @@ 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 = 0.9;
-       while ((c = getopt(argc, argv, "1:l:cHAGvLbSuP:t:p:")) >= 0) {
+       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) {
                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;
@@ -210,6 +214,7 @@ int bcfview(int argc, char *argv[])
                case 'H': vc.flag |= VC_HWE; break;
                case 't': vc.theta = atof(optarg); break;
                case 'p': vc.pref = atof(optarg); break;
+               case 'Q': vc.flag |= VC_QCALL; 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;
@@ -230,6 +235,8 @@ int bcfview(int argc, char *argv[])
                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");
                fprintf(stderr, "         -l FILE   list of sites to output [all sites]\n");
                fprintf(stderr, "         -t FLOAT  scaled mutation rate [%.4lg]\n", vc.theta);
@@ -248,7 +255,7 @@ int bcfview(int argc, char *argv[])
        bp = vcf_open(argv[optind], moder);
        h = vcf_hdr_read(bp);
        bout = vcf_open("-", modew);
-       vcf_hdr_write(bout, h);
+       if (!(vc.flag & VC_QCALL)) vcf_hdr_write(bout, h);
        if (vc.flag & VC_CALL) {
                p1 = bcf_p1_init(h->n_smpl);
                if (vc.prior_file) {
@@ -257,7 +264,10 @@ 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.n1 > 0) {
+                       bcf_p1_set_n1(p1, vc.n1);
+                       bcf_p1_init_subprior(p1, vc.prior_type, vc.theta);
+               }
        }
        if (vc.fn_list) hash = bcf_load_pos(vc.fn_list, h);
        if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
@@ -274,10 +284,18 @@ int bcfview(int argc, char *argv[])
                }
        }
        while (vcf_read(bp, h, b) > 0) {
+               if (vc.flag & VC_ACGT_ONLY) {
+                       int x;
+                       if (b->ref[0] == 0 || b->ref[1] != 0) continue;
+                       x = toupper(b->ref[0]);
+                       if (x != 'A' && x != 'C' && x != 'G' && x != 'T') continue;
+               }
                if (hash) {
                        uint64_t x = (uint64_t)b->tid<<32 | b->pos;
                        khint_t k = kh_get(set64, hash, x);
+                       if (kh_size(hash) == 0) break;
                        if (k == kh_end(hash)) continue;
+                       kh_del(set64, hash, k);
                }
                if (tid >= 0) {
                        int l = strlen(b->ref);
@@ -285,21 +303,28 @@ int bcfview(int argc, char *argv[])
                        if (b->tid != tid || b->pos >= end) break;
                        if (!(l > begin && end > b->pos)) continue;
                }
-               if (vc.flag & VC_CALL) {
+               ++n_processed;
+               if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
+                       bcf_2qcall(h, b);
+                       continue;
+               }
+               if (vc.flag & VC_CALL) { // call variants
                        bcf_p1rst_t pr;
                        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 + 1) % 50000 == 0) bcf_p1_dump_afs(p1);
+                       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);
                }
-               if (vc.flag & VC_NO_GENO) {
+               if (vc.flag & VC_NO_GENO) { // do not output GENO fields
                        b->n_gi = 0;
                        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);