added the format column

[samtools.git] / bam_plcmd.c

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 285477050d19f489d7be34722846eca8e6407e14..34366332ecf895188c29b2a88e8a062d871ffe0d 100644 (file)
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -450,7 +450,8 @@ int bam_pileup(int argc, char *argv[])
  ***********/
 
 typedef struct {
-       int vcf, max_mq, min_mq, var_only;
+       int vcf, max_mq, min_mq, var_only, prior_type;
+       double theta;
        char *reg, *fn_pos;
        faidx_t *fai;
        kh_64_t *hash;
@@ -521,6 +522,10 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                kstring_t s;
                s.l = s.m = 0; s.s = 0;
                puts("##fileformat=VCFv4.0");
+               puts("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total read depth\">");
+               puts("##INFO=<ID=AF,Number=1,Type=Float,Description=\"Posterior non-reference allele frequency\">");
+               puts("##INFO=<ID=AFEM,Number=1,Type=Float,Description=\"Prior-free non-reference allele frequency by EM\">");
+               puts("##FILTER=<ID=Q13,Description=\"All min{baseQ,mapQ} below 13\">");
                kputs("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT", &s);
                for (i = 0; i < n; ++i) {
                        const char *p;
@@ -533,7 +538,10 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                free(s.s);
        }
        // mpileup
-       if (conf->vcf) ma = mc_init(n);
+       if (conf->vcf) {
+               ma = mc_init(n);
+               mc_init_prior(ma, conf->prior_type, conf->theta);
+       }
        ref_tid = -1; ref = 0;
        iter = bam_mplp_init(n, mplp_func, (void**)data);
        while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
@@ -549,20 +557,24 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                        ref_tid = tid;
                }
                if (conf->vcf) {
-                       double f0, f, pref = -1.0; // the reference allele frequency
+                       double f0, f, fpost, pref = -1.0; // the reference allele frequency
                        int j, _ref, _alt, _ref0, depth, rms_q, _ref0b, is_var = 0, qref = 0;
                        uint64_t sqr_sum;
                        _ref0 = _ref0b = (ref && pos < ref_len)? ref[pos] : 'N';
                        _ref0 = bam_nt16_nt4_table[bam_nt16_table[_ref0]];
-                       f = f0 = mc_freq0(_ref0, n_plp, plp, ma, &_ref, &_alt);
+                       f = f0 = fpost = mc_freq0(_ref0, n_plp, plp, ma, &_ref, &_alt);
                        if (f >= 0.0) {
                                double q, flast = f;
-                               for (j = 0; j < 10; ++j){
+                               for (j = 0; j < 10; ++j) {
                                        f = mc_freq_iter(flast, ma);
+//                                     printf("%lg->%lg\n", flast, f);
                                        if (fabs(f - flast) < 1e-3) break;
                                        flast = f;
                                }
                                pref = mc_ref_prob(ma);
+                               fpost = mc_freq_post(ma);
+                               if (1. - f < 1e-4) f = 1.;
+                               if (1. - fpost < 1e-4) fpost = 1.;
                                is_var = (pref < .5);
                                q = is_var? pref : 1. - pref;
                                if (q < 1e-308) q = 1e-308;
@@ -587,13 +599,16 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                                }
                        }
                        rms_q = (int)(sqrt((double)sqr_sum / depth) + .499);
-                       printf("DP=%d;MQ=%d;AF=%.3lg", depth, rms_q, f0<0.?0.:1.-f);
+                       printf("DP=%d;MQ=%d", depth, rms_q);
+                       if (fpost >= 0. && fpost <= 1.) printf(";AF=%.3lg", 1. - fpost);
+                       if (f >= 0. && f <= 1.) printf(";AFEM=%.3lg", 1. - f);
                        printf("\tGT:GQ:DP");
-                       // output genotype information; FIXME: to be implmented...
-                       for (i = 0; i < n; ++i) {
-                               int x = mc_call_gt(ma, f, i);
-                               printf("\t%c/%c:%d:%d", "10"[((x&3)==2)], "10"[((x&3)>0)], x>>2, n_plp[i]);
-                       }
+                       if (fpost >= 0. && fpost <= 1.) {
+                               for (i = 0; i < n; ++i) {
+                                       int x = mc_call_gt(ma, fpost, i);
+                                       printf("\t%c/%c:%d:%d", "10"[((x&3)==2)], "10"[((x&3)>0)], x>>2, n_plp[i]);
+                               }
+                       } else for (i = 0; i < n; ++i) printf("\t./.:0:0");
                        putchar('\n');
                } else {
                        printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
@@ -640,8 +655,12 @@ int bam_mpileup(int argc, char *argv[])
        mplp_conf_t mplp;
        memset(&mplp, 0, sizeof(mplp_conf_t));
        mplp.max_mq = 60;
-       while ((c = getopt(argc, argv, "f:r:l:VvM:q:")) >= 0) {
+       mplp.prior_type = MC_PTYPE_FULL;
+       mplp.theta = 1e-3;
+       while ((c = getopt(argc, argv, "f:r:l:VvM:q:t:2")) >= 0) {
                switch (c) {
+               case 't': mplp.theta = atof(optarg); break;
+               case '2': mplp.prior_type = MC_PTYPE_COND2; break;
                case 'f':
                        mplp.fai = fai_load(optarg);
                        if (mplp.fai == 0) return 1;
@@ -655,7 +674,19 @@ int bam_mpileup(int argc, char *argv[])
                }
        }
        if (argc == 1) {
-               fprintf(stderr, "Usage: samtools mpileup [-r reg] [-f in.fa] [-l pos] in1.bam [in2.bam [...]]\n");
+               fprintf(stderr, "\n");
+               fprintf(stderr, "Usage:   samtools mpileup [options] in1.bam [in2.bam [...]]\n\n");
+               fprintf(stderr, "Options: -f FILE     reference sequence file [null]\n");
+               fprintf(stderr, "         -r STR      region in which pileup is generated [null]\n");
+               fprintf(stderr, "         -l FILE     list of positions (format: chr pos) [null]\n");
+               fprintf(stderr, "         -M INT      cap mapping quality at INT [%d]\n", mplp.max_mq);
+               fprintf(stderr, "         -q INT      filter out alignment with MQ smaller than INT [%d]\n", mplp.min_mq);
+               fprintf(stderr, "         -t FLOAT    scaled mutation rate [%lg]\n", mplp.theta);
+               fprintf(stderr, "         -V          generate VCF output (SNP calling)\n");
+               fprintf(stderr, "         -v          show variant sites only\n");
+               fprintf(stderr, "         -2          conditional prior\n");
+               fprintf(stderr, "\n");
+               fprintf(stderr, "Notes: Assuming error independency and diploid individuals.\n\n");
                return 1;
        }
        mpileup(&mplp, argc - optind, argv + optind);