another minor fix to mpileup

[samtools.git] / bam_plcmd.c

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 258b83fd72b6ac21db863ea837d46c92d0ff0aef..a5bcdf560a5bfdfd14a9a2eccffb46be1164e8cc 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;
@@ -537,7 +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) {
@@ -554,7 +558,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                }
                if (conf->vcf) {
                        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;
+                       int j, _ref, _alt, _ref0, depth, rms_q, _ref0b, is_var = 0, qref = 0, filtered = 0;
                        uint64_t sqr_sum;
                        _ref0 = _ref0b = (ref && pos < ref_len)? ref[pos] : 'N';
                        _ref0 = bam_nt16_nt4_table[bam_nt16_table[_ref0]];
@@ -577,6 +581,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                                qref = (int)(-3.434 * log(q) + .499);
                                if (qref > 99) qref = 99;
                        }
+                       filtered = (f >= 0. && f <= 1.)? 0 : 1;
                        if (conf->var_only && !is_var) continue;
                        printf("%s\t%d\t.\t%c\t", h->target_name[tid], pos + 1, _ref0b);
                        if (is_var) {
@@ -584,7 +589,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                                else printf("%c,%c", "ACGTN"[_ref], "ACGTN"[_alt]);
                        } else putchar('.');
                        printf("\t%d\t", qref);
-                       if (f0 < 0.) printf("Q13\t");
+                       if (filtered) printf("Q13\t");
                        else printf(".\t");
                        for (i = depth = 0, sqr_sum = 0; i < n; ++i) {
                                depth += n_plp[i];
@@ -598,6 +603,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                        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");
                        if (fpost >= 0. && fpost <= 1.) {
                                for (i = 0; i < n; ++i) {
                                        int x = mc_call_gt(ma, fpost, i);
@@ -650,8 +656,12 @@ 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;
@@ -672,8 +682,10 @@ int bam_mpileup(int argc, char *argv[])
                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;