* samtools-0.1.8-18 (r763)

[samtools.git] / bam_plcmd.c

diff --git a/bam_plcmd.c b/bam_plcmd.c
index a23477f40ec3147db38529140b09bac5531af54b..c8ab7ba8dc2535940ca6b30fe43d0d2206207594 100644 (file)
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -375,6 +375,8 @@ int bam_pileup(int argc, char *argv[])
                default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
                }
        }
+       if (d->c->errmod != BAM_ERRMOD_MAQ2) d->c->theta += 0.02;
+       if (d->c->theta > 1.0) d->c->theta = 1.0;
        if (fn_list) is_SAM = 1;
        if (optind == argc) {
                fprintf(stderr, "\n");
@@ -454,10 +456,11 @@ int bam_pileup(int argc, char *argv[])
 
 #define MPLP_GLF   0x10
 #define MPLP_NO_COMP 0x20
+#define MPLP_NO_ORPHAN 0x40
+#define MPLP_REALN   0x80
 
 typedef struct {
-       int max_mq, min_mq, flag, min_baseQ;
-       double theta;
+       int max_mq, min_mq, flag, min_baseQ, capQ_thres;
        char *reg, *fn_pos;
        faidx_t *fai;
        kh_64_t *hash;
@@ -466,7 +469,8 @@ typedef struct {
 typedef struct {
        bamFile fp;
        bam_iter_t iter;
-       int min_mq;
+       int min_mq, flag, ref_id, capQ_thres;
+       char *ref;
 } mplp_aux_t;
 
 typedef struct {
@@ -477,11 +481,25 @@ typedef struct {
 
 static int mplp_func(void *data, bam1_t *b)
 {
+       extern int bam_realn(bam1_t *b, const char *ref);
+       extern int bam_prob_realn(bam1_t *b, const char *ref);
+       extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
        mplp_aux_t *ma = (mplp_aux_t*)data;
-       int ret;
+       int ret, skip = 0;
        do {
+               int has_ref = (ma->ref && ma->ref_id == b->core.tid)? 1 : 0;
                ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
-       } while (b->core.qual < ma->min_mq && ret >= 0);
+               if (ret < 0) break;
+               skip = 0;
+               if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn(b, ma->ref);
+               if (has_ref && ma->capQ_thres > 10) {
+                       int q = bam_cap_mapQ(b, ma->ref, ma->capQ_thres);
+                       if (q < 0) skip = 1;
+                       else if (b->core.qual > q) b->core.qual = q;
+               } else if (b->core.flag&BAM_FUNMAP) skip = 1;
+               else if (b->core.qual < ma->min_mq) skip = 1; 
+               else if ((ma->flag&MPLP_NO_ORPHAN) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
+       } while (skip);
        return ret;
 }
 
@@ -541,7 +559,9 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                bam_header_t *h_tmp;
                data[i] = calloc(1, sizeof(mplp_aux_t));
                data[i]->min_mq = conf->min_mq;
-               data[i]->fp = bam_open(fn[i], "r");
+               data[i]->flag = conf->flag;
+               data[i]->capQ_thres = conf->capQ_thres;
+               data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
                h_tmp = bam_header_read(data[i]->fp);
                bam_smpl_add(sm, fn[i], h_tmp->text);
                if (conf->reg) {
@@ -610,8 +630,9 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                        if (k == kh_end(hash)) continue;
                }
                if (tid != ref_tid) {
-                       free(ref);
+                       free(ref); ref = 0;
                        if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
+                       for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
                        ref_tid = tid;
                }
                if (conf->flag & MPLP_GLF) {
@@ -676,11 +697,10 @@ int bam_mpileup(int argc, char *argv[])
        mplp_conf_t mplp;
        memset(&mplp, 0, sizeof(mplp_conf_t));
        mplp.max_mq = 60;
-       mplp.theta = 1e-3;
        mplp.min_baseQ = 13;
-       while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:u")) >= 0) {
+       mplp.capQ_thres = 0;
+       while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:")) >= 0) {
                switch (c) {
-               case 't': mplp.theta = atof(optarg); break;
                case 'f':
                        mplp.fai = fai_load(optarg);
                        if (mplp.fai == 0) return 1;
@@ -688,7 +708,11 @@ int bam_mpileup(int argc, char *argv[])
                case 'r': mplp.reg = strdup(optarg); break;
                case 'l': mplp.fn_pos = strdup(optarg); break;
                case 'g': mplp.flag |= MPLP_GLF; break;
-               case 'u': mplp.flag |= MPLP_NO_COMP; break;
+               case 'u': mplp.flag |= MPLP_NO_COMP | MPLP_GLF; break;
+               case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
+               case 'O': mplp.flag |= MPLP_NO_ORPHAN; break;
+               case 'R': mplp.flag |= MPLP_REALN; break;
+               case 'C': mplp.capQ_thres = atoi(optarg); break;
                case 'M': mplp.max_mq = atoi(optarg); break;
                case 'q': mplp.min_mq = atoi(optarg); break;
                case 'Q': mplp.min_baseQ = atoi(optarg); break;
@@ -703,8 +727,8 @@ int bam_mpileup(int argc, char *argv[])
                fprintf(stderr, "         -M INT      cap mapping quality at INT [%d]\n", mplp.max_mq);
                fprintf(stderr, "         -Q INT      min base quality [%d]\n", mplp.min_baseQ);
                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, "         -g          generate GLF output\n");
+               fprintf(stderr, "         -g          generate BCF output\n");
+               fprintf(stderr, "         -u          do not compress BCF output\n");
                fprintf(stderr, "\n");
                fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
                return 1;