* samtools-0.1.9-2 (r787)

[samtools.git] / bam_plcmd.c

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 6d42471c007a6b34d5b7b1db3ee535c974564839..f44627f79d56a0389f52dc3b1a25d054bbe19864 100644 (file)
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -473,6 +473,7 @@ int bam_pileup(int argc, char *argv[])
                        bam_plp_set_mask(iter, d->mask);
                        while ((ret = samread(fp, b)) >= 0) {
                                int skip = 0;
+                               if ((int)b->core.tid < 0) break;
                                // update d->ref if necessary
                                if (d->fai && (int)b->core.tid != d->tid) {
                                        free(d->ref);
@@ -527,9 +528,11 @@ int bam_pileup(int argc, char *argv[])
 #define MPLP_NO_COMP 0x20
 #define MPLP_NO_ORPHAN 0x40
 #define MPLP_REALN   0x80
+#define MPLP_FMT_DP 0x100
+#define MPLP_FMT_SP 0x200
 
 typedef struct {
-       int max_mq, min_mq, flag, min_baseQ, capQ_thres;
+       int max_mq, min_mq, flag, min_baseQ, capQ_thres, max_depth;
        char *reg, *fn_pos;
        faidx_t *fai;
        kh_64_t *hash;
@@ -598,7 +601,7 @@ static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
 static int mpileup(mplp_conf_t *conf, int n, char **fn)
 {
        mplp_aux_t **data;
-       int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid;
+       int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid, max_depth;
        const bam_pileup1_t **plp;
        bam_mplp_t iter;
        bam_header_t *h = 0;
@@ -691,6 +694,14 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
        }
        ref_tid = -1; ref = 0;
        iter = bam_mplp_init(n, mplp_func, (void**)data);
+       max_depth = conf->max_depth;
+       if (max_depth * sm->n > 1<<20)
+               fprintf(stderr, "(%s) Max depth is above 1M. Potential memory hog!\n", __func__);
+       if (max_depth * sm->n < 8000) {
+               max_depth = 8000 / sm->n;
+               fprintf(stderr, "<%s> Set max per-sample depth to %d\n", __func__, max_depth);
+       }
+       bam_mplp_set_maxcnt(iter, max_depth);
        while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
                if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
                if (hash) {
@@ -713,7 +724,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                        for (i = 0; i < gplp.n; ++i)
                                bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
                        bcf_call_combine(gplp.n, bcr, ref16, &bc);
-                       bcf_call2bcf(tid, pos, &bc, b);
+                       bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0, (conf->flag&MPLP_FMT_SP));
                        bcf_write(bp, bh, b);
                        bcf_destroy(b);
                } else {
@@ -768,19 +779,25 @@ int bam_mpileup(int argc, char *argv[])
        mplp.max_mq = 60;
        mplp.min_baseQ = 13;
        mplp.capQ_thres = 0;
-       while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:")) >= 0) {
+       mplp.max_depth = 250;
+       mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
+       while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:BDSd:")) >= 0) {
                switch (c) {
                case 'f':
                        mplp.fai = fai_load(optarg);
                        if (mplp.fai == 0) return 1;
                        break;
+               case 'd': mplp.max_depth = atoi(optarg); break;
                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 | MPLP_GLF; break;
                case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
+               case 'B': mplp.flag &= ~MPLP_REALN & ~MPLP_NO_ORPHAN; break;
                case 'O': mplp.flag |= MPLP_NO_ORPHAN; break;
                case 'R': mplp.flag |= MPLP_REALN; break;
+               case 'D': mplp.flag |= MPLP_FMT_DP; break;
+               case 'S': mplp.flag |= MPLP_FMT_SP; 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;
@@ -796,8 +813,12 @@ 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, "         -d INT      max per-sample depth [%d]\n", mplp.max_depth);
                fprintf(stderr, "         -g          generate BCF output\n");
                fprintf(stderr, "         -u          do not compress BCF output\n");
+               fprintf(stderr, "         -B          disable BAQ computation\n");
+               fprintf(stderr, "         -D          output per-sample DP\n");
+               fprintf(stderr, "         -S          output per-sample SP (strand bias P-value, slow)\n");
                fprintf(stderr, "\n");
                fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
                return 1;