#define BAM_PLF_VAR_ONLY 0x10
#define BAM_PLF_2ND 0x20
#define BAM_PLF_RANBASE 0x40
+#define BAM_PLF_1STBASE 0x80
+#define BAM_PLF_ALLBASE 0x100
typedef struct {
bam_header_t *h;
uint32_t format;
int tid, len, last_pos;
int mask;
+ int max_depth; // for indel calling, ignore reads with the depth too high. 0 for unlimited
char *ref;
glfFile fp_glf; // for glf output only
} pu_data_t;
g3->offset = pos - d->last_pos;
d->last_pos = pos;
glf3_write1(d->fp_glf, g3);
- if (pos < d->len) {
+ if (pos < d->len) {
+ int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
if (proposed_indels)
- r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
- else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+ r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
}
if (r) { // then write indel line
int het = 3 * n, min;
// update d->ref if necessary
if (d->fai && (int)tid != d->tid) {
free(d->ref);
- d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
+ d->ref = faidx_fetch_seq(d->fai, d->h->target_name[tid], 0, 0x7fffffff, &d->len);
d->tid = tid;
}
rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
}
// call the consensus and indel
if (d->format & BAM_PLF_CNS) { // call consensus
- if (d->format & BAM_PLF_RANBASE) {
- const bam_pileup1_t *p = pu + (int)(drand48() * n);
+ if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE)) { // use a random base or the 1st base as the consensus call
+ const bam_pileup1_t *p = (d->format & BAM_PLF_1STBASE)? pu : pu + (int)(drand48() * n);
int q = bam1_qual(p->b)[p->qpos];
int mapQ = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
uint32_t b = bam1_seqi(bam1_seq(p->b), p->qpos);
cns = b<<28 | 0xf<<24 | mapQ<<16 | q<<8;
+ } else if (d->format & BAM_PLF_ALLBASE) { // collapse all bases
+ uint64_t rmsQ = 0;
+ uint32_t b = 0;
+ for (i = 0; i < n; ++i) {
+ const bam_pileup1_t *p = pu + i;
+ int q = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+ b |= bam1_seqi(bam1_seq(p->b), p->qpos);
+ rmsQ += q * q;
+ }
+ rmsQ = (uint64_t)(sqrt((double)rmsQ / n) + .499);
+ cns = b<<28 | 0xf<<24 | rmsQ<<16 | 60<<8;
} else cns = bam_maqcns_call(n, pu, d->c);
}
- if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
- if (proposed_indels) // the first element gives the size of the array
- r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
- else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+ if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
+ int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+ if (proposed_indels) // the first element gives the size of the array
+ r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
}
// when only variant sites are asked for, test if the site is a variant
if ((d->format & BAM_PLF_CNS) && (d->format & BAM_PLF_VAR_ONLY)) {
int c, is_SAM = 0;
char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
+ d->max_depth = 0;
d->tid = -1; d->mask = BAM_DEF_MASK;
d->c = bam_maqcns_init();
d->ido = bam_maqindel_opt_init();
- while ((c = getopt(argc, argv, "st:f:cT:N:r:l:im:gI:G:vM:S2aR")) >= 0) {
+ while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:")) >= 0) {
switch (c) {
case 'a': d->c->is_soap = 1; break;
case 's': d->format |= BAM_PLF_SIMPLE; break;
case 'N': d->c->n_hap = atoi(optarg); break;
case 'r': d->c->het_rate = atof(optarg); break;
case 'M': d->c->cap_mapQ = atoi(optarg); break;
+ case 'd': d->max_depth = atoi(optarg); break;
case 'c': d->format |= BAM_PLF_CNS; break;
case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
case 'v': d->format |= BAM_PLF_VAR_ONLY; break;
case 'm': d->mask = strtol(optarg, 0, 0); break;
case 'g': d->format |= BAM_PLF_GLF; break;
case '2': d->format |= BAM_PLF_2ND; break;
- case 'R': d->format |= BAM_PLF_RANBASE; break;
case 'I': d->ido->q_indel = atoi(optarg); break;
case 'G': d->ido->r_indel = atof(optarg); break;
case 'S': is_SAM = 1; break;
+ case 'R':
+ if (strcmp(optarg, "random") == 0) d->format |= BAM_PLF_RANBASE;
+ else if (strcmp(optarg, "first") == 0) d->format |= BAM_PLF_1STBASE;
+ else if (strcmp(optarg, "all") == 0) d->format |= BAM_PLF_ALLBASE;
+ else fprintf(stderr, "[bam_pileup] unrecognized -R\n");
+ break;
default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
}
}
fprintf(stderr, " -i only show lines/consensus with indels\n");
fprintf(stderr, " -m INT filtering reads with bits in INT [%d]\n", d->mask);
fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
+ fprintf(stderr, " -d INT limit maximum depth for indels [unlimited]\n");
fprintf(stderr, " -t FILE list of reference sequences (force -S)\n");
fprintf(stderr, " -l FILE list of sites at which pileup is output\n");
fprintf(stderr, " -f FILE reference sequence in the FASTA format\n\n");
fprintf(stderr, " -c output the maq consensus sequence\n");
- fprintf(stderr, " -R randomly choose a random base as the consensus (force -c)\n");
fprintf(stderr, " -v print variants only (for -c)\n");
fprintf(stderr, " -g output in the GLFv3 format (suppressing -c/-i/-s)\n");
fprintf(stderr, " -T FLOAT theta in maq consensus calling model (for -c/-g) [%f]\n", d->c->theta);
free(fn_list); free(fn_fa); free(d);
return 1;
}
- if (d->format & BAM_PLF_RANBASE) d->format |= BAM_PLF_CNS;
+ if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE|BAM_PLF_ALLBASE)) d->format |= BAM_PLF_CNS;
if (fn_fa) d->fai = fai_load(fn_fa);
if (d->format & (BAM_PLF_CNS|BAM_PLF_GLF)) bam_maqcns_prepare(d->c); // consensus calling
if (d->format & BAM_PLF_GLF) { // for glf output
projects / samtools.git / blobdiff