#include "sam.h"
#include "faidx.h"
#include "bam_maqcns.h"
-#include "bam2bcf.h"
#include "khash.h"
#include "glf.h"
#include "kstring.h"
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");
* mpileup *
***********/
+#include <assert.h>
+#include "bam2bcf.h"
+#include "sample.h"
+
#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;
+ int max_mq, min_mq, flag, min_baseQ, capQ_thres;
double theta;
char *reg, *fn_pos;
faidx_t *fai;
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 {
+ int n;
+ int *n_plp, *m_plp;
+ bam_pileup1_t **plp;
+} mplp_pileup_t;
+
static int mplp_func(void *data, bam1_t *b)
{
+ extern int bam_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_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;
}
+static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
+ int n, char *const*fn, int *n_plp, const bam_pileup1_t **plp)
+{
+ int i, j;
+ memset(m->n_plp, 0, m->n * sizeof(int));
+ for (i = 0; i < n; ++i) {
+ for (j = 0; j < n_plp[i]; ++j) {
+ const bam_pileup1_t *p = plp[i] + j;
+ uint8_t *q;
+ int id = -1;
+ q = bam_aux_get(p->b, "RG");
+ if (q) id = bam_smpl_rg2smid(sm, fn[i], (char*)q+1, buf);
+ if (id < 0) id = bam_smpl_rg2smid(sm, fn[i], 0, buf);
+ assert(id >= 0 && id < m->n);
+ if (m->n_plp[id] == m->m_plp[id]) {
+ m->m_plp[id] = m->m_plp[id]? m->m_plp[id]<<1 : 8;
+ m->plp[id] = realloc(m->plp[id], sizeof(bam_pileup1_t) * m->m_plp[id]);
+ }
+ m->plp[id][m->n_plp[id]++] = *p;
+ }
+ }
+}
+
static int mpileup(mplp_conf_t *conf, int n, char **fn)
{
mplp_aux_t **data;
bcf_t *bp = 0;
bcf_hdr_t *bh = 0;
+ bam_sample_t *sm = 0;
+ kstring_t buf;
+ mplp_pileup_t gplp;
+
+ memset(&gplp, 0, sizeof(mplp_pileup_t));
+ memset(&buf, 0, sizeof(kstring_t));
memset(&bc, 0, sizeof(bcf_call_t));
data = calloc(n, sizeof(void*));
plp = calloc(n, sizeof(void*));
n_plp = calloc(n, sizeof(int*));
+ sm = bam_smpl_init();
// read the header and initialize data
for (i = 0; i < n; ++i) {
bam_header_t *h_tmp;
data[i] = calloc(1, sizeof(mplp_aux_t));
data[i]->min_mq = conf->min_mq;
+ data[i]->flag = conf->flag;
+ data[i]->capQ_thres = conf->capQ_thres;
data[i]->fp = 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) {
int beg, end;
bam_index_t *idx;
bam_header_destroy(h_tmp);
}
}
+ gplp.n = sm->n;
+ gplp.n_plp = calloc(sm->n, sizeof(int));
+ gplp.m_plp = calloc(sm->n, sizeof(int));
+ gplp.plp = calloc(sm->n, sizeof(void*));
+
+ fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
if (conf->fn_pos) hash = load_pos(conf->fn_pos, h);
// write the VCF header
if (conf->flag & MPLP_GLF) {
bh->name = malloc(s.l);
memcpy(bh->name, s.s, s.l);
s.l = 0;
- for (i = 0; i < n; ++i) {
- const char *p;
- if ((p = strstr(fn[i], ".bam")) != 0)
- kputsn(fn[i], p - fn[i], &s);
- else kputs(fn[i], &s);
- kputc('\0', &s);
+ for (i = 0; i < sm->n; ++i) {
+ kputs(sm->smpl[i], &s); kputc('\0', &s);
}
bh->l_smpl = s.l;
bh->sname = malloc(s.l);
free(s.s);
bcf_hdr_sync(bh);
bcf_hdr_write(bp, bh);
- }
- // mpileup
- if (conf->flag & MPLP_GLF) {
bca = bcf_call_init(-1., conf->min_baseQ);
- bcr = calloc(n, sizeof(bcf_callret1_t));
+ bcr = calloc(sm->n, sizeof(bcf_callret1_t));
}
ref_tid = -1; ref = 0;
iter = bam_mplp_init(n, mplp_func, (void**)data);
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) {
int _ref0, ref16;
bcf1_t *b = calloc(1, sizeof(bcf1_t));
+ group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp);
_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
ref16 = bam_nt16_table[_ref0];
- for (i = 0; i < n; ++i)
- bcf_call_glfgen(n_plp[i], plp[i], ref16, bca, bcr + i);
- bcf_call_combine(n, bcr, ref16, &bc);
+ 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_write(bp, bh, b);
- //fprintf(stderr, "%d,%d,%d\n", b->tid, b->pos, b->l_str);
bcf_destroy(b);
} else {
printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
putchar('\n');
}
}
+
bcf_close(bp);
+ bam_smpl_destroy(sm); free(buf.s);
+ for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]);
+ free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp);
if (hash) { // free the hash table
khint_t k;
for (k = kh_begin(hash); k < kh_end(hash); ++k)
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:t:Q:uaORC:")) >= 0) {
switch (c) {
case 't': mplp.theta = atof(optarg); break;
case 'f':
case 'l': mplp.fn_pos = strdup(optarg); break;
case 'g': mplp.flag |= MPLP_GLF; break;
case 'u': mplp.flag |= MPLP_NO_COMP; 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;
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;