#define MPLP_REALN 0x80
#define MPLP_FMT_DP 0x100
#define MPLP_FMT_SP 0x200
+#define MPLP_NO_INDEL 0x400
+#define MPLP_EXT_BAQ 0x800
typedef struct {
int max_mq, min_mq, flag, min_baseQ, capQ_thres, max_depth;
- char *reg, *fn_pos;
+ int openQ, extQ, tandemQ, min_support; // for indels
+ double min_frac; // for indels
+ char *reg, *fn_pos, *pl_list;
faidx_t *fai;
kh_64_t *hash;
} mplp_conf_t;
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_prob_realn_core(bam1_t *b, const char *ref, int);
extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
mplp_aux_t *ma = (mplp_aux_t*)data;
int ret, skip = 0;
do {
- int has_ref = (ma->ref && ma->ref_id == b->core.tid)? 1 : 0;
+ int has_ref;
ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
if (ret < 0) break;
+ has_ref = (ma->ref && ma->ref_id == b->core.tid)? 1 : 0;
skip = 0;
- if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn(b, ma->ref);
+ if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn_core(b, ma->ref, (ma->flag & MPLP_EXT_BAQ)? 3 : 1);
if (has_ref && ma->capQ_thres > 10) {
int q = bam_cap_mapQ(b, ma->ref, ma->capQ_thres);
if (q < 0) skip = 1;
static int mpileup(mplp_conf_t *conf, int n, char **fn)
{
+ extern void *bcf_call_add_rg(void *rghash, const char *hdtext, const char *list);
+ extern void bcf_call_del_rghash(void *rghash);
mplp_aux_t **data;
int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid, max_depth;
const bam_pileup1_t **plp;
bam_header_t *h = 0;
char *ref;
khash_t(64) *hash = 0;
+ void *rghash = 0;
bcf_callaux_t *bca = 0;
bcf_callret1_t *bcr = 0;
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);
+ rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
if (conf->reg) {
int beg, end;
bam_index_t *idx;
bcf_hdr_write(bp, bh);
bca = bcf_call_init(-1., conf->min_baseQ);
bcr = calloc(sm->n, sizeof(bcf_callret1_t));
+ bca->rghash = rghash;
+ bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
+ bca->min_frac = conf->min_frac;
+ bca->min_support = conf->min_support;
}
ref_tid = -1; ref = 0;
iter = bam_mplp_init(n, mplp_func, (void**)data);
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, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0, (conf->flag&MPLP_FMT_SP));
+ bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
+ (conf->flag&MPLP_FMT_SP), 0, 0);
bcf_write(bp, bh, b);
bcf_destroy(b);
- // test if there is an indel
- if (conf->flag&MPLP_REALN) {
- for (i = 0; i < gplp.n; ++i) {
- int j;
- for (j = 0; j < gplp.n_plp[i]; ++j) {
- const bam_pileup1_t *p = gplp.plp[i] + j;
- if (p->indel != 0) break;
- }
- if (j != gplp.n_plp[i]) break;
- }
- if (i < gplp.n) { // at least one of the read contains a gap
- for (i = 0; i < gplp.n; ++i)
- bcf_call_glfgen_gap(pos, gplp.n_plp[i], gplp.plp[i], bca, bcr + i);
+ // call indels
+ if (!(conf->flag&MPLP_NO_INDEL) && bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0) {
+ for (i = 0; i < gplp.n; ++i)
+ bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], -1, bca, bcr + i);
+ if (bcf_call_combine(gplp.n, bcr, -1, &bc) >= 0) {
+ b = calloc(1, sizeof(bcf1_t));
+ bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
+ (conf->flag&MPLP_FMT_SP), bca, ref);
+ bcf_write(bp, bh, b);
+ bcf_destroy(b);
}
}
} else {
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);
+ bcf_call_del_rghash(rghash);
if (hash) { // free the hash table
khint_t k;
for (k = kh_begin(hash); k < kh_end(hash); ++k)
int c;
const char *file_list = NULL;
char **fn = NULL;
- int nfiles = 0;
+ int nfiles = 0, use_orphan = 0;
mplp_conf_t mplp;
memset(&mplp, 0, sizeof(mplp_conf_t));
mplp.max_mq = 60;
mplp.min_baseQ = 13;
mplp.capQ_thres = 0;
mplp.max_depth = 250;
+ mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
+ mplp.min_frac = 0.002; mplp.min_support = 1;
mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
- while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:BDSd:b:")) >= 0) {
+ while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:b:P:o:e:h:Im:F:E")) >= 0) {
switch (c) {
case 'f':
mplp.fai = fai_load(optarg);
case 'd': mplp.max_depth = atoi(optarg); break;
case 'r': mplp.reg = strdup(optarg); break;
case 'l': mplp.fn_pos = strdup(optarg); break;
+ case 'P': mplp.pl_list = 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 'I': mplp.flag |= MPLP_NO_INDEL; break;
+ case 'E': mplp.flag |= MPLP_EXT_BAQ; 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;
case 'b': file_list = optarg; break;
+ case 'o': mplp.openQ = atoi(optarg); break;
+ case 'e': mplp.extQ = atoi(optarg); break;
+ case 'h': mplp.tandemQ = atoi(optarg); break;
+ case 'A': use_orphan = 1; break;
+ case 'F': mplp.min_frac = atof(optarg); break;
+ case 'm': mplp.min_support = atoi(optarg); break;
}
}
+ if (use_orphan) mplp.flag &= ~MPLP_NO_ORPHAN;
if (argc == 1) {
fprintf(stderr, "\n");
fprintf(stderr, "Usage: samtools mpileup [options] in1.bam [in2.bam [...]]\n\n");
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, " -P STR comma separated list of platforms for indels [all]\n");
+ fprintf(stderr, " -o INT Phred-scaled gap open sequencing error probability [%d]\n", mplp.openQ);
+ fprintf(stderr, " -e INT Phred-scaled gap extension seq error probability [%d]\n", mplp.extQ);
+ fprintf(stderr, " -h INT coefficient for homopolyer errors [%d]\n", mplp.tandemQ);
+ fprintf(stderr, " -m INT minimum gapped reads for indel candidates [%d]\n", mplp.min_support);
+ fprintf(stderr, " -F FLOAT minimum fraction of gapped reads for candidates [%g]\n", mplp.min_frac);
+ fprintf(stderr, " -A use anomalous read pairs in SNP/INDEL calling\n");
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, " -I do not perform indel calling\n");
fprintf(stderr, "\n");
fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
return 1;
}
else
mpileup(&mplp, argc - optind, argv + optind);
- free(mplp.reg);
+ free(mplp.reg); free(mplp.pl_list);
if (mplp.fai) fai_destroy(mplp.fai);
return 0;
}
projects / samtools.git / blobdiff