void closeBamFile(BamReaderData * data) {
// Seriously, does samtools not provide any convience destructors!??
bam_mplp_destroy(data->iter);
//bam_header_destroy(data->data->h);
bam_close(data->data->fp);
if (data->data->iter)
bam_iter_destroy(data->data->iter);
free(data->data);
bam_index_destroy(data->idx);
}
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, tid0 = -1, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid = -1, max_depth, max_indel_depth;
const bam_pileup1_t **plp;
bam_mplp_t iter;
bam_header_t *h = 0;
char *ref;
void *rghash = 0;
bcf_callaux_t *bca = 0;
bcf_callret1_t *bcr = 0;
bcf_call_t bc;
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]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
data[i]->conf = conf;
h_tmp = bam_header_read(data[i]->fp);
data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
bam_smpl_add(sm, fn[i], (conf->flag&MPLP_IGNORE_RG)? 0 : 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;
idx = bam_index_load(fn[i]);
if (idx == 0) {
fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
exit(1);
}
if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
exit(1);
}
if (i == 0) tid0 = tid, beg0 = beg, end0 = end;
data[i]->iter = bam_iter_query(idx, tid, beg, end);
bam_index_destroy(idx);
}
if (i == 0) h = h_tmp;
else {
// FIXME: to check consistency
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);
// write the VCF header
if (conf->flag & MPLP_GLF) {
kstring_t s;
bh = calloc(1, sizeof(bcf_hdr_t));
s.l = s.m = 0; s.s = 0;
bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
for (i = 0; i < h->n_targets; ++i) {
kputs(h->target_name[i], &s);
kputc('\0', &s);
}
bh->l_nm = s.l;
bh->name = malloc(s.l);
memcpy(bh->name, s.s, s.l);
s.l = 0;
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);
memcpy(bh->sname, s.s, s.l);
bh->txt = malloc(strlen(BAM_VERSION) + 64);
bh->l_txt = 1 + sprintf(bh->txt, "##samtoolsVersion=%s\n", BAM_VERSION);
free(s.s);
bcf_hdr_sync(bh);
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;
}
if (tid0 >= 0 && conf->fai) { // region is set
ref = faidx_fetch_seq(conf->fai, h->target_name[tid0], 0, 0x7fffffff, &ref_len);
ref_tid = tid0;
for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid0;
} else 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-file depth to %d\n", __func__, max_depth);
}
max_indel_depth = conf->max_indel_depth * sm->n;
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 (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
if (tid != ref_tid) {
free(ref); ref = 0;
if (conf->fai) ref = faidx_fetch_seq(conf->fai, h->target_name[tid], 0, 0x7fffffff, &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 total_depth, _ref0, ref16;
bcf1_t *b = calloc(1, sizeof(bcf1_t));
for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp, conf->flag & MPLP_IGNORE_RG);
_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
ref16 = bam_nt16_table[_ref0];
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), 0, 0);
bcf_write(bp, bh, b);
bcf_destroy(b);
// call indels
if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && 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 {
printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
for (i = 0; i < n; ++i) {
int j;
printf("\t%d\t", n_plp[i]);
if (n_plp[i] == 0) {
printf("*\t*"); // FIXME: printf() is very slow...
if (conf->flag & MPLP_PRINT_POS) printf("\t*");
} else {
for (j = 0; j < n_plp[i]; ++j)
pileup_seq(plp[i] + j, pos, ref_len, ref);
putchar('\t');
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
int c = bam1_qual(p->b)[p->qpos] + 33;
if (c > 126) c = 126;
putchar(c);
}
if (conf->flag & MPLP_PRINT_MAPQ) {
putchar('\t');
for (j = 0; j < n_plp[i]; ++j) {
int c = plp[i][j].b->core.qual + 33;
if (c > 126) c = 126;
putchar(c);
}
}
if (conf->flag & MPLP_PRINT_POS) {
putchar('\t');
for (j = 0; j < n_plp[i]; ++j) {
if (j > 0) putchar(',');
printf("%d", plp[i][j].qpos + 1); // FIXME: printf() is very slow...
}
}
}
}
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);
bcf_call_del_rghash(rghash);
bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
bam_mplp_destroy(iter);
bam_header_destroy(h);
for (i = 0; i < n; ++i) {
bam_close(data[i]->fp);
if (data[i]->iter) bam_iter_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(plp); free(ref); free(n_plp);
return 0;
}
int main_depth(int argc, char *argv[])
#endif
{
int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0;
const bam_pileup1_t **plp;
char *reg = 0; // specified region
void *bed = 0; // BED data structure
bam_header_t *h = 0; // BAM header of the 1st input
aux_t **data;
bam_mplp_t mplp;
// parse the command line
while ((n = getopt(argc, argv, "r:b:q:Q:")) >= 0) {
switch (n) {
case 'r': reg = strdup(optarg); break; // parsing a region requires a BAM header
case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
case 'q': baseQ = atoi(optarg); break; // base quality threshold
case 'Q': mapQ = atoi(optarg); break; // mapping quality threshold
}
}
if (optind == argc) {
fprintf(stderr, "Usage: bam2depth [-r reg] [-q baseQthres] [-Q mapQthres] [-b in.bed] <in1.bam> [...]\n");
return 1;
}
// initialize the auxiliary data structures
n = argc - optind; // the number of BAMs on the command line
data = (aux_t **) calloc(n, sizeof(void*)); // data[i] for the i-th input
beg = 0; end = 1<<30; tid = -1; // set the default region
for (i = 0; i < n; ++i) {
bam_header_t *htmp;
data[i] = (aux_t *) calloc(1, sizeof(aux_t));
data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
data[i]->min_mapQ = mapQ; // set the mapQ filter
htmp = bam_header_read(data[i]->fp); // read the BAM header
if (i == 0) {
h = htmp; // keep the header of the 1st BAM
if (reg) bam_parse_region(h, reg, &tid, &beg, &end); // also parse the region
} else bam_header_destroy(htmp); // if not the 1st BAM, trash the header
if (tid >= 0) { // if a region is specified and parsed successfully
bam_index_t *idx = bam_index_load(argv[optind+i]); // load the index
data[i]->iter = bam_iter_query(idx, tid, beg, end); // set the iterator
bam_index_destroy(idx); // the index is not needed any more; phase out of the memory
}
}
// the core multi-pileup loop
mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
n_plp = (int*) calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = (bam_pileup1_t **) calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
if (pos < beg || pos >= end) continue; // out of range; skip
if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
for (i = 0; i < n; ++i) { // base level filters have to go here
int j, m = 0;
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
else if (bam1_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
}
printf("\t%d", n_plp[i] - m); // this the depth to output
}
putchar('\n');
}
free(n_plp); free(plp);
bam_mplp_destroy(mplp);
bam_header_destroy(h);
for (i = 0; i < n; ++i) {
bam_close(data[i]->fp);
if (data[i]->iter) bam_iter_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(reg);
if (bed) bed_destroy(bed);
return 0;
}
int main_bedcov(int argc, char *argv[])
{
gzFile fp;
kstring_t str;
kstream_t *ks;
hts_idx_t **idx;
aux_t **aux;
int *n_plp, dret, i, n, c, min_mapQ = 0;
int64_t *cnt;
const bam_pileup1_t **plp;
int usage = 0;
sam_global_args ga = SAM_GLOBAL_ARGS_INIT;
static const struct option lopts[] = {
SAM_OPT_GLOBAL_OPTIONS('-', 0, '-', '-', 0),
{ NULL, 0, NULL, 0 }
};
while ((c = getopt_long(argc, argv, "Q:", lopts, NULL)) >= 0) {
switch (c) {
case 'Q': min_mapQ = atoi(optarg); break;
default: if (parse_sam_global_opt(c, optarg, lopts, &ga) == 0) break;
/* else fall-through */
case '?': usage = 1; break;
}
if (usage) break;
}
if (usage || optind + 2 > argc) {
fprintf(pysam_stderr, "Usage: samtools bedcov [options] <in.bed> <in1.bam> [...]\n\n");
fprintf(pysam_stderr, " -Q INT Only count bases of at least INT quality [0]\n");
sam_global_opt_help(pysam_stderr, "-.--.");
return 1;
}
memset(&str, 0, sizeof(kstring_t));
n = argc - optind - 1;
aux = calloc(n, sizeof(aux_t*));
idx = calloc(n, sizeof(hts_idx_t*));
for (i = 0; i < n; ++i) {
aux[i] = calloc(1, sizeof(aux_t));
aux[i]->min_mapQ = min_mapQ;
aux[i]->fp = sam_open_format(argv[i+optind+1], "r", &ga.in);
if (aux[i]->fp)
idx[i] = sam_index_load(aux[i]->fp, argv[i+optind+1]);
if (aux[i]->fp == 0 || idx[i] == 0) {
fprintf(pysam_stderr, "ERROR: fail to open index BAM file '%s'\n", argv[i+optind+1]);
return 2;
}
// TODO bgzf_set_cache_size(aux[i]->fp, 20);
aux[i]->header = sam_hdr_read(aux[i]->fp);
if (aux[i]->header == NULL) {
fprintf(pysam_stderr, "ERROR: failed to read header for '%s'\n",
argv[i+optind+1]);
return 2;
}
}
cnt = calloc(n, 8);
fp = gzopen(argv[optind], "rb");
ks = ks_init(fp);
n_plp = calloc(n, sizeof(int));
plp = calloc(n, sizeof(bam_pileup1_t*));
while (ks_getuntil(ks, KS_SEP_LINE, &str, &dret) >= 0) {
char *p, *q;
int tid, beg, end, pos;
bam_mplp_t mplp;
for (p = q = str.s; *p && *p != '\t'; ++p);
if (*p != '\t') goto bed_error;
*p = 0; tid = bam_name2id(aux[0]->header, q); *p = '\t';
if (tid < 0) goto bed_error;
for (q = p = p + 1; isdigit(*p); ++p);
if (*p != '\t') goto bed_error;
*p = 0; beg = atoi(q); *p = '\t';
for (q = p = p + 1; isdigit(*p); ++p);
if (*p == '\t' || *p == 0) {
int c = *p;
*p = 0; end = atoi(q); *p = c;
} else goto bed_error;
for (i = 0; i < n; ++i) {
if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
aux[i]->iter = sam_itr_queryi(idx[i], tid, beg, end);
}
mplp = bam_mplp_init(n, read_bam, (void**)aux);
bam_mplp_set_maxcnt(mplp, 64000);
memset(cnt, 0, 8 * n);
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0)
if (pos >= beg && pos < end)
for (i = 0; i < n; ++i) cnt[i] += n_plp[i];
for (i = 0; i < n; ++i) {
kputc('\t', &str);
kputl(cnt[i], &str);
}
fputs(str.s, pysam_stdout) & fputc('\n', pysam_stdout);
bam_mplp_destroy(mplp);
continue;
bed_error:
fprintf(pysam_stderr, "Errors in BED line '%s'\n", str.s);
}
free(n_plp); free(plp);
ks_destroy(ks);
gzclose(fp);
free(cnt);
for (i = 0; i < n; ++i) {
if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
hts_idx_destroy(idx[i]);
bam_hdr_destroy(aux[i]->header);
sam_close(aux[i]->fp);
free(aux[i]);
}
free(aux); free(idx);
free(str.s);
sam_global_args_free(&ga);
return 0;
}
int main_depth(int argc, char *argv[])
{
int i, n, tid, reg_tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0, min_len = 0;
int all = 0, status = EXIT_SUCCESS, nfiles, max_depth = -1;
const bam_pileup1_t **plp;
char *reg = 0; // specified region
void *bed = 0; // BED data structure
char *file_list = NULL, **fn = NULL;
bam_hdr_t *h = NULL; // BAM header of the 1st input
aux_t **data;
bam_mplp_t mplp;
int last_pos = -1, last_tid = -1, ret;
sam_global_args ga = SAM_GLOBAL_ARGS_INIT;
static const struct option lopts[] = {
SAM_OPT_GLOBAL_OPTIONS('-', 0, '-', '-', 0),
{ NULL, 0, NULL, 0 }
};
// parse the command line
while ((n = getopt_long(argc, argv, "r:b:q:Q:l:f:am:d:", lopts, NULL)) >= 0) {
switch (n) {
case 'l':
min_len = atoi(optarg);
break; // minimum query length
case 'r':
reg = strdup(optarg);
break; // parsing a region requires a BAM header
case 'b':
bed = bed_read(optarg); // BED or position list file can be parsed now
if (!bed) {
print_error_errno("depth", "Could not read file \"%s\"", optarg);
return 1;
}
break;
case 'q':
baseQ = atoi(optarg);
break; // base quality threshold
case 'Q':
mapQ = atoi(optarg);
break; // mapping quality threshold
case 'f':
file_list = optarg;
break;
case 'a':
all++;
break;
case 'd':
case 'm':
max_depth = atoi(optarg);
break; // maximum coverage depth
default:
if (parse_sam_global_opt(n, optarg, lopts, &ga) == 0) break;
/* else fall-through */
case '?':
return usage();
}
}
if (optind == argc && !file_list)
return usage();
// initialize the auxiliary data structures
if (file_list)
{
if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
n = nfiles;
argv = fn;
optind = 0;
}
else
n = argc - optind; // the number of BAMs on the command line
data = calloc(n, sizeof(aux_t*)); // data[i] for the i-th input
reg_tid = 0;
beg = 0;
end = INT_MAX; // set the default region
for (i = 0; i < n; ++i) {
int rf;
data[i] = calloc(1, sizeof(aux_t));
data[i]->fp = sam_open_format(argv[optind+i], "r", &ga.in); // open BAM
if (data[i]->fp == NULL) {
print_error_errno("depth", "Could not open \"%s\"", argv[optind+i]);
status = EXIT_FAILURE;
goto depth_end;
}
rf = SAM_FLAG | SAM_RNAME | SAM_POS | SAM_MAPQ | SAM_CIGAR | SAM_SEQ;
if (baseQ) rf |= SAM_QUAL;
if (hts_set_opt(data[i]->fp, CRAM_OPT_REQUIRED_FIELDS, rf)) {
fprintf(stderr, "Failed to set CRAM_OPT_REQUIRED_FIELDS value\n");
return 1;
}
if (hts_set_opt(data[i]->fp, CRAM_OPT_DECODE_MD, 0)) {
fprintf(stderr, "Failed to set CRAM_OPT_DECODE_MD value\n");
return 1;
}
data[i]->min_mapQ = mapQ; // set the mapQ filter
data[i]->min_len = min_len; // set the qlen filter
data[i]->hdr = sam_hdr_read(data[i]->fp); // read the BAM header
if (data[i]->hdr == NULL) {
fprintf(stderr, "Couldn't read header for \"%s\"\n",
argv[optind+i]);
status = EXIT_FAILURE;
goto depth_end;
}
if (reg) { // if a region is specified
hts_idx_t *idx = sam_index_load(data[i]->fp, argv[optind+i]); // load the index
if (idx == NULL) {
print_error("depth", "can't load index for \"%s\"", argv[optind+i]);
status = EXIT_FAILURE;
goto depth_end;
}
data[i]->iter = sam_itr_querys(idx, data[i]->hdr, reg); // set the iterator
hts_idx_destroy(idx); // the index is not needed any more; free the memory
if (data[i]->iter == NULL) {
print_error("depth", "can't parse region \"%s\"", reg);
status = EXIT_FAILURE;
goto depth_end;
}
}
}
h = data[0]->hdr; // easy access to the header of the 1st BAM
if (reg) {
beg = data[0]->iter->beg; // and to the parsed region coordinates
end = data[0]->iter->end;
reg_tid = data[0]->iter->tid;
}
// the core multi-pileup loop
mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
if (0 < max_depth)
bam_mplp_set_maxcnt(mplp,max_depth); // set maximum coverage depth
n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = calloc(n, sizeof(bam_pileup1_t*)); // plp[i] points to the array of covering reads (internal in mplp)
while ((ret=bam_mplp_auto(mplp, &tid, &pos, n_plp, plp)) > 0) { // come to the next covered position
if (pos < beg || pos >= end) continue; // out of range; skip
if (tid >= h->n_targets) continue; // diff number of @SQ lines per file?
if (all) {
while (tid > last_tid) {
if (last_tid >= 0 && !reg) {
// Deal with remainder or entirety of last tid.
while (++last_pos < h->target_len[last_tid]) {
// Horribly inefficient, but the bed API is an obfuscated black box.
if (bed && bed_overlap(bed, h->target_name[last_tid], last_pos, last_pos + 1) == 0)
continue;
fputs(h->target_name[last_tid], stdout);
printf("\t%d", last_pos+1);
for (i = 0; i < n; i++)
putchar('\t'), putchar('0');
putchar('\n');
}
}
last_tid++;
last_pos = -1;
if (all < 2)
break;
}
// Deal with missing portion of current tid
while (++last_pos < pos) {
if (last_pos < beg) continue; // out of range; skip
if (bed && bed_overlap(bed, h->target_name[tid], last_pos, last_pos + 1) == 0)
continue;
fputs(h->target_name[tid], stdout);
printf("\t%d", last_pos+1);
for (i = 0; i < n; i++)
putchar('\t'), putchar('0');
putchar('\n');
}
last_tid = tid;
last_pos = pos;
}
if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue;
fputs(h->target_name[tid], stdout);
printf("\t%d", pos+1); // a customized printf() would be faster
for (i = 0; i < n; ++i) { // base level filters have to go here
int j, m = 0;
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
else if (bam_get_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
}
printf("\t%d", n_plp[i] - m); // this the depth to output
}
putchar('\n');
}
if (ret < 0) status = EXIT_FAILURE;
free(n_plp);
free(plp);
bam_mplp_destroy(mplp);
if (all) {
// Handle terminating region
if (last_tid < 0 && reg && all > 1) {
last_tid = reg_tid;
last_pos = beg-1;
}
while (last_tid >= 0 && last_tid < h->n_targets) {
while (++last_pos < h->target_len[last_tid]) {
if (last_pos >= end) break;
if (bed && bed_overlap(bed, h->target_name[last_tid], last_pos, last_pos + 1) == 0)
continue;
fputs(h->target_name[last_tid], stdout);
printf("\t%d", last_pos+1);
for (i = 0; i < n; i++)
putchar('\t'), putchar('0');
putchar('\n');
}
last_tid++;
last_pos = -1;
if (all < 2 || reg)
break;
}
}
depth_end:
for (i = 0; i < n && data[i]; ++i) {
bam_hdr_destroy(data[i]->hdr);
if (data[i]->fp) sam_close(data[i]->fp);
hts_itr_destroy(data[i]->iter);
free(data[i]);
}
free(data);
free(reg);
if (bed) bed_destroy(bed);
if ( file_list )
{
for (i=0; i<n; i++) free(fn[i]);
free(fn);
}
sam_global_args_free(&ga);
return status;
}
int main_depth(int argc, char *argv[])
#endif
{
int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0, min_len = 0, nfiles;
const bam_pileup1_t **plp;
char *reg = 0; // specified region
void *bed = 0; // BED data structure
char *file_list = NULL, **fn = NULL;
bam_header_t *h = 0; // BAM header of the 1st input
aux_t **data;
bam_mplp_t mplp;
// parse the command line
while ((n = getopt(argc, argv, "r:b:q:Q:l:f:")) >= 0) {
switch (n) {
case 'l': min_len = atoi(optarg); break; // minimum query length
case 'r': reg = strdup(optarg); break; // parsing a region requires a BAM header
case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
case 'q': baseQ = atoi(optarg); break; // base quality threshold
case 'Q': mapQ = atoi(optarg); break; // mapping quality threshold
case 'f': file_list = optarg; break;
}
}
if (optind == argc && !file_list) {
fprintf(stderr, "\n");
fprintf(stderr, "Usage: samtools depth [options] in1.bam [in2.bam [...]]\n");
fprintf(stderr, "Options:\n");
fprintf(stderr, " -b <bed> list of positions or regions\n");
fprintf(stderr, " -f <list> list of input BAM filenames, one per line [null]\n");
fprintf(stderr, " -l <int> minQLen\n");
fprintf(stderr, " -q <int> base quality threshold\n");
fprintf(stderr, " -Q <int> mapping quality threshold\n");
fprintf(stderr, " -r <chr:from-to> region\n");
fprintf(stderr, "\n");
return 1;
}
// initialize the auxiliary data structures
if (file_list)
{
if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
n = nfiles;
argv = fn;
optind = 0;
}
else
n = argc - optind; // the number of BAMs on the command line
data = calloc(n, sizeof(void*)); // data[i] for the i-th input
beg = 0; end = 1<<30; tid = -1; // set the default region
for (i = 0; i < n; ++i) {
bam_header_t *htmp;
data[i] = calloc(1, sizeof(aux_t));
data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
data[i]->min_mapQ = mapQ; // set the mapQ filter
data[i]->min_len = min_len; // set the qlen filter
htmp = bam_header_read(data[i]->fp); // read the BAM header
if (i == 0) {
h = htmp; // keep the header of the 1st BAM
if (reg) bam_parse_region(h, reg, &tid, &beg, &end); // also parse the region
} else bam_header_destroy(htmp); // if not the 1st BAM, trash the header
if (tid >= 0) { // if a region is specified and parsed successfully
bam_index_t *idx = bam_index_load(argv[optind+i]); // load the index
data[i]->iter = bam_iter_query(idx, tid, beg, end); // set the iterator
bam_index_destroy(idx); // the index is not needed any more; phase out of the memory
}
}
// the core multi-pileup loop
mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
bam_mplp_set_maxcnt(mplp,2147483647); // set max_depth to int max
n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
if (pos < beg || pos >= end) continue; // out of range; skip
if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
for (i = 0; i < n; ++i) { // base level filters have to go here
int j, m = 0;
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
else if (bam1_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
}
printf("\t%d", n_plp[i] - m); // this the depth to output
}
putchar('\n');
}
free(n_plp); free(plp);
bam_mplp_destroy(mplp);
bam_header_destroy(h);
for (i = 0; i < n; ++i) {
bam_close(data[i]->fp);
if (data[i]->iter) bam_iter_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(reg);
if (bed) bed_destroy(bed);
if ( file_list )
{
for (i=0; i<n; i++) free(fn[i]);
free(fn);
}
return 0;
}
static int mpileup(mplp_conf_t *conf)
{
if (conf->nfiles == 0) {
fprintf(stderr,"[%s] no input file/data given\n", __func__);
exit(EXIT_FAILURE);
}
mplp_ref_t mp_ref = MPLP_REF_INIT;
conf->gplp = (mplp_pileup_t *) calloc(1,sizeof(mplp_pileup_t));
conf->mplp_data = (mplp_aux_t**) calloc(conf->nfiles, sizeof(mplp_aux_t*));
conf->plp = (const bam_pileup1_t**) calloc(conf->nfiles, sizeof(bam_pileup1_t*));
conf->n_plp = (int*) calloc(conf->nfiles, sizeof(int));
// Allow to run mpileup on multiple regions in one go. This comes at cost: the bai index
// must be kept in the memory for the whole time which can be a problem with many bams.
// Therefore if none or only one region is requested, we initialize the bam iterator as
// before and free the index. Only when multiple regions are queried, we keep the index.
int nregs = 0;
if ( conf->reg_fname )
{
if ( conf->reg_is_file )
{
conf->reg = regidx_init(conf->reg_fname,NULL,NULL,0,NULL);
if ( !conf->reg ) {
fprintf(stderr,"Could not parse the regions: %s\n", conf->reg_fname);
exit(EXIT_FAILURE);
}
}
else
{
conf->reg = regidx_init(NULL,regidx_parse_reg,NULL,sizeof(char*),NULL);
if ( regidx_insert_list(conf->reg,conf->reg_fname,',') !=0 ) {
fprintf(stderr,"Could not parse the regions: %s\n", conf->reg_fname);
exit(EXIT_FAILURE);
}
}
nregs = regidx_nregs(conf->reg);
conf->reg_itr = regitr_init(conf->reg);
regitr_loop(conf->reg_itr); // region iterator now positioned at the first region
}
// read the header of each file in the list and initialize data
// beware: mpileup has always assumed that tid's are consistent in the headers, add sanity check at least!
bam_hdr_t *hdr = NULL; // header of first file in input list
int i;
for (i = 0; i < conf->nfiles; ++i) {
bam_hdr_t *h_tmp;
conf->mplp_data[i] = (mplp_aux_t*) calloc(1, sizeof(mplp_aux_t));
conf->mplp_data[i]->fp = sam_open(conf->files[i], "rb");
if ( !conf->mplp_data[i]->fp )
{
fprintf(stderr, "[%s] failed to open %s: %s\n", __func__, conf->files[i], strerror(errno));
exit(EXIT_FAILURE);
}
if (hts_set_opt(conf->mplp_data[i]->fp, CRAM_OPT_DECODE_MD, 0)) {
fprintf(stderr, "Failed to set CRAM_OPT_DECODE_MD value\n");
exit(EXIT_FAILURE);
}
if (conf->fai_fname && hts_set_fai_filename(conf->mplp_data[i]->fp, conf->fai_fname) != 0) {
fprintf(stderr, "[%s] failed to process %s: %s\n",
__func__, conf->fai_fname, strerror(errno));
exit(EXIT_FAILURE);
}
conf->mplp_data[i]->conf = conf;
conf->mplp_data[i]->ref = &mp_ref;
h_tmp = sam_hdr_read(conf->mplp_data[i]->fp);
if ( !h_tmp ) {
fprintf(stderr,"[%s] fail to read the header of %s\n", __func__, conf->files[i]);
exit(EXIT_FAILURE);
}
conf->mplp_data[i]->h = i ? hdr : h_tmp; // for j==0, "h" has not been set yet
conf->mplp_data[i]->bam_id = bam_smpl_add_bam(conf->bsmpl,h_tmp->text,conf->files[i]);
if ( conf->mplp_data[i]->bam_id<0 )
{
// no usable readgroups in this bam, it can be skipped
sam_close(conf->mplp_data[i]->fp);
free(conf->mplp_data[i]);
bam_hdr_destroy(h_tmp);
free(conf->files[i]);
if ( i+1<conf->nfiles ) memmove(&conf->files[i],&conf->files[i+1],sizeof(*conf->files)*(conf->nfiles-i-1));
conf->nfiles--;
i--;
continue;
}
if (conf->reg) {
hts_idx_t *idx = sam_index_load(conf->mplp_data[i]->fp, conf->files[i]);
if (idx == NULL) {
fprintf(stderr, "[%s] fail to load index for %s\n", __func__, conf->files[i]);
exit(EXIT_FAILURE);
}
conf->buf.l = 0;
ksprintf(&conf->buf,"%s:%u-%u",conf->reg_itr->seq,conf->reg_itr->beg+1,conf->reg_itr->end+1);
conf->mplp_data[i]->iter = sam_itr_querys(idx, conf->mplp_data[i]->h, conf->buf.s);
if ( !conf->mplp_data[i]->iter )
{
conf->mplp_data[i]->iter = sam_itr_querys(idx, conf->mplp_data[i]->h, conf->reg_itr->seq);
if ( conf->mplp_data[i]->iter ) {
fprintf(stderr,"[E::%s] fail to parse region '%s'\n", __func__, conf->buf.s);
exit(EXIT_FAILURE);
}
fprintf(stderr,"[E::%s] the sequence \"%s\" not found: %s\n",__func__,conf->reg_itr->seq,conf->files[i]);
exit(EXIT_FAILURE);
}
if ( nregs==1 ) // no need to keep the index in memory
hts_idx_destroy(idx);
else
conf->mplp_data[i]->idx = idx;
}
if ( !hdr ) hdr = h_tmp; /* save the header of first file in list */
else {
// FIXME: check consistency between h and h_tmp
bam_hdr_destroy(h_tmp);
// we store only the first file's header; it's (alleged to be)
// compatible with the i-th file's target_name lookup needs
conf->mplp_data[i]->h = hdr;
}
}
// allocate data storage proportionate to number of samples being studied sm->n
bam_smpl_get_samples(conf->bsmpl, &conf->gplp->n);
conf->gplp->n_plp = (int*) calloc(conf->gplp->n, sizeof(int));
conf->gplp->m_plp = (int*) calloc(conf->gplp->n, sizeof(int));
conf->gplp->plp = (bam_pileup1_t**) calloc(conf->gplp->n, sizeof(bam_pileup1_t*));
fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, conf->gplp->n, conf->nfiles);
// write the VCF header
conf->bcf_fp = hts_open(conf->output_fname?conf->output_fname:"-", hts_bcf_wmode(conf->output_type));
if (conf->bcf_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname? conf->output_fname : "standard output", strerror(errno));
exit(EXIT_FAILURE);
}
if ( conf->n_threads ) hts_set_threads(conf->bcf_fp, conf->n_threads);
// BCF header creation
conf->bcf_hdr = bcf_hdr_init("w");
conf->buf.l = 0;
if (conf->record_cmd_line)
{
ksprintf(&conf->buf, "##bcftoolsVersion=%s+htslib-%s\n",bcftools_version(),hts_version());
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
conf->buf.l = 0;
ksprintf(&conf->buf, "##bcftoolsCommand=mpileup");
for (i=1; i<conf->argc; i++) ksprintf(&conf->buf, " %s", conf->argv[i]);
kputc('\n', &conf->buf);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
if (conf->fai_fname)
{
conf->buf.l = 0;
ksprintf(&conf->buf, "##reference=file://%s\n", conf->fai_fname);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
// Translate BAM @SQ tags to BCF ##contig tags
// todo: use/write new BAM header manipulation routines, fill also UR, M5
for (i=0; i<hdr->n_targets; i++)
{
conf->buf.l = 0;
ksprintf(&conf->buf, "##contig=<ID=%s,length=%d>", hdr->target_name[i], hdr->target_len[i]);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
conf->buf.l = 0;
bcf_hdr_append(conf->bcf_hdr,"##ALT=<ID=*,Description=\"Represents allele(s) other than observed.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=IDV,Number=1,Type=Integer,Description=\"Maximum number of reads supporting an indel\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=IMF,Number=1,Type=Float,Description=\"Maximum fraction of reads supporting an indel\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias for filtering splice-site artefacts in RNA-seq data (bigger is better)\",Version=\"3\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=RPB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=BQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQSB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias (bigger is better)\">");
#if CDF_MWU_TESTS
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=RPB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=BQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQSB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias [CDF] (bigger is better)\">");
#endif
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=SGB,Number=1,Type=Float,Description=\"Segregation based metric.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQ0F,Number=1,Type=Float,Description=\"Fraction of MQ0 reads (smaller is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=I16,Number=16,Type=Float,Description=\"Auxiliary tag used for calling, see description of bcf_callret1_t in bam2bcf.h\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=QS,Number=R,Type=Float,Description=\"Auxiliary tag used for calling\">");
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">");
if ( conf->fmt_flag&B2B_FMT_DP )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Number of high-quality bases\">");
if ( conf->fmt_flag&B2B_FMT_DV )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"Number of high-quality non-reference bases\">");
if ( conf->fmt_flag&B2B_FMT_DPR )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_INFO_DPR )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_FMT_DP4 )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DP4,Number=4,Type=Integer,Description=\"Number of high-quality ref-fwd, ref-reverse, alt-fwd and alt-reverse bases\">");
if ( conf->fmt_flag&B2B_FMT_SP )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">");
if ( conf->fmt_flag&B2B_FMT_AD )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=AD,Number=R,Type=Integer,Description=\"Allelic depths\">");
if ( conf->fmt_flag&B2B_FMT_ADF )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=ADF,Number=R,Type=Integer,Description=\"Allelic depths on the forward strand\">");
if ( conf->fmt_flag&B2B_FMT_ADR )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=ADR,Number=R,Type=Integer,Description=\"Allelic depths on the reverse strand\">");
if ( conf->fmt_flag&B2B_INFO_AD )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=AD,Number=R,Type=Integer,Description=\"Total allelic depths\">");
if ( conf->fmt_flag&B2B_INFO_ADF )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=ADF,Number=R,Type=Integer,Description=\"Total allelic depths on the forward strand\">");
if ( conf->fmt_flag&B2B_INFO_ADR )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=ADR,Number=R,Type=Integer,Description=\"Total allelic depths on the reverse strand\">");
if ( conf->gvcf )
gvcf_update_header(conf->gvcf, conf->bcf_hdr);
int nsmpl;
const char **smpl = bam_smpl_get_samples(conf->bsmpl, &nsmpl);
for (i=0; i<nsmpl; i++)
bcf_hdr_add_sample(conf->bcf_hdr, smpl[i]);
bcf_hdr_write(conf->bcf_fp, conf->bcf_hdr);
conf->bca = bcf_call_init(-1., conf->min_baseQ);
conf->bcr = (bcf_callret1_t*) calloc(nsmpl, sizeof(bcf_callret1_t));
conf->bca->openQ = conf->openQ, conf->bca->extQ = conf->extQ, conf->bca->tandemQ = conf->tandemQ;
conf->bca->min_frac = conf->min_frac;
conf->bca->min_support = conf->min_support;
conf->bca->per_sample_flt = conf->flag & MPLP_PER_SAMPLE;
conf->bc.bcf_hdr = conf->bcf_hdr;
conf->bc.n = nsmpl;
conf->bc.PL = (int32_t*) malloc(15 * nsmpl * sizeof(*conf->bc.PL));
if (conf->fmt_flag)
{
assert( sizeof(float)==sizeof(int32_t) );
conf->bc.DP4 = (int32_t*) malloc(nsmpl * sizeof(int32_t) * 4);
conf->bc.fmt_arr = (uint8_t*) malloc(nsmpl * sizeof(float)); // all fmt_flag fields, float and int32
if ( conf->fmt_flag&(B2B_INFO_DPR|B2B_FMT_DPR|B2B_INFO_AD|B2B_INFO_ADF|B2B_INFO_ADR|B2B_FMT_AD|B2B_FMT_ADF|B2B_FMT_ADR) )
{
// first B2B_MAX_ALLELES fields for total numbers, the rest per-sample
conf->bc.ADR = (int32_t*) malloc((nsmpl+1)*B2B_MAX_ALLELES*sizeof(int32_t));
conf->bc.ADF = (int32_t*) malloc((nsmpl+1)*B2B_MAX_ALLELES*sizeof(int32_t));
for (i=0; i<nsmpl; i++)
{
conf->bcr[i].ADR = conf->bc.ADR + (i+1)*B2B_MAX_ALLELES;
conf->bcr[i].ADF = conf->bc.ADF + (i+1)*B2B_MAX_ALLELES;
}
}
}
// init mpileup
conf->iter = bam_mplp_init(conf->nfiles, mplp_func, (void**)conf->mplp_data);
if ( conf->flag & MPLP_SMART_OVERLAPS ) bam_mplp_init_overlaps(conf->iter);
if ( (double)conf->max_depth * conf->nfiles > 1<<20)
fprintf(stderr, "Warning: Potential memory hog, up to %.0fM reads in the pileup!\n", (double)conf->max_depth*conf->nfiles);
if ( (double)conf->max_depth * conf->nfiles / nsmpl < 250 )
fprintf(stderr, "Note: The maximum per-sample depth with -d %d is %.1fx\n", conf->max_depth,(double)conf->max_depth * conf->nfiles / nsmpl);
bam_mplp_set_maxcnt(conf->iter, conf->max_depth);
conf->max_indel_depth = conf->max_indel_depth * nsmpl;
conf->bcf_rec = bcf_init1();
bam_mplp_constructor(conf->iter, pileup_constructor);
// Run mpileup for multiple regions
if ( nregs )
{
int ireg = 0;
do
{
// first region is already positioned
if ( ireg++ > 0 )
{
conf->buf.l = 0;
ksprintf(&conf->buf,"%s:%u-%u",conf->reg_itr->seq,conf->reg_itr->beg,conf->reg_itr->end);
for (i=0; i<conf->nfiles; i++)
{
hts_itr_destroy(conf->mplp_data[i]->iter);
conf->mplp_data[i]->iter = sam_itr_querys(conf->mplp_data[i]->idx, conf->mplp_data[i]->h, conf->buf.s);
if ( !conf->mplp_data[i]->iter )
{
conf->mplp_data[i]->iter = sam_itr_querys(conf->mplp_data[i]->idx, conf->mplp_data[i]->h, conf->reg_itr->seq);
if ( conf->mplp_data[i]->iter ) {
fprintf(stderr,"[E::%s] fail to parse region '%s'\n", __func__, conf->buf.s);
exit(EXIT_FAILURE);
}
fprintf(stderr,"[E::%s] the sequence \"%s\" not found: %s\n",__func__,conf->reg_itr->seq,conf->files[i]);
exit(EXIT_FAILURE);
}
bam_mplp_reset(conf->iter);
}
}
mpileup_reg(conf,conf->reg_itr->beg,conf->reg_itr->end);
}
while ( regitr_loop(conf->reg_itr) );
}
else
mpileup_reg(conf,0,0);
flush_bcf_records(conf, conf->bcf_fp, conf->bcf_hdr, NULL);
// clean up
free(conf->bc.tmp.s);
bcf_destroy1(conf->bcf_rec);
if (conf->bcf_fp)
{
hts_close(conf->bcf_fp);
bcf_hdr_destroy(conf->bcf_hdr);
bcf_call_destroy(conf->bca);
free(conf->bc.PL);
free(conf->bc.DP4);
free(conf->bc.ADR);
free(conf->bc.ADF);
free(conf->bc.fmt_arr);
free(conf->bcr);
}
if ( conf->gvcf ) gvcf_destroy(conf->gvcf);
free(conf->buf.s);
for (i = 0; i < conf->gplp->n; ++i) free(conf->gplp->plp[i]);
free(conf->gplp->plp); free(conf->gplp->n_plp); free(conf->gplp->m_plp); free(conf->gplp);
bam_mplp_destroy(conf->iter);
bam_hdr_destroy(hdr);
for (i = 0; i < conf->nfiles; ++i) {
if ( nregs>1 ) hts_idx_destroy(conf->mplp_data[i]->idx);
sam_close(conf->mplp_data[i]->fp);
if ( conf->mplp_data[i]->iter) hts_itr_destroy(conf->mplp_data[i]->iter);
free(conf->mplp_data[i]);
}
if ( conf->reg_itr ) regitr_destroy(conf->reg_itr);
free(conf->mplp_data); free(conf->plp); free(conf->n_plp);
free(mp_ref.ref[0]);
free(mp_ref.ref[1]);
return 0;
}
/*
* Performs pileup
* @param conf configuration for this pileup
* @param n number of files specified in fn
* @param fn filenames
*/
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, tid0 = -1, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid = -1, max_depth, max_indel_depth;
const bam_pileup1_t **plp;
bam_mplp_t iter;
bam_hdr_t *h = NULL; /* header of first file in input list */
char *ref;
void *rghash = NULL;
FILE *pileup_fp = NULL;
bcf_callaux_t *bca = NULL;
bcf_callret1_t *bcr = NULL;
bcf_call_t bc;
htsFile *bcf_fp = NULL;
bcf_hdr_t *bcf_hdr = NULL;
bam_sample_t *sm = NULL;
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(mplp_aux_t*));
plp = calloc(n, sizeof(bam_pileup1_t*));
n_plp = calloc(n, sizeof(int));
sm = bam_smpl_init();
if (n == 0) {
fprintf(stderr,"[%s] no input file/data given\n", __func__);
exit(1);
}
// read the header of each file in the list and initialize data
for (i = 0; i < n; ++i) {
bam_hdr_t *h_tmp;
data[i] = calloc(1, sizeof(mplp_aux_t));
data[i]->fp = sam_open(fn[i], "rb");
if ( !data[i]->fp )
{
fprintf(stderr, "[%s] failed to open %s: %s\n", __func__, fn[i], strerror(errno));
exit(1);
}
hts_set_fai_filename(data[i]->fp, conf->fai_fname);
data[i]->conf = conf;
h_tmp = sam_hdr_read(data[i]->fp);
if ( !h_tmp ) {
fprintf(stderr,"[%s] fail to read the header of %s\n", __func__, fn[i]);
exit(1);
}
data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
bam_smpl_add(sm, fn[i], (conf->flag&MPLP_IGNORE_RG)? 0 : h_tmp->text);
// Collect read group IDs with PL (platform) listed in pl_list (note: fragile, strstr search)
rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
if (conf->reg) {
hts_idx_t *idx = sam_index_load(data[i]->fp, fn[i]);
if (idx == 0) {
fprintf(stderr, "[%s] fail to load index for %s\n", __func__, fn[i]);
exit(1);
}
if ( (data[i]->iter=sam_itr_querys(idx, data[i]->h, conf->reg)) == 0) {
fprintf(stderr, "[E::%s] fail to parse region '%s'\n", __func__, conf->reg);
exit(1);
}
if (i == 0) tid0 = data[i]->iter->tid, beg0 = data[i]->iter->beg, end0 = data[i]->iter->end;
hts_idx_destroy(idx);
}
if (i == 0) h = h_tmp; /* save the header of first file in list */
else {
// FIXME: to check consistency
bam_hdr_destroy(h_tmp);
}
}
// allocate data storage proportionate to number of samples being studied sm->n
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(bam_pileup1_t*));
fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
// write the VCF header
if (conf->flag & MPLP_BCF)
{
const char *mode;
if ( conf->flag & MPLP_VCF )
mode = (conf->flag&MPLP_NO_COMP)? "wu" : "wz"; // uncompressed VCF or compressed VCF
else
mode = (conf->flag&MPLP_NO_COMP)? "wub" : "wb"; // uncompressed BCF or compressed BCF
bcf_fp = bcf_open(conf->output_fname? conf->output_fname : "-", mode);
if (bcf_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname? conf->output_fname : "standard output", strerror(errno));
exit(1);
}
bcf_hdr = bcf_hdr_init("w");
kstring_t str = {0,0,0};
ksprintf(&str, "##samtoolsVersion=%s+htslib-%s\n",samtools_version(),hts_version());
bcf_hdr_append(bcf_hdr, str.s);
str.l = 0;
ksprintf(&str, "##samtoolsCommand=samtools mpileup");
for (i=1; i<conf->argc; i++) ksprintf(&str, " %s", conf->argv[i]);
kputc('\n', &str);
bcf_hdr_append(bcf_hdr, str.s);
if (conf->fai_fname)
{
str.l = 0;
ksprintf(&str, "##reference=file://%s\n", conf->fai_fname);
bcf_hdr_append(bcf_hdr, str.s);
}
// todo: use/write new BAM header manipulation routines, fill also UR, M5
for (i=0; i<h->n_targets; i++)
{
str.l = 0;
ksprintf(&str, "##contig=<ID=%s,length=%d>", h->target_name[i], h->target_len[i]);
bcf_hdr_append(bcf_hdr, str.s);
}
free(str.s);
bcf_hdr_append(bcf_hdr,"##ALT=<ID=X,Description=\"Represents allele(s) other than observed.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=IDV,Number=1,Type=Integer,Description=\"Maximum number of reads supporting an indel\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=IMF,Number=1,Type=Float,Description=\"Maximum fraction of reads supporting an indel\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias for filtering splice-site artefacts in RNA-seq data (bigger is better)\",Version=\"3\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=RPB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=BQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQSB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias (bigger is better)\">");
#if CDF_MWU_TESTS
bcf_hdr_append(bcf_hdr,"##INFO=<ID=RPB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=BQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQSB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias [CDF] (bigger is better)\">");
#endif
bcf_hdr_append(bcf_hdr,"##INFO=<ID=SGB,Number=1,Type=Float,Description=\"Segregation based metric.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQ0F,Number=1,Type=Float,Description=\"Fraction of MQ0 reads (smaller is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=I16,Number=16,Type=Float,Description=\"Auxiliary tag used for calling, see description of bcf_callret1_t in bam2bcf.h\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=QS,Number=R,Type=Float,Description=\"Auxiliary tag used for calling\">");
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">");
if ( conf->fmt_flag&B2B_FMT_DP )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Number of high-quality bases\">");
if ( conf->fmt_flag&B2B_FMT_DV )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"Number of high-quality non-reference bases\">");
if ( conf->fmt_flag&B2B_FMT_DPR )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_INFO_DPR )
bcf_hdr_append(bcf_hdr,"##INFO=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_FMT_DP4 )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DP4,Number=4,Type=Integer,Description=\"Number of high-quality ref-fwd, ref-reverse, alt-fwd and alt-reverse bases\">");
if ( conf->fmt_flag&B2B_FMT_SP )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">");
for (i=0; i<sm->n; i++)
bcf_hdr_add_sample(bcf_hdr, sm->smpl[i]);
bcf_hdr_add_sample(bcf_hdr, NULL);
bcf_hdr_write(bcf_fp, bcf_hdr);
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;
bca->per_sample_flt = conf->flag & MPLP_PER_SAMPLE;
bc.bcf_hdr = bcf_hdr;
bc.n = sm->n;
bc.PL = malloc(15 * sm->n * sizeof(*bc.PL));
if (conf->fmt_flag)
{
assert( sizeof(float)==sizeof(int32_t) );
bc.DP4 = malloc(sm->n * sizeof(int32_t) * 4);
bc.fmt_arr = malloc(sm->n * sizeof(float)); // all fmt_flag fields
if ( conf->fmt_flag&(B2B_INFO_DPR|B2B_FMT_DPR) )
{
// first B2B_MAX_ALLELES fields for total numbers, the rest per-sample
bc.DPR = malloc((sm->n+1)*B2B_MAX_ALLELES*sizeof(int32_t));
for (i=0; i<sm->n; i++)
bcr[i].DPR = bc.DPR + (i+1)*B2B_MAX_ALLELES;
}
}
}
else {
pileup_fp = conf->output_fname? fopen(conf->output_fname, "w") : stdout;
if (pileup_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname, strerror(errno));
exit(1);
}
}
if (tid0 >= 0 && conf->fai) { // region is set
ref = faidx_fetch_seq(conf->fai, h->target_name[tid0], 0, 0x7fffffff, &ref_len);
ref_tid = tid0;
for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid0;
} else ref_tid = -1, ref = 0;
// begin pileup
iter = bam_mplp_init(n, mplp_func, (void**)data);
if ( conf->flag & MPLP_SMART_OVERLAPS ) bam_mplp_init_overlaps(iter);
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-file depth to %d\n", __func__, max_depth);
}
max_indel_depth = conf->max_indel_depth * sm->n;
bam_mplp_set_maxcnt(iter, max_depth);
bcf1_t *bcf_rec = bcf_init1();
int ret;
while ( (ret=bam_mplp_auto(iter, &tid, &pos, n_plp, plp)) > 0) {
if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
if (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
if (tid != ref_tid) {
free(ref); ref = 0;
if (conf->fai) ref = faidx_fetch_seq(conf->fai, h->target_name[tid], 0, 0x7fffffff, &ref_len);
for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
ref_tid = tid;
}
if (conf->flag & MPLP_BCF) {
int total_depth, _ref0, ref16;
for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp, conf->flag & MPLP_IGNORE_RG);
_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
ref16 = seq_nt16_table[_ref0];
bcf_callaux_clean(bca, &bc);
for (i = 0; i < gplp.n; ++i)
bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
bc.tid = tid; bc.pos = pos;
bcf_call_combine(gplp.n, bcr, bca, ref16, &bc);
bcf_clear1(bcf_rec);
bcf_call2bcf(&bc, bcf_rec, bcr, conf->fmt_flag, 0, 0);
bcf_write1(bcf_fp, bcf_hdr, bcf_rec);
// call indels; todo: subsampling with total_depth>max_indel_depth instead of ignoring?
if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0)
{
bcf_callaux_clean(bca, &bc);
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, bca, -1, &bc) >= 0) {
bcf_clear1(bcf_rec);
bcf_call2bcf(&bc, bcf_rec, bcr, conf->fmt_flag, bca, ref);
bcf_write1(bcf_fp, bcf_hdr, bcf_rec);
}
}
} else {
fprintf(pileup_fp, "%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
for (i = 0; i < n; ++i) {
int j, cnt;
for (j = cnt = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
if (bam_get_qual(p->b)[p->qpos] >= conf->min_baseQ) ++cnt;
}
fprintf(pileup_fp, "\t%d\t", cnt);
if (n_plp[i] == 0) {
fputs("*\t*", pileup_fp);
if (conf->flag & MPLP_PRINT_MAPQ) fputs("\t*", pileup_fp);
if (conf->flag & MPLP_PRINT_POS) fputs("\t*", pileup_fp);
} else {
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
if (bam_get_qual(p->b)[p->qpos] >= conf->min_baseQ)
pileup_seq(pileup_fp, plp[i] + j, pos, ref_len, ref);
}
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
int c = bam_get_qual(p->b)[p->qpos];
if (c >= conf->min_baseQ) {
c = c + 33 < 126? c + 33 : 126;
putc(c, pileup_fp);
}
}
if (conf->flag & MPLP_PRINT_MAPQ) {
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
int c = bam_get_qual(p->b)[p->qpos];
if ( c < conf->min_baseQ ) continue;
c = plp[i][j].b->core.qual + 33;
if (c > 126) c = 126;
putc(c, pileup_fp);
}
}
if (conf->flag & MPLP_PRINT_POS) {
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
if (j > 0) putc(',', pileup_fp);
fprintf(pileup_fp, "%d", plp[i][j].qpos + 1); // FIXME: printf() is very slow...
}
}
}
}
putc('\n', pileup_fp);
}
}
// clean up
free(bc.tmp.s);
bcf_destroy1(bcf_rec);
if (bcf_fp)
{
hts_close(bcf_fp);
bcf_hdr_destroy(bcf_hdr);
bcf_call_destroy(bca);
free(bc.PL);
free(bc.DP4);
free(bc.DPR);
free(bc.fmt_arr);
free(bcr);
}
if (pileup_fp && conf->output_fname) fclose(pileup_fp);
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);
bam_mplp_destroy(iter);
bam_hdr_destroy(h);
for (i = 0; i < n; ++i) {
sam_close(data[i]->fp);
if (data[i]->iter) hts_itr_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(plp); free(ref); free(n_plp);
return ret;
}
int main_bedcov(int argc, char *argv[])
{
gzFile fp;
kstring_t str;
kstream_t *ks;
hts_idx_t **idx;
aux_t **aux;
int *n_plp, dret, i, n, c, min_mapQ = 0;
int64_t *cnt;
const bam_pileup1_t **plp;
while ((c = getopt(argc, argv, "Q:")) >= 0) {
switch (c) {
case 'Q': min_mapQ = atoi(optarg); break;
}
}
if (optind + 2 > argc) {
fprintf(stderr, "Usage: samtools bedcov <in.bed> <in1.bam> [...]\n");
return 1;
}
memset(&str, 0, sizeof(kstring_t));
n = argc - optind - 1;
aux = calloc(n, sizeof(aux_t*));
idx = calloc(n, sizeof(hts_idx_t*));
for (i = 0; i < n; ++i) {
aux[i] = calloc(1, sizeof(aux_t));
aux[i]->min_mapQ = min_mapQ;
aux[i]->fp = sam_open(argv[i+optind+1], "r");
idx[i] = sam_index_load(aux[i]->fp, argv[i+optind+1]);
if (aux[i]->fp == 0 || idx[i] == 0) {
fprintf(stderr, "ERROR: fail to open index BAM file '%s'\n", argv[i+optind+1]);
return 2;
}
// TODO bgzf_set_cache_size(aux[i]->fp, 20);
aux[i]->header = sam_hdr_read(aux[i]->fp);
}
cnt = calloc(n, 8);
fp = gzopen(argv[optind], "rb");
ks = ks_init(fp);
n_plp = calloc(n, sizeof(int));
plp = calloc(n, sizeof(bam_pileup1_t*));
while (ks_getuntil(ks, KS_SEP_LINE, &str, &dret) >= 0) {
char *p, *q;
int tid, beg, end, pos;
bam_mplp_t mplp;
for (p = q = str.s; *p && *p != '\t'; ++p);
if (*p != '\t') goto bed_error;
*p = 0; tid = bam_name2id(aux[0]->header, q); *p = '\t';
if (tid < 0) goto bed_error;
for (q = p = p + 1; isdigit(*p); ++p);
if (*p != '\t') goto bed_error;
*p = 0; beg = atoi(q); *p = '\t';
for (q = p = p + 1; isdigit(*p); ++p);
if (*p == '\t' || *p == 0) {
int c = *p;
*p = 0; end = atoi(q); *p = c;
} else goto bed_error;
for (i = 0; i < n; ++i) {
if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
aux[i]->iter = sam_itr_queryi(idx[i], tid, beg, end);
}
mplp = bam_mplp_init(n, read_bam, (void**)aux);
bam_mplp_set_maxcnt(mplp, 64000);
memset(cnt, 0, 8 * n);
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0)
if (pos >= beg && pos < end)
for (i = 0; i < n; ++i) cnt[i] += n_plp[i];
for (i = 0; i < n; ++i) {
kputc('\t', &str);
kputl(cnt[i], &str);
}
puts(str.s);
bam_mplp_destroy(mplp);
continue;
bed_error:
fprintf(stderr, "Errors in BED line '%s'\n", str.s);
}
free(n_plp); free(plp);
ks_destroy(ks);
gzclose(fp);
free(cnt);
for (i = 0; i < n; ++i) {
if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
hts_idx_destroy(idx[i]);
bam_hdr_destroy(aux[i]->header);
sam_close(aux[i]->fp);
free(aux[i]);
}
free(aux); free(idx);
free(str.s);
return 0;
}
int main_depth(int argc, char *argv[])
{
int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0, min_len = 0, status = EXIT_SUCCESS, nfiles;
const bam_pileup1_t **plp;
char *reg = 0; // specified region
void *bed = 0; // BED data structure
char *file_list = NULL, **fn = NULL;
bam_hdr_t *h = NULL; // BAM header of the 1st input
aux_t **data;
bam_mplp_t mplp;
// parse the command line
while ((n = getopt(argc, argv, "r:b:q:Q:l:f:")) >= 0) {
switch (n) {
case 'l': min_len = atoi(optarg); break; // minimum query length
case 'r': reg = strdup(optarg); break; // parsing a region requires a BAM header
case 'b':
bed = bed_read(optarg); // BED or position list file can be parsed now
if (!bed) { print_error_errno("Could not read file \"%s\"", optarg); return 1; }
break;
case 'q': baseQ = atoi(optarg); break; // base quality threshold
case 'Q': mapQ = atoi(optarg); break; // mapping quality threshold
case 'f': file_list = optarg; break;
}
}
if (optind == argc && !file_list) {
fprintf(pysamerr, "\n");
fprintf(pysamerr, "Usage: samtools depth [options] in1.bam [in2.bam [...]]\n");
fprintf(pysamerr, "Options:\n");
fprintf(pysamerr, " -b <bed> list of positions or regions\n");
fprintf(pysamerr, " -f <list> list of input BAM filenames, one per line [null]\n");
fprintf(pysamerr, " -l <int> read length threshold (ignore reads shorter than <int>)\n");
fprintf(pysamerr, " -q <int> base quality threshold\n");
fprintf(pysamerr, " -Q <int> mapping quality threshold\n");
fprintf(pysamerr, " -r <chr:from-to> region\n");
fprintf(pysamerr, "\n");
return 1;
}
// initialize the auxiliary data structures
if (file_list)
{
if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
n = nfiles;
argv = fn;
optind = 0;
}
else
n = argc - optind; // the number of BAMs on the command line
data = calloc(n, sizeof(aux_t*)); // data[i] for the i-th input
beg = 0; end = 1<<30; // set the default region
for (i = 0; i < n; ++i) {
data[i] = calloc(1, sizeof(aux_t));
data[i]->fp = sam_open(argv[optind+i], "r"); // open BAM
if (data[i]->fp == NULL) {
print_error_errno("Could not open \"%s\"", argv[optind+i]);
status = EXIT_FAILURE;
goto depth_end;
}
if (hts_set_opt(data[i]->fp, CRAM_OPT_REQUIRED_FIELDS,
SAM_FLAG | SAM_RNAME | SAM_POS | SAM_MAPQ | SAM_CIGAR |
SAM_SEQ)) {
fprintf(pysamerr, "Failed to set CRAM_OPT_REQUIRED_FIELDS value\n");
return 1;
}
if (hts_set_opt(data[i]->fp, CRAM_OPT_DECODE_MD, 0)) {
fprintf(pysamerr, "Failed to set CRAM_OPT_DECODE_MD value\n");
return 1;
}
data[i]->min_mapQ = mapQ; // set the mapQ filter
data[i]->min_len = min_len; // set the qlen filter
data[i]->hdr = sam_hdr_read(data[i]->fp); // read the BAM header
if (reg) { // if a region is specified
hts_idx_t *idx = sam_index_load(data[i]->fp, argv[optind+i]); // load the index
if (idx == NULL) {
print_error("can't load index for \"%s\"", argv[optind+i]);
status = EXIT_FAILURE;
goto depth_end;
}
data[i]->iter = sam_itr_querys(idx, data[i]->hdr, reg); // set the iterator
hts_idx_destroy(idx); // the index is not needed any more; free the memory
if (data[i]->iter == NULL) {
print_error("can't parse region \"%s\"", reg);
status = EXIT_FAILURE;
goto depth_end;
}
}
}
h = data[0]->hdr; // easy access to the header of the 1st BAM
if (reg) {
beg = data[0]->iter->beg; // and to the parsed region coordinates
end = data[0]->iter->end;
}
// the core multi-pileup loop
mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = calloc(n, sizeof(bam_pileup1_t*)); // plp[i] points to the array of covering reads (internal in mplp)
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
if (pos < beg || pos >= end) continue; // out of range; skip
if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
for (i = 0; i < n; ++i) { // base level filters have to go here
int j, m = 0;
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
else if (bam_get_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
}
printf("\t%d", n_plp[i] - m); // this the depth to output
}
putchar('\n');
}
free(n_plp); free(plp);
bam_mplp_destroy(mplp);
depth_end:
for (i = 0; i < n && data[i]; ++i) {
bam_hdr_destroy(data[i]->hdr);
if (data[i]->fp) sam_close(data[i]->fp);
hts_itr_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(reg);
if (bed) bed_destroy(bed);
if ( file_list )
{
for (i=0; i<n; i++) free(fn[i]);
free(fn);
}
return status;
}
int bam2depth(const std::string& chromosomeName, const int startPos, const int endPos, const int minBaseQuality, const int minMappingQuality, const std::vector <std::string> & listOfFiles,
std::vector< double > & averageCoveragePerBam
)
{
int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0;
const bam_pileup1_t **plp;
char *reg = 0; // specified region
//void *bed = 0; // BED data structure
aux_t **data;
bam_mplp_t mplp;
// initialize the auxiliary data structures
n = listOfFiles.size(); // the number of BAMs on the command line
data = (aux_t **)calloc(n, sizeof(void*)); // data[i] for the i-th input
beg = 0; end = 1<<30; tid = -1; // set the default region
beg = startPos;
end = endPos;
for (i = 0; i < n; ++i) {
data[i] = (aux_t*)calloc(1, sizeof(aux_t));
data[i]->fp = sam_open(listOfFiles[i].c_str(), "r"); // open BAM
data[i]->min_mapQ = mapQ; // set the mapQ filter
data[i]->hdr = sam_hdr_read(data[i]->fp); // read the BAM header
tid = bam_name2id(data[i]->hdr, chromosomeName.c_str());
if (tid >= 0) { // if a region is specified and parsed successfully
hts_idx_t *idx = sam_index_load(data[i]->fp, listOfFiles[i].c_str()); // load the index
data[i]->iter = sam_itr_queryi(idx, tid, beg, end); // set the iterator
hts_idx_destroy(idx); // the index is not needed any more; phase out of the memory
}
}
// the core multi-pileup loop
mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
n_plp = (int*)calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = (const bam_pileup1_t **)calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
std::vector<int> sumOfReadDepths( n, 0);
while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
if (pos < beg || pos >= end) continue; // out of range; skip
for (i = 0; i < n; ++i) { // base level filters have to go here
int j, m = 0;
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
else if (bam_get_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
}
sumOfReadDepths[ i ] += n_plp[i] - m;
}
}
averageCoveragePerBam.resize( n );
for (int fileIndex=0; fileIndex< n; fileIndex++ ) {
averageCoveragePerBam[ fileIndex ] = (double)sumOfReadDepths[ fileIndex ] / (end - beg );
}
free(n_plp); free(plp);
bam_mplp_destroy(mplp);
for (i = 0; i < n; ++i) {
bam_hdr_destroy(data[i]->hdr);
sam_close(data[i]->fp);
if (data[i]->iter) hts_itr_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(reg);
return 0;
}
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, tid0 = -1, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid = -1, max_depth, max_indel_depth;
const bam_pileup1_t **plp;
bam_mplp_t iter;
bam_header_t *h = 0;
char *ref;
void *rghash = 0;
bcf_callaux_t *bca = 0;
bcf_callret1_t *bcr = 0;
bcf_call_t bc;
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]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
data[i]->conf = conf;
h_tmp = bam_header_read(data[i]->fp);
data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
bam_smpl_add(sm, fn[i], (conf->flag&MPLP_IGNORE_RG)? 0 : 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;
idx = bam_index_load(fn[i]);
if (idx == 0) {
fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
exit(1);
}
if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
exit(1);
}
if (i == 0) tid0 = tid, beg0 = beg, end0 = end;
data[i]->iter = bam_iter_query(idx, tid, beg, end);
bam_index_destroy(idx);
}
if (i == 0) h = h_tmp;
else {
// FIXME: to check consistency
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);
// write the VCF header
if (conf->flag & MPLP_GLF) {
kstring_t s;
bh = calloc(1, sizeof(bcf_hdr_t));
s.l = s.m = 0; s.s = 0;
bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
for (i = 0; i < h->n_targets; ++i) {
kputs(h->target_name[i], &s);
kputc('\0', &s);
}
bh->l_nm = s.l;
bh->name = malloc(s.l);
memcpy(bh->name, s.s, s.l);
s.l = 0;
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);
memcpy(bh->sname, s.s, s.l);
bh->txt = malloc(strlen(BAM_VERSION) + 64);
bh->l_txt = 1 + sprintf(bh->txt, "##samtoolsVersion=%s\n", BAM_VERSION);
free(s.s);
bcf_hdr_sync(bh);
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;
}
if (tid0 >= 0 && conf->fai) { // region is set
ref = faidx_fetch_seq(conf->fai, h->target_name[tid0], 0, 0x7fffffff, &ref_len);
ref_tid = tid0;
for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid0;
} else 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-file depth to %d\n", __func__, max_depth);
}
max_indel_depth = conf->max_indel_depth * sm->n;
bam_mplp_set_maxcnt(iter, max_depth);
int storeSize = 100;
int delStore[2][100] = {{0},{0}};
typedef char * mstring;
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 (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
if (tid != ref_tid) {
free(ref); ref = 0;
if (conf->fai) ref = faidx_fetch_seq(conf->fai, h->target_name[tid], 0, 0x7fffffff, &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 total_depth, _ref0, ref16;
bcf1_t *b = calloc(1, sizeof(bcf1_t));
for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp, conf->flag & MPLP_IGNORE_RG);
_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
ref16 = bam_nt16_table[_ref0];
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), 0, 0);
bcf_write(bp, bh, b);
bcf_destroy(b);
// call indels
if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && 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 {
printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
for (i = 0; i < n; ++i) {
int j;
printf("\t%d\t", n_plp[i]);
if (n_plp[i] == 0) {
printf("*\t*"); // FIXME: printf() is very slow...
if (conf->flag & MPLP_PRINT_POS) printf("\t*");
} else {
//MDW start
//for each position in the pileup column
int charLen = 16;
int countChars[ charLen ][2];
int countiChars[ charLen ][2];
int countGap[2]={0,0};
//double qvTotal=0;
int numStruck=0;
int numGood=0;
int tti;
int ttj;
mstring insAllele[100];
int insAlleleCnt[100];
int sf=0;
int flag=0;
//typedef char * string;
char insStr0[10000];
int iCnt0=0;
char insStr1[10000];
int iCnt1=0;
char delStr0[10000];
int dCnt0=0;
char delStr1[10000];
int dCnt1=0;
float qposP[10000];
int qposCnt=0;
//initialize with zeros
for(tti=0;tti<charLen;tti++){
countChars[tti][0]=0;
countChars[tti][1]=0;
}
// define repeat length here; look back up to 10 prior positions
// start one position away.
int replC=0; //
for(tti=1;tti<=15;tti++){
// check for greater than zero
if(toupper(ref[pos-1])==toupper(ref[pos-tti])){
replC++;
}else{ // breaks the chain at first non identical to current position not strict homopolymer
break;
}
}
int reprC=0; //
for(tti=1;tti<=15;tti++){
// check for greater than zero
if(toupper(ref[pos+1])==toupper(ref[pos+tti])){
reprC++;
}else{ // breaks the chain at first non identical to current position not strict homopolymer
break;
}
}
int repT = replC;
if(replC < reprC){
repT=reprC;
}
for (j = 0; j < n_plp[i]; ++j){
const bam_pileup1_t *p = plp[i] + j;
/*
SAME LOGIC AS pileup_seq()
*/
if(p->is_refskip){ // never count intron gaps in numStruck
continue;
}
if(p->is_del){ // skip deletion gap, after first position which is the first aligned char
continue;
}
if( p->b->core.qual < conf->min_mqToCount || // mapping quality
conf->maxrepC < (repT) || // max homopolymer run, this will not
(!p->is_del && bam1_qual(p->b)[p->qpos] < conf->min_baseQ) || // base quality for matches
p->alignedQPosBeg <= (conf->trimEnd ) || p->alignedQPosEnd <= (conf->trimEnd ) || // trimEnd is 1-based
p->zf == 1 || // fusion tag
p->ih > conf->maxIH || // max hit index
(p->nmd > conf->maxNM) || // max mismatch
(conf->flagFilter == 1 && !(p->b->core.flag&BAM_FPROPER_PAIR)) || // optionally keep only proper pairs
(conf->flagFilter == 2 && p->b->core.flag&BAM_FSECONDARY) || // optionally strike secondary
(conf->flagFilter == 3 && p->b->core.flag&BAM_FDUP) || // optionally strike dup
(conf->flagFilter == 4 && (p->b->core.flag&BAM_FDUP || p->b->core.flag&BAM_FSECONDARY)) || // optionally strike secondary or dup
(conf->flagFilter == 5 && (p->b->core.flag&BAM_FDUP || p->b->core.flag&BAM_FSECONDARY || p->b->core.flag&BAM_FQCFAIL || !(p->b->core.flag&BAM_FPROPER_PAIR) )) // optionally strike secondary, dup and QCfail
){
numStruck++;
continue;
}
//printf("repT=%d: %d %c %c %c %c \n",repT,p->indel,ref[pos],ref[pos-1],ref[pos-2],ref[pos-3]);
if(!p->is_del && p->indel==0){
countChars[ bam1_seqi(bam1_seq(p->b), p->qpos) ][ bam1_strand(p->b) ] ++;
numGood++;
}else if(p->is_refskip){
countGap[ bam1_strand(p->b) ]++;
}
if(p->indel<0){
numGood++;
if(bam1_strand(p->b) ==0){
for(tti=1;tti<= -p->indel; tti++) {
// current spot, starting at 0 in store, because indel<0 refers to next position
delStr0[dCnt0] = ref[pos+tti];
dCnt0++;
}
delStr0[dCnt0] = ',';
dCnt0++;
}else{
for(tti=1;tti<= -p->indel; tti++) {
// current spot, starting at 0 in store, because indel<0 refers to next position
delStr1[dCnt1] = ref[pos+tti];
dCnt1++;
}
delStr1[dCnt1] = ',';
dCnt1++;
}
}else if(p->indel>0){
numGood++;
if(bam1_strand(p->b) ==0){
for(tti=1;tti<= p->indel; tti++) {
// current spot, starting at 0 in store, because indel<0 refers to next position
insStr0[iCnt0] = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + tti)];
iCnt0++;
}
insStr0[iCnt0] = ',';
iCnt0++;
}else{
for(tti=1;tti<= p->indel; tti++) {
// current spot, starting at 0 in store, because indel<0 refers to next position
insStr1[iCnt1] = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + tti)];
iCnt1++;
}
insStr1[iCnt1] = ',';
iCnt1++;
}
}
//calculate position of variant within aligned read - no soft clips
if( toupper(ref[pos]) != toupper(bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)]) || p->indel>0 || p->indel<0 ){
//distance to end; calculate distance to end of aligned read. removes soft clips.
int distToEnd = (p->alignedQPosBeg < p->alignedQPosEnd) ? p->alignedQPosBeg : p->alignedQPosEnd;
qposP[qposCnt] = distToEnd;
qposCnt++;
// printf("id=%s, pos=%d",bam1_qname(p->b),distToEnd);
}
}
//
//print A,C,G,T, by +/-
printf("\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d", countChars[1][0],countChars[1][1],
countChars[2][0],countChars[2][1],
countChars[4][0],countChars[4][1],
countChars[8][0],countChars[8][1],
countChars[7][0],countChars[7][1]);
putchar('\t');
for(tti=0;tti<dCnt0;tti++){
putchar(delStr0[tti]);
}
putchar('\t');
for(tti=0;tti<dCnt1;tti++){
putchar(delStr1[tti]);
}
putchar('\t');
for(tti=0;tti<iCnt0;tti++){
putchar(insStr0[tti]);
}
putchar('\t');
for(tti=0;tti<iCnt1;tti++){
putchar(insStr1[tti]);
}
printf("\t%d\t%d",numGood,numStruck);
// get non-ref qpos variation
float medqpos = -1;
float medAbsDev = -1;
if(qposCnt>0){
medqpos = median(qposCnt,qposP);
float absDev[qposCnt];
for(tti=0;tti<qposCnt;tti++){
absDev[tti] = abs(medqpos - qposP[tti]);
}
medAbsDev = median(qposCnt-1,absDev);
}
printf("\t%f",medAbsDev);
///END MDW
}
}
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);
bcf_call_del_rghash(rghash);
bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
bam_mplp_destroy(iter);
bam_header_destroy(h);
for (i = 0; i < n; ++i) {
bam_close(data[i]->fp);
if (data[i]->iter) bam_iter_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(plp); free(ref); free(n_plp);
return 0;
}
void extractCalls(Config *config) {
bam_hdr_t *hdr = sam_hdr_read(config->fp);
bam_mplp_t iter;
int ret, tid, pos, i, seqlen, type, rv, o = 0;
int beg0 = 0, end0 = 1u<<29;
int n_plp; //This will need to be modified for multiple input files
int ctid = -1; //The tid of the contig whose sequence is stored in "seq"
int idxBED = 0, strand;
uint32_t nmethyl = 0, nunmethyl = 0;
const bam_pileup1_t **plp = NULL;
char *seq = NULL, base;
mplp_data *data = NULL;
struct lastCall *lastCpG = NULL;
struct lastCall *lastCHG = NULL;
if(config->merge) {
if(config->keepCpG) {
lastCpG = calloc(1, sizeof(struct lastCall));
assert(lastCpG);
lastCpG->tid = -1;
}
if(config->keepCHG) {
lastCHG = calloc(1, sizeof(struct lastCall));
assert(lastCHG);
lastCHG->tid = -1;
}
}
data = calloc(1,sizeof(mplp_data));
if(data == NULL) {
fprintf(stderr, "Couldn't allocate space for the data structure in extractCalls()!\n");
return;
}
data->config = config;
data->hdr = hdr;
if (config->reg) {
if((data->iter = sam_itr_querys(config->bai, hdr, config->reg)) == 0) {
fprintf(stderr, "failed to parse regions %s", config->reg);
return;
}
}
if(config->bedName) {
config->bed = parseBED(config->bedName, hdr);
if(config->bed == NULL) return;
}
plp = calloc(1, sizeof(bam_pileup1_t *)); //This will have to be modified for multiple input files
if(plp == NULL) {
fprintf(stderr, "Couldn't allocate space for the plp structure in extractCalls()!\n");
return;
}
//Start the pileup
iter = bam_mplp_init(1, filter_func, (void **) &data);
bam_mplp_init_overlaps(iter);
bam_mplp_set_maxcnt(iter, config->maxDepth);
while((ret = cust_mplp_auto(iter, &tid, &pos, &n_plp, plp)) > 0) {
//Do we need to process this position?
if (config->reg) {
beg0 = data->iter->beg, end0 = data->iter->end;
if ((pos < beg0 || pos >= end0)) continue; // out of the region requested
}
if(tid != ctid) {
if(seq != NULL) free(seq);
seq = faidx_fetch_seq(config->fai, hdr->target_name[tid], 0, faidx_seq_len(config->fai, hdr->target_name[tid]), &seqlen);
if(seqlen < 0) {
fprintf(stderr, "faidx_fetch_seq returned %i while trying to fetch the sequence for tid %i (%s)!\n",\
seqlen, tid, hdr->target_name[tid]);
fprintf(stderr, "Note that the output will be truncated!\n");
return;
}
ctid = tid;
}
if(config->bed) { //Handle -l
while((o = posOverlapsBED(tid, pos, config->bed, idxBED)) == -1) idxBED++;
if(o == 0) continue; //Wrong strand
}
if(isCpG(seq, pos, seqlen)) {
if(!config->keepCpG) continue;
type = 0;
} else if(isCHG(seq, pos, seqlen)) {
if(!config->keepCHG) continue;
type = 1;
} else if(isCHH(seq, pos, seqlen)) {
if(!config->keepCHH) continue;
type = 2;
} else {
continue;
}
nmethyl = nunmethyl = 0;
base = *(seq+pos);
for(i=0; i<n_plp; i++) {
if(plp[0][i].is_del) continue;
if(plp[0][i].is_refskip) continue;
if(config->bed) if(!readStrandOverlapsBED(plp[0][i].b, config->bed->region[idxBED])) continue;
strand = getStrand((plp[0]+i)->b);
if(strand & 1) {
if(base != 'C' && base != 'c') continue;
} else {
if(base != 'G' && base != 'g') continue;
}
rv = updateMetrics(config, plp[0]+i);
if(rv > 0) nmethyl++;
else if(rv<0) nunmethyl++;
}
if(nmethyl+nunmethyl==0) continue;
if(!config->merge || type==2) {
writeCall(config->output_fp[type], config, hdr->target_name[tid], pos, 1, nmethyl, nunmethyl);
} else {
//Merge into per-CpG/CHG metrics
if(type==0) {
if(base=='G' || base=='g') pos--;
processLast(config->output_fp[0], config, lastCpG, hdr, tid, pos, 2, nmethyl, nunmethyl);
} else {
if(base=='G' || base=='g') pos-=2;
processLast(config->output_fp[1], config, lastCHG, hdr, tid, pos, 3, nmethyl, nunmethyl);
}
}
}
//Don't forget the last CpG/CHG
if(config->merge) {
if(config->keepCpG && lastCpG->tid != -1) {
processLast(config->output_fp[0], config, lastCpG, hdr, tid, pos, 2, nmethyl, nunmethyl);
}
if(config->keepCHG && lastCHG->tid != -1) {
processLast(config->output_fp[1], config, lastCHG, hdr, tid, pos, 3, nmethyl, nunmethyl);
}
}
bam_hdr_destroy(hdr);
if(data->iter) hts_itr_destroy(data->iter);
bam_mplp_destroy(iter);
free(data);
free(plp);
if(seq != NULL) free(seq);
}
int main(int argc, char *argv[])
{
int c, i, n, ret, res;
int tid, pos, *n_plp;
cmdopt_t o;
bam_mplp_t mplp;
const bam_pileup1_t **plp;
aux_t **data;
bam_hdr_t *h = 0;
sv_t sv1;
qual_sum_t qual2;
khiter_t k_iter;
khash_t(sv_hash) *sv_h = kh_init(sv_hash);
khash_t(sv_geno) *geno_h = kh_init(sv_geno);
khash_t(colmap) *smp_cols;
khash_t(ped) *ped_h = 0;
mempool_t *mp;
char **samples;
o.min_q = 40; o.min_s = 80; o.min_len = 150; o.min_dp = 10; o.bed = 0, o.fnped = 0, o.mi_prob=0.005;
while ((c = getopt(argc, argv, "hq:s:l:d:b:p:m:")) >= 0) {
if (c == 'h') { usage(stderr, &o); return 0; }
else if (c == 'q') o.min_q = atoi(optarg);
else if (c == 's') o.min_s = atoi(optarg);
else if (c == 'l') o.min_len = atoi(optarg);
else if (c == 'd') o.min_dp = atoi(optarg);
else if (c == 'p') o.fnped = optarg;
else if (c == 'm') o.mi_prob = atof(optarg);
else if (c == 'b') {
if ((o.bed = bed_read(optarg)) == NULL) {
return -1;
}
}
}
if (o.mi_prob < 0.0000000000001 || o.mi_prob > 0.1) {
fprintf(stderr, "Error. Probability of a mendelian inconsistency must be between 0.1 and 0.0000000000001.\n");
}
if (argc - optind < 1) {
usage(stderr, &o);
return 1;
}
// Open files and initalize aux data //
n = argc - optind;
data = calloc(n, sizeof(aux_t*));
samples = (char**)malloc(n * sizeof(char*));
for (i = 0; i < n; ++i) {
data[i] = calloc(1, sizeof (aux_t));
data[i]->fp = sam_open(argv[optind + i], "r");
if (!data[i]->fp) {
fprintf(stderr, "Input file \"%s\" could not be opened.\n", argv[optind + 1]);
return 1;
}
data[i]->min_mapq = o.min_q;
data[i]->min_as = o.min_s;
data[i]->min_len = o.min_len;
data[i]->hdr = sam_hdr_read(data[i]->fp);
if (!data[i]->hdr) {
fprintf(stderr, "Could not read the header for input file \"%s\".\n", argv[optind + 1]);
return 1;
}
samples[i] = find_sample(data[i]->hdr, &res);
if (!samples[i]) {
fprintf(stderr, "Warning. No sample name detected for bam %s. Using filename\n", argv[optind + i]);
samples[i] = argv[optind + i];
}
}
h = data[0]->hdr;
smp_cols = map_samples(samples, n);
if (o.fnped) {
if ((ped_h = read_ped(o.fnped, smp_cols)) == 0) { return -1; }
}
// The core data processing loop //
mplp = bam_mplp_init(n, read_bam, (void**)data);
n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
plp = calloc(n, sizeof(bam_pileup1_t*)); // plp[i] points to the array of covering reads in mplp
//quals = (qual_vec_t*)calloc(n, sizeof(qual_vec_t));
mp = mp_init();
while ((ret = bam_mplp_auto(mplp, &tid, &pos, n_plp, plp)) > 0) { // iterate of positions with coverage
int n_sv;
if (o.bed && tid >= 0 && !bed_overlap(o.bed, h->target_name[tid], pos, pos+1)) continue;
n_sv = plp2sv(h, tid, pos, n, n_plp, plp, sv_h);
if (n_sv > 1) { fprintf(stderr, "Warning: more than two alleles detected at %s:%d\n", h->target_name[tid], pos); }
if (n_sv) {
fprintf(stderr, "SV detected at %d:%d\n", tid, pos);
for (k_iter = kh_begin(sv_h); k_iter != kh_end(sv_h); ++k_iter) {
if (kh_exist(sv_h, k_iter)) {
sv1 = kh_value(sv_h, k_iter);
fprintf(stderr, "SV tid1=%d, tid2=%d, pos1=%d, pos2=%d, ori1=%d, ori2=%d, allele=%d\n", sv1.tid1, sv1.tid2, sv1.pos1, sv1.pos2, sv1.ori1, sv1.ori2, sv1.allele);
}
}
res = get_qual_data(h, tid, pos, n, n_plp, plp, n_sv + 1, sv_h, geno_h, mp);
if (res < 0) {
fprintf(stderr, "Error collecting quality data from reads\n");
return -1;
}
kh_clear(sv_hash, sv_h);
}
}
print_header(h, optind, n, argv);
genotype_sv(h, n, geno_h, o.min_dp, ped_h, o.mi_prob);
free(n_plp);
free(plp);
bam_mplp_destroy(mplp);
mp_destroy(mp);
if (o.bed) bed_destroy(o.bed);
for (i = 0; i < n; ++i) {
bam_hdr_destroy(data[i]->hdr);
sam_close(data[i]->fp);
free(data[i]);
free(samples[i]);
}
free(data);
free(samples);
kh_destroy(sv_hash, sv_h);
kh_destroy(sv_geno, geno_h);
kh_destroy(colmap, smp_cols);
kh_destroy(ped, ped_h);
return 0;
}
