void bamFetchAlreadyOpen(samfile_t *samfile, bam_index_t *idx, char *bamFileName,
char *position, bam_fetch_f callbackFunc, void *callbackData)
/* With the open bam file, return items the same way with the callbacks as with bamFetch() */
/* except in this case use an already-open bam file and index (use bam_index_load and free() for */
/* the index). It seems a little strange to pass the filename in with the open bam, but */
/* it's just used to report errors. */
{
int chromId, start, end;
int ret = bam_parse_region(samfile->header, position, &chromId, &start, &end);
if (ret != 0 && startsWith("chr", position))
ret = bam_parse_region(samfile->header, position+strlen("chr"), &chromId, &start, &end);
if (ret != 0)
// If the bam file does not cover the current chromosome, OK
return;
ret = bam_fetch(samfile->x.bam, idx, chromId, start, end, callbackData, callbackFunc);
if (ret != 0)
warn("bam_fetch(%s, %s (chromId=%d) failed (%d)", bamFileName, position, chromId, ret);
}
void
bam_streamer::
set_new_region(const char* region)
{
int ref,beg,end;
bam_parse_region(_bfp->header, region, &ref, &beg, &end); // parse the region
set_new_region(ref,beg,end);
_region=region;
}
int MyBamWrap::myPassRegion(region_t & region,string& chrName, uint32_t& lpos, uint32_t& rpos)
{
char tmp[1024]="";
char tmpnum[128]="";
strcat(tmp,chrName.c_str());
strcat(tmp,":");
sprintf(&tmpnum[0],"%u",lpos);
strcat(tmp,tmpnum);
strcat(tmp,"-");
sprintf(&tmpnum[0],"%u",rpos);
strcat(tmp,tmpnum);
bam_parse_region(in->header, tmp, ®ion.tid, ®ion.lpos , ®ion.rpos);
return 0;
}
bam_streamer::
bam_streamer(
const char* filename,
const char* region)
: _is_record_set(false),
_bfp(nullptr),
_hidx(nullptr),
_hitr(nullptr),
_record_no(0),
_stream_name(filename),
_is_region(false)
{
assert(nullptr != filename);
if ('\0' == *filename)
{
throw blt_exception("Can't initialize bam_streamer with empty filename\n");
}
_bfp = samopen(filename, "rb", 0);
if (nullptr == _bfp)
{
log_os << "ERROR: Failed to open SAM/BAM/CRAM file: " << filename << "\n";
exit(EXIT_FAILURE);
}
if (nullptr == region)
{
// read the whole BAM file:
if (_bfp->header->n_targets)
{
// parse a fake region so that header->hash is created
std::string fake_region(target_id_to_name(0));
fake_region += ":1-1";
int ref,beg,end;
bam_parse_region(_bfp->header, fake_region.c_str(), &ref, &beg, &end);
}
return;
}
// read a specific region of the bam file:
set_new_region(region);
}
void seekRegion(BamReaderData * data) {
int tid, beg, end;
if (data->conf->reg) {
// Create BAM iterator at region
if (bam_parse_region(data->data->h, data->conf->reg, &tid, &beg, &end) < 0) {
fprintf(stderr, "[%s] malformatted region or wrong seqname for input.\n", __func__);
exit(1);
}
data->data->iter = bam_iter_query(data->idx, tid, beg, end);
data->ref_tid = tid;
} else {
// Create general BAM iterator
data->data->iter = NULL;
}
// Create pileup iterator
data->iter = bam_mplp_init(1, mplp_func, (void**) &data->data);
}
int sam_fetch(char *ifn, char *ofn, char *reg, void *data, sam_fetch_f func) {
int ret = 0;
samfile_t *in = samopen(ifn, "rb", 0);
samfile_t *out = 0;
if (ofn) out = samopen(ofn, "wb", in->header);
if (reg) {
bam_index_t *idx = bam_index_load(ifn);
if (idx == 0) {
fprintf(stderr, "[%s:%d] Random alignment retrieval only works for indexed BAM files.\n",
__func__, __LINE__);
exit(1);
}
int tid, beg, end;
bam_parse_region(in->header, reg, &tid, &beg, &end);
if (tid < 0) {
fprintf(stderr, "[%s:%d] Region \"%s\" specifies an unknown reference name. \n",
__func__, __LINE__, reg);
exit(1);
}
bam_iter_t iter;
bam1_t *b = bam_init1();
iter = bam_iter_query(idx, tid, beg, end);
while ((ret = bam_iter_read(in->x.bam, iter, b)) >= 0) func(b, in, out, data);
bam_iter_destroy(iter);
bam_destroy1(b);
bam_index_destroy(idx);
} else {
bam1_t *b = bam_init1();
while ((ret = samread(in, b)) >= 0) func(b, in, out, data);
bam_destroy1(b);
}
if (out) samclose(out);
samclose(in);
if (ret != -1) { /* truncated is -2 */
fprintf(stderr, "[%s:%d] Alignment retrieval failed due to truncated file\n",
__func__, __LINE__);
exit(1);
}
return ret;
}
int main(int argc, char *argv[])
{
tmpstruct_t tmp;
if (argc == 1) {
fprintf(stderr, "Usage: calDepth <in.bam> [region]\n");
return 1;
}
tmp.beg = 0; tmp.end = 0x7fffffff;
tmp.in = samopen(argv[1], "rb", 0);
if (tmp.in == 0) {
fprintf(stderr, "Fail to open BAM file %s\n", argv[1]);
return 1;
}
if (argc == 2) { // if a region is not specified
sampileup(tmp.in, -1, pileup_func, &tmp);
} else {
int ref;
bam_index_t *idx;
bam_plbuf_t *buf;
idx = bam_index_load(argv[1]); // load BAM index
if (idx == 0) {
fprintf(stderr, "BAM indexing file is not available.\n");
return 1;
}
bam_parse_region(tmp.in->header, argv[2], &ref,
&tmp.beg, &tmp.end); // parse the region
if (ref < 0) {
fprintf(stderr, "Invalid region %s\n", argv[2]);
return 1;
}
buf = bam_plbuf_init(pileup_func, &tmp); // initialize pileup
bam_fetch(tmp.in->x.bam, idx, ref, tmp.beg, tmp.end, buf, fetch_func);
bam_plbuf_push(0, buf); // finalize pileup
bam_index_destroy(idx);
bam_plbuf_destroy(buf);
}
samclose(tmp.in);
return 0;
}
bam_streamer::
bam_streamer(const char* filename,
const char* region)
: _is_record_set(false), _bfp(NULL), _bidx(NULL), _biter(NULL),
_record_no(0), _stream_name(filename), _is_region(false)
{
assert(NULL != filename);
assert('\0' != *filename);
_bfp = samopen(filename, "rb", 0);
if (NULL == _bfp)
{
log_os << "ERROR: Failed to open SAM/BAM file: " << filename << "\n";
exit(EXIT_FAILURE);
}
if (NULL == region)
{
// read the whole BAM file:
if (_bfp->header->n_targets)
{
// parse a fake region so that header->hash is created
std::string fake_region(target_id_to_name(0));
fake_region += ":1-1";
int ref,beg,end;
bam_parse_region(_bfp->header, fake_region.c_str(), &ref, &beg, &end);
}
return;
}
// read a specific region of the bam file:
set_new_region(region);
}
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_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;
}
int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg, int level)
#endif
{
bamFile fpout, *fp;
heap1_t *heap;
bam_header_t *hout = 0;
bam_header_t *hheaders = NULL;
int i, j, *RG_len = 0;
uint64_t idx = 0;
char **RG = 0, mode[8];
bam_iter_t *iter = 0;
if (headers) {
tamFile fpheaders = sam_open(headers);
if (fpheaders == 0) {
const char *message = strerror(errno);
fprintf(stderr, "[bam_merge_core] cannot open '%s': %s\n", headers, message);
return -1;
}
hheaders = sam_header_read(fpheaders);
sam_close(fpheaders);
}
g_is_by_qname = by_qname;
fp = (bamFile*)calloc(n, sizeof(bamFile));
heap = (heap1_t*)calloc(n, sizeof(heap1_t));
iter = (bam_iter_t*)calloc(n, sizeof(bam_iter_t));
// prepare RG tag
if (flag & MERGE_RG) {
RG = (char**)calloc(n, sizeof(void*));
RG_len = (int*)calloc(n, sizeof(int));
for (i = 0; i != n; ++i) {
int l = strlen(fn[i]);
const char *s = fn[i];
if (l > 4 && strcmp(s + l - 4, ".bam") == 0) l -= 4;
for (j = l - 1; j >= 0; --j) if (s[j] == '/') break;
++j; l -= j;
RG[i] = calloc(l + 1, 1);
RG_len[i] = l;
strncpy(RG[i], s + j, l);
}
}
// read the first
for (i = 0; i != n; ++i) {
bam_header_t *hin;
fp[i] = bam_open(fn[i], "r");
if (fp[i] == 0) {
int j;
fprintf(stderr, "[bam_merge_core] fail to open file %s\n", fn[i]);
for (j = 0; j < i; ++j) bam_close(fp[j]);
free(fp); free(heap);
// FIXME: possible memory leak
return -1;
}
hin = bam_header_read(fp[i]);
if (i == 0) { // the first BAM
hout = hin;
} else { // validate multiple baf
int min_n_targets = hout->n_targets;
if (hin->n_targets < min_n_targets) min_n_targets = hin->n_targets;
for (j = 0; j < min_n_targets; ++j)
if (strcmp(hout->target_name[j], hin->target_name[j]) != 0) {
fprintf(stderr, "[bam_merge_core] different target sequence name: '%s' != '%s' in file '%s'\n",
hout->target_name[j], hin->target_name[j], fn[i]);
return -1;
}
// If this input file has additional target reference sequences,
// add them to the headers to be output
if (hin->n_targets > hout->n_targets) {
swap_header_targets(hout, hin);
// FIXME Possibly we should also create @SQ text headers
// for the newly added reference sequences
}
bam_header_destroy(hin);
}
}
if (hheaders) {
// If the text headers to be swapped in include any @SQ headers,
// check that they are consistent with the existing binary list
// of reference information.
if (hheaders->n_targets > 0) {
if (hout->n_targets != hheaders->n_targets) {
fprintf(stderr, "[bam_merge_core] number of @SQ headers in '%s' differs from number of target sequences\n", headers);
if (!reg) return -1;
}
for (j = 0; j < hout->n_targets; ++j)
if (strcmp(hout->target_name[j], hheaders->target_name[j]) != 0) {
fprintf(stderr, "[bam_merge_core] @SQ header '%s' in '%s' differs from target sequence\n", hheaders->target_name[j], headers);
if (!reg) return -1;
}
}
swap_header_text(hout, hheaders);
bam_header_destroy(hheaders);
}
if (reg) {
int tid, beg, end;
if (bam_parse_region(hout, reg, &tid, &beg, &end) < 0) {
fprintf(stderr, "[%s] Malformated region string or undefined reference name\n", __func__);
return -1;
}
for (i = 0; i < n; ++i) {
bam_index_t *idx;
idx = bam_index_load(fn[i]);
iter[i] = bam_iter_query(idx, tid, beg, end);
bam_index_destroy(idx);
}
}
for (i = 0; i < n; ++i) {
heap1_t *h = heap + i;
h->i = i;
h->b = (bam1_t*)calloc(1, sizeof(bam1_t));
if (bam_iter_read(fp[i], iter[i], h->b) >= 0) {
h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)((int32_t)h->b->core.pos+1)<<1 | bam1_strand(h->b);
h->idx = idx++;
}
else h->pos = HEAP_EMPTY;
}
if (flag & MERGE_UNCOMP) level = 0;
else if (flag & MERGE_LEVEL1) level = 1;
strcpy(mode, "w");
if (level >= 0) sprintf(mode + 1, "%d", level < 9? level : 9);
if ((fpout = strcmp(out, "-")? bam_open(out, "w") : bam_dopen(fileno(stdout), "w")) == 0) {
fprintf(stderr, "[%s] fail to create the output file.\n", __func__);
return -1;
}
bam_header_write(fpout, hout);
bam_header_destroy(hout);
#ifndef _PBGZF_USE
if (!(flag & MERGE_UNCOMP)) bgzf_mt(fpout, n_threads, 256);
#endif
ks_heapmake(heap, n, heap);
while (heap->pos != HEAP_EMPTY) {
bam1_t *b = heap->b;
if (flag & MERGE_RG) {
uint8_t *rg = bam_aux_get(b, "RG");
if (rg) bam_aux_del(b, rg);
bam_aux_append(b, "RG", 'Z', RG_len[heap->i] + 1, (uint8_t*)RG[heap->i]);
}
bam_write1_core(fpout, &b->core, b->data_len, b->data);
if ((j = bam_iter_read(fp[heap->i], iter[heap->i], b)) >= 0) {
heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)((int)b->core.pos+1)<<1 | bam1_strand(b);
heap->idx = idx++;
} else if (j == -1) {
heap->pos = HEAP_EMPTY;
free(heap->b->data); free(heap->b);
heap->b = 0;
} else fprintf(stderr, "[bam_merge_core] '%s' is truncated. Continue anyway.\n", fn[heap->i]);
ks_heapadjust(heap, 0, n, heap);
}
if (flag & MERGE_RG) {
for (i = 0; i != n; ++i) free(RG[i]);
free(RG); free(RG_len);
}
for (i = 0; i != n; ++i) {
bam_iter_destroy(iter[i]);
bam_close(fp[i]);
}
bam_close(fpout);
free(fp); free(heap); free(iter);
return 0;
}
BamX::BamX(pars & Params1) // optional constructor
{
// parameters
Params=Params1;
Nread=0;
Npair=0;
Nproper=0;
Nout=0;
LFlow=INT_MIN;
LFhigh=INT_MAX;
region.limit=false;
IlluminizeBam=0;
outFragTailBam=false;
outInterChromBam=false;
outUniqueMultipleBam=false;
outUniquePartialBam=false;
outUniqueUnmappedBam=false;
outAllPairsBam=false;
outReadPairPosBam=false;
//output file
//samfile_t *fp;
bam_header_t *bam_header;
string s = Params.getInput();
BamUtil bam1(s);
Bam = bam1;
string filename=extractfilename(s);
// parameters
string fragPosFile = Params.getString("ReadPairPosFile");
string r = Params.getString("ChromRegion");
int maxReads = Params.getInt("MaxReads");
Qmin = Params.getInt("Qmin");
LRmin = Params.getInt("MinReadLength");
maxmismatchPC=Params.getDouble("FractionMaxMisMatches");
FragLengthWindow=Params.getInt("FragmentLengthWindow");
int cmd_MateMode=Params.getInt("ReadPairSenseConfig");
string ReferenceFastaFile=Params.getString("ReferenceFastaFile");
FragmentTailPercent=Params.getDouble("FragmentTailPercent");
IlluminizeBam=Params.getInt("Illuminize")>0;
outputDirectory=Params.getString("OutputDirectory");
int minLR=Params.getInt("MinReadLength");
int SplitBracketMin=Params.getInt("SplitBracketMin");
int SplitBaseQmin=Params.getInt("SplitBaseQmin");
string StatFile=Params.getString("StatFile");
if (StatFile.size()>0) {
hists H1(StatFile);
hist HLF=H1.h["LF"];
hist HLR=H1.h["LR"];
Params.setHist("LF",HLF);
Params.setHist("LR",HLR);
H1.h.clear(); // free some memory
if (FragmentTailPercent>0) {
LFlow=int(HLF.p2xTrim(FragmentTailPercent/100.));
LFhigh=int(HLF.p2xTrim(1-FragmentTailPercent/100.));
}
}
int dbg = Params.getInt("Dbg");
time(&tprev);
if (ReferenceFastaFile.size()>0) {
FastaObj RF1(ReferenceFastaFile, "");
Reference=RF1;
RF1.seq.clear(); // free some memory
}
bam_header= Bam.fp->header;
string bamheadertext = bam_header->text;
ReadGroup = extractBamTag(bamheadertext,"@RG");
outAllPairsBam=(r.size()>0);
if (!outAllPairsBam) {
outFragTailBam=true; //FragmentTailPercent>=0;
outInterChromBam=true;
outUniqueMultipleBam=true;
outUniquePartialBam=true;
outUniqueUnmappedBam=true;
}
// output Bams
outputBam.clear();
/*
// test BamHeaderContainer
vector<BamHeaderContainer> x;
string sv=SpannerVersion;
string q="@PG\tID:FragmentTail\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
while (true) {
string outfile=outputDirectory+"/"+filename+".fragtail.bam";
q=q+"\n@PG\tID:FragmentTail\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
BamHeaderContainer x1( bam_header, q);
x.push_back(x1);
bam_header_t* h1=x[x.size()-1].header();
cout<< h1->text << endl;
}
cout << x.size() << endl;
*/
samfile_t *fpFT=0;
samfile_t *fpIC=0;
samfile_t *fpUM=0;
samfile_t *fpUP=0;
samfile_t *fpUZ=0;
samfile_t *fpAP=0;
samfile_t *fpWP=0;
//region
if (r.size()>0) {
int r1,r2,r3;
C_region r0(r);
region=r0;
string bamRegion=region.region;
size_t k=bamRegion.find("chr");
if (k!=string::npos) {
bamRegion=bamRegion.substr(3);
}
if ( bam_parse_region(bam_header, bamRegion.c_str(), &r1, &r2, &r3)==0) {
region.limit=true;
region.anchor=r1;
region.start=r2;
region.end=r3;
} else {
cerr << "region not found\t" << r << endl;
exit(111);
}
}
//fragPosFile
if (fragPosFile.size()>0) {
FragmentPosFileObj fp(fragPosFile);
if (fp.fragmentPosList.size()>0) {
FragPos=fp;
} else {
cerr << "Read Pair Pos file not found\t" << fragPosFile << endl;
exit(112);
}
outFragTailBam=false;
outInterChromBam=false;
outUniqueMultipleBam=false;
outUniquePartialBam=false;
outUniqueUnmappedBam=false;
outReadPairPosBam=true;
}
if (outAllPairsBam) {
string outfile=outputDirectory+"/"+filename+"."+r+".bam";
string sv=SpannerVersion;
string q="@PG\tID:Region\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["AP"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["AP"].header();
if ((fpAP = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(160);
}
}
if (outFragTailBam) {
string outfile=outputDirectory+"/"+filename+".fragtail.bam";
string sv=SpannerVersion;
string q="@PG\tID:FragmentTail\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["FT"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["FT"].header();
if ((fpFT = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(161);
}
}
if (outInterChromBam) {
string outfile=outputDirectory+"/"+filename+".interchrom.bam";
string sv=SpannerVersion;
string q="@PG\tID:InterChromPairs\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["IC"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["IC"].header();
if ((fpIC = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(162);
}
}
if (outUniqueMultipleBam) {
string outfile=outputDirectory+"/"+filename+".uMult.bam";
string sv=SpannerVersion;
string q="@PG\tID:uniqMultiplyMappedPairs\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["UM"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["IUM"].header();
if ((fpUM = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(163);
}
}
if (outUniquePartialBam) {
string outfile=outputDirectory+"/"+filename+".uPart.bam";
string sv=SpannerVersion;
string q="@PG\tID:uniqPartiallyMappedPairs\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["UP"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["UP"].header();
if ((fpUP = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(164);
}
}
if (outUniqueUnmappedBam) {
string outfile=outputDirectory+"/"+filename+".uUnmapped.bam";
string sv=SpannerVersion;
string q="@PG\tID:uniqUnMappedPairs\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["UZ"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["UZ"].header();
if ((fpUZ = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(165);
}
}
if (outReadPairPosBam) {
string outfile=outputDirectory+"/"+filename+".weirdpairs.bam";
string sv=SpannerVersion;
string q="@PG\tID:weirdpairs\tPN:SpannerX\tVN"+sv+"\tCL:"+Params.getCmdLine();
outputBam["WP"]=BamHeaderContainer(bam_header,q);
bam_header_t* h1=outputBam["WP"].header();
if ((fpWP = samopen(outfile.c_str(), "wb", h1)) == 0) {
fprintf(stderr, "samopen: Fail to open output BAM file %s\n", filename.c_str());
exit(165);
}
}
cout << ReadGroup << endl << endl;
//extractMateMode();
if (cmd_MateMode>=0) MateMode=cmd_MateMode;
BamContainerPair bampair;
bool more = true;
while (more)
{
bampair=Bam.getNextBamPair();
// skip if neither end within region
more=(bampair.BamEnd.size()>1);
Npair++;
if (Npair>=maxReads) break;
//
if ( (dbg!=0)&&(elapsedtime()>float(dbg))) {
time(&tprev);
cout << " pairs:" << Npair << "\toutput:" << Nout;
cout << "\tchr:" << bampair.BamEnd[0].b.core.tid+1;
cout << "\tpos:" << bampair.BamEnd[0].b.core.pos;
cout << endl;
}
if (!more) continue;
if (region.limit) {
bool overlap = false;
for (int e=0; e<=1; e++) {
int a1=bampair.BamEnd[e].b.core.tid;
int p1=bampair.BamEnd[e].b.core.pos;
int p2=p1+bampair.BamEnd[e].len;
overlap=region.overlap(a1,p1,p2);
if (overlap) break;
}
if (!overlap) continue;
}
bampair.Illuminize(IlluminizeBam);
bampair.calcFragmentLengths();
more=(bampair.BamEnd[1].packeddata.size()>1);
//if (bampair.BamEnd[0].b.core.tid==bampair.BamEnd[1].b.core.tid)
// cout<< bampair << endl;
bool bothmap = ((bampair.BamEnd[0].b.core.flag&BAM_FUNMAP)==0)&&((bampair.BamEnd[0].b.core.flag&BAM_FMUNMAP)==0);
if (outAllPairsBam) {
Nout++;
int s1=samwrite(fpAP, &(bampair.BamEnd[0].b));
int s2=samwrite(fpAP, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to pairs.bam" << endl;
exit(150);
}
}
if (outReadPairPosBam) {
int ichr1=bampair.BamEnd[0].b.core.tid+1;
int istd1=bampair.BamEnd[0].sense=='+'? 0: 1;
int ista1=bampair.BamEnd[0].b.core.pos+1;
int iq1=bampair.BamEnd[0].q;
int ichr2=bampair.BamEnd[1].b.core.tid+1;
int istd2=bampair.BamEnd[1].sense=='+'? 0: 1;
int ista2=bampair.BamEnd[1].b.core.pos+1;
int iq2=bampair.BamEnd[1].q;
FragmentPosObj fp1(0,ichr1,istd1,ista1,0,ichr2,istd2,ista2,0,iq1, iq2,0);
/*
if ((fp1.chr1==10)&&(fp1.start1>=89687801)&&(fp1.end1<=89700722)) {
cout << "read "<< fp1 << endl;
}
*/
if (FragPos.find(fp1)) {
Nout++;
int s1=samwrite(fpWP, &(bampair.BamEnd[0].b));
int s2=samwrite(fpWP, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to weirdpairs.bam" << endl;
exit(156);
}
}
}
bool ok[2];
for (int e=0; e<2; e++) {
uint8_t* bq=bam1_qual(&(bampair.BamEnd[e].b));
int LR=bampair.BamEnd[0].b.core.l_qseq;
double bok=0;
for (int ib=0; ib<LR; ib++) {
if (bq[ib]>SplitBaseQmin) {
bok++;
}
}
ok[e]=(bok>LRmin);
}
if (! (ok[0]&ok[1]) )
continue;
if ( (outFragTailBam) & ((bampair.BamEnd[0].q>=Qmin)|(bampair.BamEnd[1].q>=Qmin)) ) {
bool FT=(bampair.FragmentLength>LFhigh)|((bampair.FragmentLength<LFlow)&(bampair.FragmentLength>INT_MIN))&bothmap;
if (FT && (fpFT!=0)) {
Nout++;
int s1=samwrite(fpFT, &(bampair.BamEnd[0].b));
int s2=samwrite(fpFT, &(bampair.BamEnd[1].b));
//if (outputBam["FT"].write(&(bampair.BamEnd[0].b),&(bampair.BamEnd[1].b))) {
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to fragtail.bam" << endl;
exit(151);
}
}
}
if ((outInterChromBam) & ((bampair.BamEnd[0].q>=Qmin)&(bampair.BamEnd[1].q>=Qmin))) {
bool IC=(bampair.BamEnd[0].b.core.tid!=bampair.BamEnd[1].b.core.tid)&&bothmap;
if (IC && (fpIC!=0)) {
Nout++;
int s1=samwrite(fpIC, &(bampair.BamEnd[0].b));
int s2=samwrite(fpIC, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to interchrom.bam" << endl;
exit(152);
}
}
}
if ((outUniqueMultipleBam) & ((bampair.BamEnd[0].q>=Qmin)|(bampair.BamEnd[1].q>=Qmin))){
int im=bampair.BamEnd[0].nmap>1? 0: 1;
int iu=bampair.BamEnd[0].q>=Qmin? 0: 1;
bool UM=(bampair.BamEnd[iu].nmap>1)&&(iu!=im)&&bothmap;
if (UM && (fpUM!=0)) {
Nout++;
int s1=samwrite(fpUM, &(bampair.BamEnd[0].b));
int s2=samwrite(fpUM, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to uMult.bam" << endl;
exit(153);
}
}
}
if ( (outUniquePartialBam) && ((bampair.BamEnd[0].q>=Qmin)|(bampair.BamEnd[1].q>=Qmin)) && bothmap) {
int c0=bampair.BamEnd[0].clip[0]+bampair.BamEnd[0].clip[1];
int LR=bampair.BamEnd[0].b.core.l_qseq;
bool split0=((LR-c0)>SplitBracketMin)&(c0>SplitBracketMin);
int ib0=0;
if ((split0)&(bampair.BamEnd[0].clip[0]>SplitBracketMin)) {
ib0=bampair.BamEnd[0].clip[0];
} else if ((split0)&(bampair.BamEnd[0].clip[1]>SplitBracketMin) ) {
ib0=LR-bampair.BamEnd[0].clip[1];
}
split0=split0&(ib0>0);
if (split0) {
uint8_t* bq=bam1_qual(&(bampair.BamEnd[0].b));
for (int ib=(ib0-SplitBracketMin); ib<(ib0+SplitBracketMin); ib++) {
if (bq[ib]<SplitBaseQmin) {
split0=false;
break;
}
}
}
int c1=bampair.BamEnd[1].clip[0]+bampair.BamEnd[1].clip[1];
LR=bampair.BamEnd[1].b.core.l_qseq;
bool split1=((LR-c0)>SplitBracketMin)&(c1>SplitBracketMin);;
int ib1=0;
if ((split1)&(bampair.BamEnd[1].clip[0]>SplitBracketMin)) {
ib1=bampair.BamEnd[1].clip[0];
} else if ((split1)&(bampair.BamEnd[1].clip[1]>SplitBracketMin) ) {
ib1=LR-bampair.BamEnd[1].clip[1];
}
split1=split1&(ib1>0);
if (split1) {
uint8_t* bq=bam1_qual(&(bampair.BamEnd[1].b));
for (int ib=(ib1-SplitBracketMin); ib<(ib1+SplitBracketMin); ib++) {
if (bq[ib]<SplitBaseQmin) {
split1=false;
break;
}
}
}
bool UP=(split0|split1)&((c1+c0)>minLR);
if (UP && (fpUP!=0)) {
Nout++;
int s1=samwrite(fpUP, &(bampair.BamEnd[0].b));
int s2=samwrite(fpUP, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to uPart.bam" << endl;
exit(154);
}
}
}
if ( (outUniqueUnmappedBam) & ((bampair.BamEnd[0].q>=Qmin)|(bampair.BamEnd[1].q>=Qmin)) ) {
bool z0=((bampair.BamEnd[0].b.core.flag&BAM_FUNMAP)>0);
bool z1=((bampair.BamEnd[1].b.core.flag&BAM_FUNMAP)>0);
uint8_t* bq=bam1_qual(&(bampair.BamEnd[0].b));
for (int nb,ib=0; ib<bampair.BamEnd[0].b.core.l_qseq; ib++) {
if (bq[ib]<SplitBaseQmin) {
nb++;
}
}
bool UZ=(z0|z1)&(!(z1&z0));
if (UZ && (fpUZ!=0)) {
Nout++;
int s1=samwrite(fpUZ, &(bampair.BamEnd[0].b));
int s2=samwrite(fpUZ, &(bampair.BamEnd[1].b));
if ((s1*s2)>0) {
continue;
} else {
cerr << "bad write to uUnmapped.bam" << endl;
exit(155);
}
}
}
//cout<< bampair.Orientation << "\t"<< bampair.FragmentLength << "\t" <<bampair.BamEnd[1].b.core.pos << endl;
}
if (outReadPairPosBam) {
samclose(fpWP);
} else {
if (outAllPairsBam) {
samclose(fpAP);
} else {
samclose(fpFT);
samclose(fpIC);
samclose(fpUP);
samclose(fpUM);
samclose(fpUZ);
}
}
/*
for (ioutputBam=outputBam.begin(); ioutputBam!=outputBam.end(); ioutputBam++) {
(*ioutputBam).second.close();
}
if (FragmentTailPercent>0)
outputBam["FT"].close();
*/
samclose(Bam.fp);
}
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;
}
int main_ld(int argc, char *argv[])
{
int chr; //! chromosome identifier
int beg; //! beginning coordinate for analysis
int end; //! end coordinate for analysis
int ref; //! ref
long num_windows; //! number of windows
std::string msg; //! string for error message
bam_plbuf_t *buf; //! pileup buffer
ldData t;
// parse the command line options
std::string region = t.parseCommandLine(argc, argv);
// check input BAM file for errors
t.checkBAM();
// initialize the sample data structure
t.bam_smpl_init();
// add samples
t.bam_smpl_add();
// initialize error model
t.em = errmod_init(1.0-0.83);
// parse genomic region
int k = bam_parse_region(t.h, region, &chr, &beg, &end);
if (k < 0)
{
msg = "Bad genome coordinates: " + region;
fatal_error(msg, __FILE__, __LINE__, 0);
}
// fetch reference sequence
t.ref_base = faidx_fetch_seq(t.fai_file, t.h->target_name[chr], 0, 0x7fffffff, &(t.len));
// calculate the number of windows
if (t.flag & BAM_WINDOW)
num_windows = ((end-beg)-1)/t.win_size;
else
{
t.win_size = (end-beg);
num_windows = 1;
}
// iterate through all windows along specified genomic region
for (long cw=0; cw < num_windows; cw++)
{
// construct genome coordinate string
std::string scaffold_name(t.h->target_name[chr]);
std::ostringstream winc(scaffold_name);
winc.seekp(0, std::ios::end);
winc << ":" << beg+(cw*t.win_size)+1 << "-" << ((cw+1)*t.win_size)+(beg-1);
std::string winCoord = winc.str();
// initialize number of sites to zero
t.num_sites = 0;
// parse the BAM file and check if region is retrieved from the reference
if (t.flag & BAM_WINDOW)
{
k = bam_parse_region(t.h, winCoord, &ref, &(t.beg), &(t.end));
if (k < 0)
{
msg = "Bad window coordinates " + winCoord;
fatal_error(msg, __FILE__, __LINE__, 0);
}
}
else
{
ref = chr;
t.beg = beg;
t.end = end;
if (ref < 0)
{
msg = "Bad scaffold name: " + region;
fatal_error(msg, __FILE__, __LINE__, 0);
}
}
// initialize nucdiv variables
t.init_ld();
// create population assignments
t.assign_pops();
// initialize pileup
buf = bam_plbuf_init(make_ld, &t);
// fetch region from bam file
if ((bam_fetch(t.bam_in->x.bam, t.idx, ref, t.beg, t.end, buf, fetch_func)) < 0)
{
msg = "Failed to retrieve region " + region + " due to corrupted BAM index file";
fatal_error(msg, __FILE__, __LINE__, 0);
}
// finalize pileup
bam_plbuf_push(0, buf);
// calculate linkage disequilibrium statistics
ld_func fp[3] = {&ldData::calc_zns, &ldData::calc_omegamax, &ldData::calc_wall};
(t.*fp[t.output])();
// print results to stdout
t.print_ld(chr);
// take out the garbage
t.destroy_ld();
bam_plbuf_destroy(buf);
}
// end of window interation
errmod_destroy(t.em);
samclose(t.bam_in);
bam_index_destroy(t.idx);
t.bam_smpl_destroy();
free(t.ref_base);
return 0;
}
int main(int argc,char* argv[]) {
time_t timestamp, current;
int i,j,k;
int a,n;
char *pc;
FILE *input_file;
FILE *output_file;
FILE* log_file=stderr;
bamFile bam_input;
bam_header_t *header;
bam1_t* b;
bam1_core_t *c;
char cps_file_name[MAXFILEBUFFLENGTH]="";
char bam_file_name[MAXFILEBUFFLENGTH]="";
char out_file_name[MAXFILEBUFFLENGTH]="";
char log_file_name[MAXFILEBUFFLENGTH]="";
char buff[MAXFILEBUFFLENGTH];
char chr[MAXFILEBUFFLENGTH];
int beg, beg_prev, end, pos, offset;
int ref_id, ref_id_prev, label;
int s, side;
int read_type, mapped_strand;
char ch;
int limit_counts = 0;
int* contig_count[2];
int* contig_index[2];
splice_site** contig_sites[2];
long int n_reads[N_READ_TYPES][2];
long int n_total_reads = 0;
long int n_skipped_reads = 0;
int max_intron_length=0;
int min_intron_length=0;
int ignore_gene_labels = 0;
int stranded = 1;
int rev_compl[2] = {1,0};
int other_end, the_end, donor_id, acceptor_id;
int *cigar;
int flagged = 0;
int margin = 4;
/** reading input from the command line **/
timestamp = time(NULL);
if(argc==1) {
fprintf(stderr, "BAM2SSJ is the utility for fast counting reads covering splice junctions\nCommand line use:\n");
fprintf(stderr, "%s -cps <cps_file> -bam <bam_file> [-out <out_file>] [-log <log_file>] [-maxlen <max_intron_length>] [-minlen <min_intron_length>] [-margin <length>] ",argv[0]);
fprintf(stderr, "[-v suppress verbose output] [-read1 0/1] [-read2 0/1] [-g ignore gene labels] [-u unstranded] [-f count reads flagged 0x800 only]\ntype %s -h for more info\n",argv[0]);
exit(1);
}
for(i=1;i<argc;i++) {
pc = argv[i];
if(*pc == '-') {
if(strcmp(pc+1,"cps") == 0) sscanf(argv[++i], "%s", &cps_file_name[0]);
if(strcmp(pc+1,"bam") == 0) sscanf(argv[++i], "%s", &bam_file_name[0]);
if(strcmp(pc+1,"out") == 0) sscanf(argv[++i], "%s", &out_file_name[0]);
if(strcmp(pc+1,"log") == 0) sscanf(argv[++i], "%s", &log_file_name[0]);
if(strcmp(pc+1,"read1") == 0) sscanf(argv[++i], "%i", &rev_compl[0]);
if(strcmp(pc+1,"read2") == 0) sscanf(argv[++i], "%i", &rev_compl[1]);
if(strcmp(pc+1,"lim") == 0) sscanf(argv[++i], "%i", &limit_counts);
if(strcmp(pc+1,"minlen") == 0) sscanf(argv[++i], "%i", &min_intron_length);
if(strcmp(pc+1,"maxlen") == 0) sscanf(argv[++i], "%i", &max_intron_length);
if(strcmp(pc+1,"margin") == 0) sscanf(argv[++i], "%i", &margin);
if(strcmp(pc+1,"v") == 0) verbose = 0;
if(strcmp(pc+1,"g") == 0) ignore_gene_labels = 1;
if(strcmp(pc+1,"u") == 0) stranded = 0;
if(strcmp(pc+1,"f") == 0) flagged = 1;
if(strcmp(pc+1,"h") ==0 ) {
fprintf(stderr, "Input: (1) sorted BAM file\n");
fprintf(stderr, "\t(2) CPS (chromosome-position-strand) tab-delimited file sorted by position (chr1 100 + etc)\n\n");
fprintf(stderr, "\tIn order to get CPS file from gtf, use the utility gtf2cps.sh\n");
fprintf(stderr, "\tImportant: CPS must be sorted by position ONLY!\n\n");
fprintf(stderr, "\tIf the 4th column contains (a numeric) gene label then only splice junctions within the same gene will be considered (unless the '-g' option is active)\n");
fprintf(stderr, "\tThe utility to generate CPS with gene labels is gtf2cps_with_gene_id.sh (or update the script accordingly if you are using genome other than human)\n\n");
fprintf(stderr, "Options:\n");
fprintf(stderr, "\t-maxlen <upper limit on intron length>; 0 = no limit (default=%i)\n",max_intron_length);
fprintf(stderr, "\t-minlen <lower limit on intron length>; 0 = no limit (default=%i)\n",min_intron_length);
fprintf(stderr, "\t-margin <length> minimum number of flanking nucleotides in the read in order to support SJ or cover EB, (default=%i)\n",margin);
fprintf(stderr, "\t-read1 0/1, reverse complement read1 no/yes (default=%i)\n",rev_compl[0]);
fprintf(stderr, "\t-read2 0/1, reverse complement read2 no/yes (default=%i)\n",rev_compl[1]);
fprintf(stderr, "\t-g ignore gene labels (column 4 of cps), default=%s\n", ignore_gene_labels ? "ON" : "OFF");
fprintf(stderr, "\t-u ignore strand (all reads map to the correct strand), default=%s\n", stranded ? "OFF" : "ON");
fprintf(stderr, "\t-f count only reads that are flagged 0x800 (uniquely mapped reads), default=%s\n", flagged ? "ON" : "OFF");
fprintf(stderr, "Output: tab-delimited (default=stdout)\n");
fprintf(stderr, "\tColumn 1 is splice_junction_id\n");
fprintf(stderr, "\tColumns 2-6 are counts of 53, 5X, X3, 50, and 03 reads for the correct (annotated) strand\n");
fprintf(stderr, "\tColumns 7-11 are similar counts for the incorrect (opposite to annotated) strand\n");
fprintf(stderr, "Descriptive read statistics are reported to stderr\n");
exit(1);
}
}
}
if(log_file_name[0]==0) {
log_file = stderr;
}
else {
log_file = fopen(log_file_name,"w");
if(log_file == NULL) log_file = stderr;
}
if(bam_file_name[0]==0) {
fprintf(log_file,"Bam not specified, exiting\n");
exit(1);
}
if(cps_file_name[0]==0) {
fprintf(log_file,"Input not specified, exiting\n");
exit(1);
}
if(out_file_name[0]==0) {
fprintf(log_file,"[Warning: output set to stdout]\n");
output_file = stdout;
}
else {
output_file = fopen(out_file_name,"w");
if(output_file == NULL) {
fprintf(log_file,"[Warning: output set to stdout]\n");
output_file = stdout;
}
}
if(max_intron_length>0) {
if(verbose) fprintf(log_file,"[Warning: set max intron length=%i]\n",max_intron_length);
}
if(ignore_gene_labels) {
if(verbose) fprintf(log_file,"[Warning: ignoring gene labels (column 4)]\n");
}
if(flagged) {
if(verbose) fprintf(log_file,"[Warning: only look at reads flagged 0x800]\n");
}
if(margin>0) {
if(verbose) fprintf(log_file,"[Warning: read margin set to %i]\n", margin);
}
if(verbose) {
for(s = 0; s < 2; s++) if(rev_compl[s]) fprintf(log_file,"[Warning: take reverse complement of read %i]\n", s+1);
fprintf(log_file,"[Warning: stranded = %s]\n", stranded ? "TRUE" : "FALSE (always correct strand)");
if(ignore_gene_labels) fprintf(log_file,"[Warning: ignore gene labels (column 4)]\n");
}
for(i = 0; i < N_READ_TYPES; i++) for(s = 0; s < 2; s++) n_reads[i][s] = 0;
/** initatializing BAM and header **/
bam_input = bam_open(bam_file_name, "r");
header = bam_header_read(bam_input);
if(bam_input == NULL || header == NULL) {
fprintf(log_file,"BAM can't be opened or contains no header, exiting\n");
exit(1);
}
/** reading input from CPS **/
input_file = fopen(cps_file_name, "r");
if(input_file == NULL) {
fprintf(log_file,"CPS can't be opened, exiting\n");
exit(1);
}
/** populating gene structure arrays **/
for(s = 0; s < 2; s++) {
contig_count[s] = (int*) malloc(sizeof(int) * (header->n_targets + ARRAY_MARGIN));
contig_index[s] = (int*) malloc(sizeof(int) * (header->n_targets + ARRAY_MARGIN));
contig_sites[s] = (splice_site**) malloc(sizeof(splice_site*) * (header->n_targets + ARRAY_MARGIN));
if(contig_count[s] == NULL || contig_sites[s] == NULL || contig_index[s] == NULL) {
fprintf(log_file, "Not enought memory, exiting\n");
exit(1);
}
}
for(s = 0; s < 2; s++)
for(i=0; i < header->n_targets; i++)
contig_count[s][i] = contig_index[s][i] = 0;
if(verbose) fprintf(log_file, "Reading %s pass1", cps_file_name);
while(fgets(buff, MAXFILEBUFFLENGTH, input_file)) {
sscanf(buff, "%s %*i %c", &chr[0], &ch);
bam_parse_region(header, chr, &i, &beg, &end);
s = (ch == '+' ? 0 : 1);
if(i < header->n_targets && i>=0) contig_count[s][i]++;
}
for(s = 0; s < 2; s++) {
for(i = 0;i < header->n_targets; i++) {
contig_sites[s][i] = (splice_site*) malloc(sizeof(splice_site) * (contig_count[s][i] + ARRAY_MARGIN));
if(contig_sites[s][i] == NULL) {
fprintf(log_file, "Not enought memory, exiting\n");
exit(1);
}
}
}
if(verbose) fprintf(log_file, "\n");
if(verbose) fprintf(log_file, "Reading %s pass2",cps_file_name);
fseek(input_file, 0, SEEK_SET);
while(fgets(buff, MAXFILEBUFFLENGTH, input_file)) {
sscanf(buff, "%s %i %c %i", &chr[0], &pos, &ch, &label);
bam_parse_region(header, chr, &i, &beg, &end);
s = (ch == '+' ? 0 : 1);
if(i < header->n_targets && i>=0) {
if(contig_index[s][i]>0) {
if(pos < contig_sites[s][i][contig_index[s][i]-1].pos) {
fprintf(log_file, "Splice sites weren't sorted, exiting\n");
exit(1);
}
}
contig_sites[s][i][contig_index[s][i]].pos = pos;
contig_sites[s][i][contig_index[s][i]].label = ignore_gene_labels ? 0 : label;
for(side = 0; side < 2; side++) {
contig_sites[s][i][contig_index[s][i]].count00[side] = 0;
contig_sites[s][i][contig_index[s][i]].count5X[side] = 0;
contig_sites[s][i][contig_index[s][i]].countX3[side] = 0;
contig_sites[s][i][contig_index[s][i]].junctions = NULL;
}
contig_index[s][i]++;
}
}
if(verbose) fprintf(log_file, "\n");
for(s = 0; s < 2; s++)
for(i = 0;i < header->n_targets; i++)
contig_index[s][i] = 0;
/** analysis starts here **/
b = bam_init1();
k = 0;
ref_id_prev = -1;
beg_prev = -1;
while(bam_read1(bam_input, b)>=0) {
c = &b->core;
ref_id = c->tid;
if(ref_id<0) continue;
if(flagged && ((c->flag & 0x800) == 0)) {
n_skipped_reads++;
continue;
}
if(stranded && ((c->flag & BAM_FREAD1) && (c->flag & BAM_FREAD2) || !(c->flag & BAM_FREAD1) && !(c->flag & BAM_FREAD2))) {
n_skipped_reads++;
continue;
}
cigar = bam1_cigar(b);
if(ref_id != ref_id_prev && ref_id_prev >= 0) {
if(contig_index[0][ref_id_prev] + contig_index[1][ref_id_prev] < contig_count[0][ref_id_prev] + contig_count[1][ref_id_prev]) {
if(log_file==stderr) progressbar(1, 1, header->target_name[ref_id_prev], verbose);
}
beg_prev = -1;
}
/*if(ref_id < ref_id_prev) {
fprintf(log_file,"BAM file wasn't sorted, exiting\n");
exit(1);
}*/
ref_id_prev = ref_id;
beg = c->pos + 1;
if(beg < beg_prev) {
fprintf(log_file,"BAM file wasn't sorted, exiting\n");
exit(1);
}
beg_prev = beg;
s = ((c->flag & BAM_FREVERSE)>0);
mapped_strand = (c->flag & BAM_FREAD1) ? (s + rev_compl[0]) & 1 : (s + rev_compl[1]) & 1;
the_end = bam_calend(c, cigar);
for(s = 0; s < 1 + stranded; s++) {
end = beg;
side = (s == mapped_strand) ? 0 : 1;
side *= stranded;
// keep reading until the currect site is on the same chromosome downstream of the read
while(contig_sites[s][ref_id][contig_index[s][ref_id]].pos < beg && contig_index[s][ref_id] < contig_count[s][ref_id]) {
contig_index[s][ref_id]++;
if(log_file==stderr) progressbar(contig_index[0][ref_id]+contig_index[1][ref_id], contig_count[0][ref_id]+contig_count[1][ref_id], header->target_name[ref_id], verbose);
}
read_type = RT_OTHER;
if(contig_index[s][ref_id]<contig_count[s][ref_id]) {
// check if the read is a split read and find its other end
read_type = RT_GENOME;
for(i = 0; i < c->n_cigar; i++) {
offset = cigar[i] >> 4;
switch(cigar[i] & 0x0F) {
case BAM_CMATCH: end += offset; // match to the reference
break;
case BAM_CINS: end += 0; // insertion to the reference, pointer stays unchanged
break;
case BAM_CDEL: end += offset; // deletion from the reference (technically the same as 'N') pointer moves
break;
case BAM_CREF_SKIP: other_end = end + offset;
donor_id = acceptor_id = -INFTY;
if(end - beg < margin) break;
if(the_end - other_end < margin) break;
for(j = contig_index[s][ref_id]; contig_sites[s][ref_id][j].pos <= other_end && j < contig_count[s][ref_id];j++) {
if(contig_sites[s][ref_id][j].pos - end < min_intron_length && min_intron_length > 0) continue;
if(contig_sites[s][ref_id][j].pos - end > max_intron_length && max_intron_length > 0) break;
if(contig_sites[s][ref_id][j].label == contig_sites[s][ref_id][contig_index[s][ref_id]].label) {
if(contig_sites[s][ref_id][j].pos == end - 1) donor_id = j;
if(contig_sites[s][ref_id][j].pos == other_end) acceptor_id = j;
}
}
if(donor_id>0 && acceptor_id>0) {
update_count(&contig_sites[s][ref_id][donor_id].junctions, acceptor_id, side);
contig_sites[s][ref_id][donor_id].count5X[side]++;
contig_sites[s][ref_id][acceptor_id].countX3[side]++;
read_type = RT_KJUNCT;
}
else {
read_type = RT_UJUNCT;
}
end = other_end;
break;
case BAM_CSOFT_CLIP:
case BAM_CHARD_CLIP:
case BAM_CPAD: break;
default: read_type = RT_OTHER;
}
}
if(read_type == RT_GENOME) {
for(j=contig_index[s][ref_id]; beg + margin <= contig_sites[s][ref_id][j].pos && contig_sites[s][ref_id][j].pos < end - margin && j<contig_count[s][ref_id]; j++) {
contig_sites[s][ref_id][j].count00[side]++;
read_type = RT_OVRLAP;
k++;
}
}
}
n_reads[read_type][side]++;
}
n_total_reads++;
if(k>limit_counts && limit_counts>0) break;
}