int bam_pileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data)
{
bam_plbuf_t *buf;
int ret;
bam1_t *b;
b = bam_init1();
buf = bam_plbuf_init(func, func_data);
bam_plbuf_set_mask(buf, mask);
while ((ret = bam_read1(fp, b)) >= 0)
bam_plbuf_push(b, buf);
bam_plbuf_push(0, buf);
bam_plbuf_destroy(buf);
bam_destroy1(b);
return 0;
}
uint calculate_cov_params(const char* const bam_name,
const int32_t tid,
const int32_t start,
const int32_t stop)
{
bamFile fp = bam_open(bam_name, "r");
bam_index_t* fp_index = bam_index_load(bam_name);
bam_plbuf_t *buf;
covdata* cvdt = ckallocz(sizeof(covdata));
cvdt->tid = tid;
cvdt->begin = start;
cvdt->end = stop;
cvdt->coverage = ckallocz((cvdt->end - cvdt->begin) * sizeof(uint32_t));
buf = bam_plbuf_init(pileup_func, cvdt);
bam_fetch(fp, fp_index, tid, start, stop, buf, fetch_func);
bam_plbuf_push(0, buf);
bam_plbuf_destroy(buf);
// calculate the mean coverage in the region of the putative deletion
uint i, covsum;
for(i = 0, covsum = 0; i < (cvdt->end - cvdt->begin); i++){
covsum += cvdt->coverage[i];
}
uint avgcov = floor(covsum * 1.0/(cvdt->end - cvdt->begin));
ckfree(cvdt->coverage);
ckfree(cvdt);
bam_close(fp);
bam_index_destroy(fp_index);
return avgcov;
}
/* callback for bam_fetch() */
static int fetch_func(const bam1_t *b, void *data)
{
bam_plbuf_t *buf = data;
if (b->core.flag & BAM_FPROPER_PAIR) {
bam_plbuf_push(b, buf);
}
return 0;
}
int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
{
bam_plbuf_t *buf;
int ret;
bam1_t *b;
b = bam_init1();
buf = bam_plbuf_init(func, func_data);
if (mask < 0) mask = BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP;
else mask |= BAM_FUNMAP;
while ((ret = samread(fp, b)) >= 0) {
// bam_plp_push() itself now filters out unmapped reads only
if (b->core.flag & mask) b->core.flag |= BAM_FUNMAP;
bam_plbuf_push(b, buf);
}
bam_plbuf_push(0, buf);
bam_plbuf_destroy(buf);
bam_destroy1(b);
return 0;
}
int main(int argc, char *argv[])
{
char *progname;
char *bamfilename;
int32_t tid;
samfile_t *bamin;
bam_index_t *bamidx;
bam_plbuf_t *buf;
bam1_t *bam_read;
uint32_t next_pos = 1;
progname = *argv;
argv++; argc--;
if (argc < 2) {
printf("Usage: %s bam_file tid\n", progname);
exit(1);
}
else {
bamfilename = argv[0];
tid = strtol(argv[1], NULL, 10);
}
/* try to open bam file */
bamin = samopen(bamfilename, "rb", NULL);
if (!bamin) {
fprintf(stderr, "Error opening bamfile %s\n", bamfilename);
exit(1);
}
/* try to open index */
bamidx = bam_index_load(bamfilename);
if (!bamidx) {
fprintf(stderr, "Error opening index for %s\n", bamfilename);
exit(1);
}
bam_read = bam_init1();
buf = bam_plbuf_init(&pileup_func, &next_pos);
/* disable maximum pileup depth */
bam_plp_set_maxcnt(buf->iter, INT_MAX);
bam_fetch(bamin->x.bam, bamidx,
tid, 0, INT_MAX,
buf, &fetch_func);
bam_plbuf_push(0, buf); /* finish pileup */
bam_plbuf_destroy(buf);
bam_destroy1(bam_read);
bam_index_destroy(bamidx);
samclose(bamin);
return 0;
}
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;
}
void mapper( char *ref, int length, int start_base_pos, const char *bam )
{
anal_t input;
gzFile pRef;
kseq_t * seq = NULL;
char chr[8] = { 0, };
int ret;
bam_plbuf_t *buf;
bam1_t *b;
/*
fprintf( stderr, "ref: %s\n", ref );
fprintf( stderr, "length: %d\n", length );
fprintf( stderr, "start_base_pos: %d\n", start_base_pos );
fprintf( stderr, "bam: %s\n", bam );
*/
input.beg = 0; input.end = 0x7fffffff;
input.in = samopen(bam, "rb", 0);
if (input.in == 0)
{
fprintf(stderr, "Fail to open BAM file %s\n", bam);
return;
}
pRef = gzopen( ref, "r" );
fprintf( stderr, "ref : %s\n", ref );
fprintf( stderr, "pRef: %p\n", pRef );
if( pRef == NULL )
{
fprintf( stderr, "ref : %s\n", ref );
fprintf( stderr, "pRef: %p\n", pRef );
return;
}
seq = kseq_init( pRef );
b = bam_init1(); // alloc memory size of bam1_t
//fprintf( stderr, "%\pn", b );
buf = bam_plbuf_init(pileup_func, &input); // alloc memory
bam_plbuf_set_mask(buf, -1);
while ((ret = samread( input.in, b)) >= 0)
{
bam_plbuf_push(b, buf);
//fprintf( stderr, "%x\n", b->core.flag );
if( b->core.flag & 0x0004 ) // unmapped
{ // do nothing
/*
qname1 = strtok(bam1_qname(b), ":\t\n ");
qname2 = strtok(NULL, ":\t\n ");
qname3 = atoi(qname2);
fprintf( stderr, "%s:%10d:%s:%d\t%c:%d:%d:%d\n",
qname1, qname3, "*", b->core.pos,
'*', b->core.flag, b->core.qual, ret );
*/
fprintf( stdout, "%s:%s:%d\t%c:0x%x:%d:%d\n",
bam1_qname(b), "*", b->core.pos+1,
'*', b->core.flag, b->core.qual, ret );
/*
fprintf( stderr, "%s:%s:%d\t%c:0x%x:%d:%d\n",
bam1_qname(b), "*", b->core.pos,
'*', b->core.flag, b->core.qual, ret );
*/
}
else
{
// to find a base in the reference genome, seq.
if( ( seq != NULL ) &&
( strcmp( input.in->header->target_name[b->core.tid], chr ) == 0 ) )
{
// already found that
// fprintf( stderr, "found : %s\n", chr );
}else
{
if( find_chr(input.in->header->target_name[b->core.tid], seq, chr) < 0 )
{
fprintf( stderr, "ERROR : cannot find chromosome %s\n", \
input.in->header->target_name[b->core.tid] );
}else
{
fprintf( stderr, "FOUND CHR : %s\n", chr );
}
}
// remove not aligned to the chromosome
fprintf( stdout, "%s:%s:%d\t%c:%d:%d:%d\n",
bam1_qname(b),
input.in->header->target_name[b->core.tid],
b->core.pos+1,
seq->seq.s[b->core.pos], b->core.flag, b->core.qual, ret );
/*
fprintf( stderr, "%s:%s:%d\t%c:%d:%d:%d\n",
bam1_qname(b),
input.in->header->target_name[b->core.tid],
b->core.pos,
seq->seq.s[b->core.pos], b->core.flag, b->core.qual, ret );
*/
}
}
// for the last bases...
// printf("pos:%d(%c), flag:%d qual: %d(ret %d)\n",
// b->core.pos+1, seq->seq.s[b->core.pos], b->core.flag, b->core.qual, ret );
bam_plbuf_push(0, buf);
bam_plbuf_destroy(buf); // release memory
bam_destroy1(b); // release memory size of bam1_t
samclose(input.in);
kseq_destroy( seq );
gzclose( pRef );
return;
}
int bam_lplbuf_push(const bam1_t *b, bam_lplbuf_t *tv)
{
return bam_plbuf_push(b, tv->plbuf);
}
// callback for bam_fetch()
static int fetch_func(const bam1_t *b, void *data)
{
bam_plbuf_t *buf = (bam_plbuf_t*)data;
bam_plbuf_push(b, buf);
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;
}