value caml_bgzf_set_cache_size(value bgzf, value sz) {
CAMLparam2(bgzf,sz);
bgzf_set_cache_size(BGZF_val(bgzf),Val_int(sz));
CAMLreturn(Val_unit);
}
int main (int argc, char **argv) {
/////////////////////
// Parse Arguments //
/////////////////////
params *pars = new params;
init_pars(pars);
parse_cmd_args(argc, argv, pars);
if( pars->version ) {
printf("ngsF v%s\nCompiled on %s @ %s", version, __DATE__, __TIME__);
#ifdef _USE_BGZF
printf(" (BGZF library)\n");
#else
printf(" (STD library)\n");
#endif
exit(0);
}
if( pars->verbose >= 1 ) {
printf("==> Input Arguments:\n");
printf("\tglf file: %s\n\tinit_values: %s\n\tfreq_fixed: %s\n\tout file: %s\n\tn_ind: %d\n\tn_sites: %lu\n\tchunk_size: %lu\n\tfast_lkl: %s\n\tapprox_EM: %s\n\tcall_geno: %s\n\tmax_iters: %d\n\tmin_epsilon: %.10f\n\tn_threads: %d\n\tseed: %lu\n\tquick: %s\n\tversion: %s\n\tverbose: %d\n\n",
pars->in_glf, pars->init_values, pars->freq_fixed ? "true":"false", pars->out_file, pars->n_ind, pars->n_sites, pars->max_chunk_size, pars->fast_lkl ? "true":"false", pars->approx_EM ? "true":"false", pars->call_geno ? "true":"false", pars->max_iters, pars->min_epsilon, pars->n_threads, pars->seed, pars->quick ? "true":"false", version, pars->verbose);
}
if( pars->verbose > 4 ) printf("==> Verbose values greater than 4 for debugging purpose only. Expect large amounts of info on screen\n");
/////////////////////
// Check Arguments //
/////////////////////
if(pars->in_glf == NULL)
error(__FUNCTION__,"GL input file (-glf) missing!");
else if( strcmp(pars->in_glf, "-") == 0 ) {
pars->in_glf_type = new char[6];
pars->in_glf_type = strcat(pars->in_glf_type, "STDIN");
} else {
pars->in_glf_type = strrchr(pars->in_glf, '.');
if(pars->in_glf_type == NULL)
error(__FUNCTION__,"invalid file type!");
}
if(pars->out_file == NULL)
error(__FUNCTION__,"output file (-out) missing!");
if(pars->n_ind == 0)
error(__FUNCTION__,"number of individuals (-n_ind) missing!");
if(pars->n_sites == 0)
error(__FUNCTION__,"number of sites (-n_sites) missing!");
///////////////////////
// Check input files //
///////////////////////
// Get file total size
struct stat st;
stat(pars->in_glf, &st);
if( strcmp(pars->in_glf_type, "STDIN") != 0 ) {
if( pars->n_sites == st.st_size/sizeof(double)/pars->n_ind/3 && strcmp(pars->in_glf_type, ".glf") == 0 ) {
if(pars->verbose >= 1)
printf("==> UNCOMP input file (\"%s\"): number of sites (%lu) match expected file size\n", pars->in_glf_type, pars->n_sites);
} else if( strcmp(pars->in_glf_type, ".glf") != 0 ) {
if( pars->verbose >= 1)
printf("==> COMPRESSED input file (\"%s\"): number of sites (%lu) do NOT match expected file size\n", pars->in_glf_type, pars->n_sites);
} else
error(__FUNCTION__,"wrong number of sites or invalid/corrupt file!");
}
// Adjust max_chunk_size in case of fewer sites
if(pars->max_chunk_size > pars->n_sites) {
if( pars->verbose >= 1 ) printf("==> Fewer sites (%lu) than chunk_size (%lu). Reducing chunk size to match number of sites\n", pars->n_sites, pars->max_chunk_size);
pars->max_chunk_size = pars->n_sites;
}
// Calculate total number of chunks
pars->n_chunks = ceil( (double) pars->n_sites/ (double) pars->max_chunk_size );
if( pars->verbose >= 1 ) printf("==> Analysis will be run in %ld chunk(s)\n", pars->n_chunks);
// Alocate memory for the chunk index
pars->chunks_voffset = new int64_t[pars->n_chunks];
memset(pars->chunks_voffset, 0, pars->n_chunks*sizeof(int64_t));
// Adjust thread number to chunks
if(pars->n_chunks < pars->n_threads) {
if( pars->verbose >= 1 ) printf("==> Fewer chunks (%ld) than threads (%d). Reducing the number of threads to match number of chunks\n", pars->n_chunks, pars->n_threads);
pars->n_threads = pars->n_chunks;
}
// Open input file
#ifdef _USE_BGZF
if( pars->verbose >= 1 ) printf("==> Using BGZF I/O library\n");
// Open BGZIP file
if( strcmp(pars->in_glf_type, ".bgz") == 0 ) {
if( (pars->in_glf_fh = bgzf_open(pars->in_glf, "rb")) < 0 )
error(__FUNCTION__,"Cannot open BGZIP file!");
} else
error(__FUNCTION__,"BGZF library only supports BGZIP files!");
bgzf_set_cache_size(pars->in_glf_fh, CACHE_SIZE * 1024uL * 1024uL * 1024uL);
#else
if( pars->verbose >= 1 ) printf("==> Using native I/O library\n");
// Open GLF file
if( strcmp(pars->in_glf_type, "STDIN") == 0 )
pars->in_glf_fh = stdin;
else if( strcmp(pars->in_glf_type, ".glf") == 0 ) {
if( (pars->in_glf_fh = fopen(pars->in_glf, "rb")) == NULL )
error(__FUNCTION__,"Cannot open GLF file!");
} else
error(__FUNCTION__,"Standard library only supports UNCOMPRESSED GLF files!");
// Allocate memory and read from the file
pars->data = new double* [pars->n_sites];
for(uint64_t s = 0; s < pars->n_sites; s++) {
pars->data[s] = new double[pars->n_ind * 3];
if( fread (pars->data[s], sizeof(double), pars->n_ind * 3, pars->in_glf_fh) != pars->n_ind * 3)
error(__FUNCTION__,"cannot read GLF file!");
if(pars->call_geno)
call_geno(pars->data[s], pars->n_ind, 3);
}
#endif
if( pars->in_glf_fh == NULL )
error(__FUNCTION__,"cannot open GLF file!");
///////////////////////////////////
// Declare variables for results //
///////////////////////////////////
out_data *output = new out_data;
output->site_freq = new double[pars->n_sites];
output->site_freq_num = new double[pars->n_sites];
output->site_freq_den = new double[pars->n_sites];
output->site_prob_var = new double[pars->n_sites];
output->site_tmpprob_var = new double[pars->n_sites];
output->indF = new double[pars->n_ind];
output->indF_num = new double[pars->n_ind];
output->indF_den = new double[pars->n_ind];
output->ind_lkl = new double[pars->n_ind];
// Initialize output
init_output(pars, output);
//////////////////
// Analyze Data //
//////////////////
if( pars->verbose >= 1 && !pars->fast_lkl && strcmp("e", pars->init_values) != 0 ) {
printf("==> Initial LogLkl: %.15f\n", full_HWE_like(pars, output->site_freq, output->indF, 0, pars->n_ind));
fflush(stdout);
}
do_EM(pars, output);
if( pars->verbose >= 1 ) printf("\nFinal logLkl: %f\n", output->global_lkl);
//////////////////
// Print Output //
//////////////////
FILE *out_file;
if( pars->verbose >= 1 ) printf("Printing Output...\n");
out_file = fopen(pars->out_file, "w");
if(out_file == NULL)
error(__FUNCTION__,"Cannot open OUTPUT file!");
for(uint16_t i = 0; i < pars->n_ind; i++)
fprintf(out_file,"%f\n", output->indF[i]);
fclose(out_file);
//////////////////////
// Close Input File //
//////////////////////
if( pars->verbose >= 1 ) printf("Exiting...\n");
#ifdef _USE_BGZF
bgzf_close(pars->in_glf_fh);
#else
for(uint64_t s = 0; s < pars->n_sites; s++)
delete [] pars->data[s];
delete [] pars->data;
fclose(pars->in_glf_fh);
#endif
/////////////////
// Free Memory //
/////////////////
delete [] output->site_freq;
delete [] output->site_freq_num;
delete [] output->site_freq_den;
delete [] output->site_prob_var;
delete [] output->indF;
delete [] output->indF_num;
delete [] output->indF_den;
delete [] output->ind_lkl;
delete output;
//if( strcmp("e", pars->init_values) == 0 )
//delete [] pars->init_values;
delete [] pars->chunks_voffset;
delete pars;
return 0;
}
int main_bedcov(int argc, char *argv[])
{
extern void bam_init_header_hash(bam_header_t*);
gzFile fp;
kstring_t str;
kstream_t *ks;
bam_index_t **idx;
bam_header_t *h = 0;
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(bam_index_t*));
for (i = 0; i < n; ++i) {
aux[i] = calloc(1, sizeof(aux_t));
aux[i]->min_mapQ = min_mapQ;
aux[i]->fp = bam_open(argv[i+optind+1], "r");
idx[i] = bam_index_load(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;
}
bgzf_set_cache_size(aux[i]->fp, 20);
if (i == 0) h = bam_header_read(aux[0]->fp);
}
bam_init_header_hash(h);
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_get_tid(h, 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) bam_iter_destroy(aux[i]->iter);
aux[i]->iter = bam_iter_query(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) bam_iter_destroy(aux[i]->iter);
bam_index_destroy(idx[i]);
bam_close(aux[i]->fp);
free(aux[i]);
}
bam_header_destroy(h);
free(aux); free(idx);
free(str.s);
return 0;
}
