static double randu(double a, double b)
{
return a + random_uniform() * (b - a);
}
// Returns num of bases printed
size_t sim_reads(seq_file_t *reffile, gzFile out0, gzFile out1,
FileList *flist, float err_rate,
size_t insert, double insert_stddev, size_t rlen, double depth)
{
size_t i, chromcap = 16, nchroms, glen = 0, nreads, chr, pos0, pos1, tlen;
read_t *chroms;
tlen = rlen + (out1 == NULL ? 0 : insert + rlen);
chroms = malloc(chromcap * sizeof(read_t));
nchroms = 0;
// Load genome
printf(" Loaded contigs:");
while(1)
{
if(nchroms == chromcap) chroms = realloc(chroms, (chromcap*=2)*sizeof(read_t));
seq_read_alloc(&chroms[nchroms]);
if(seq_read(reffile, &chroms[nchroms]) <= 0)
{ seq_read_dealloc(&chroms[nchroms]); break; }
if(chroms[nchroms].seq.end < tlen) { seq_read_dealloc(&chroms[nchroms]); }
else {
seq_read_truncate_name(&chroms[nchroms]);
printf(" %s[%zu]", chroms[nchroms].name.b, chroms[nchroms].seq.end);
glen += chroms[nchroms].seq.end;
nchroms++;
}
}
printf("\n Genome size: %zu\n", glen);
if(nchroms == 0) {
die("No sequences long enough in ref genome file [min len: %zu]: %s",
tlen, reffile->path);
}
// Sample
nreads = (glen * depth) / (out1 == NULL ? rlen : (2 * rlen));
char read0[rlen+1], read1[rlen+1];
read0[rlen] = read1[rlen] = '\0';
printf("Sampling %zu %sreads...\n", nreads,
out1 == NULL ? "single " : "paired-end ");
// Sample paired-end if out1 != NULL
for(i = 0; i < nreads; i++)
{
chr = (nchroms == 1) ? 0 : rand_chrom(chroms, nchroms, glen);
pos0 = random_uniform(chroms[chr].seq.end - (out1 == NULL ? rlen : tlen));
pos1 = pos0;
memcpy(read0, chroms[chr].seq.b+pos0, rlen);
if(out1 != NULL) {
pos1 = pos0 + rlen + insert + ran_normal()*insert_stddev;
if(pos1 + rlen > chroms[chr].seq.end) pos1 = chroms[chr].seq.end-rlen;
memcpy(read1, chroms[chr].seq.b+pos1, rlen);
}
if(flist != NULL) {
add_seq_error_profile(read0, rlen, flist);
if(out1 != NULL)
add_seq_error_profile(read1, rlen, flist);
}
else if(err_rate >= 0) {
add_seq_error_rate(read0, rlen, err_rate);
}
gzprintf(out0, ">r%zu:%s:%zu:%zu%s\n%.*s\n", i, chroms[chr].name.b,
pos0, pos1, (out1 != NULL ? "/1" : ""), (int)rlen, read0);
if(out1 != NULL) {
dna_revcmp(read1, rlen);
gzprintf(out1, ">r%zu:%s:%zu:%zu/2\n%.*s\n", i, chroms[chr].name.b,
pos0, pos1, (int)rlen, read1);
}
}
for(i = 0; i < nchroms; i++) seq_read_dealloc(&chroms[i]);
free(chroms);
size_t num_bases = nreads * rlen;
if(out1 != NULL) num_bases *= 2;
return num_bases;
}
/** sanitize
Sanitizes a gridlabd model by clear names and position from object headers
@returns 0 on success, -2 on error
**/
extern "C" int sanitize(int argc, char *argv[])
{
OBJECT *obj;
FILE *fp;
double delta_latitude, delta_longitude;
// lat/lon change
if ( strcmp(global_sanitizeoffset,"")==0 )
{
delta_latitude = random_uniform(NULL,-5,+5);
delta_longitude = random_uniform(NULL,-180,+180);
}
else if ( global_sanitizeoffset=="destroy" )
delta_latitude = delta_longitude = QNAN;
else if ( sscanf(global_sanitizeoffset.get_string(),"%lf%*[,/]%lf",&delta_latitude,&delta_longitude)!=2 )
{
output_error("sanitize_offset lat/lon '%s' is not valid", global_sanitizeoffset.get_string());
return -2;
}
// sanitize object names
for ( obj=object_get_first() ; obj!=NULL ; obj=object_get_next(obj) )
{
if ( obj->name!=NULL && (global_sanitizeoptions&SO_NAMES)==SO_NAMES )
sanitize_name(obj);
if ( isfinite(obj->latitude) && (global_sanitizeoptions&SO_GEOCOORDS)==SO_GEOCOORDS )
{
obj->latitude += delta_latitude;
if ( obj->latitude<-90 ) obj->latitude = -90;
if ( obj->latitude>+90 ) obj->latitude = +90;
}
if ( isfinite(obj->longitude) && (global_sanitizeoptions&SO_GEOCOORDS)==SO_GEOCOORDS )
obj->longitude = fmod(obj->longitude+delta_longitude,360);
}
// dump object name index
if ( strcmp(global_sanitizeindex,".xml")==0 )
{
strcpy(global_sanitizeindex,global_modelname);
char *ext = strrchr(global_sanitizeindex,'.');
if ( ext && strcmp(ext,".glm")==0 ) strcpy(ext,"-index.xml");
else strcat(global_sanitizeindex,"-index.xml");
}
else if ( strcmp(global_sanitizeindex,".txt")==0 )
{
strcpy(global_sanitizeindex,global_modelname);
char *ext = strrchr(global_sanitizeindex,'.');
if ( ext && strcmp(ext,".glm")==0 ) strcpy(ext,"-index.txt");
else strcat(global_sanitizeindex,"-index.txt");
}
else if ( global_sanitizeindex[0]=='.' )
{
output_error("sanitization index file spec '%s' is not recognized", global_sanitizeindex.get_string());
return -2;
}
if ( strcmp(global_sanitizeindex,"")!=0 )
{
char *ext = strrchr(global_sanitizeindex,'.');
bool use_xml = (ext && strcmp(ext,".xml")==0) ;
fp = fopen(global_sanitizeindex,"w");
if ( fp )
{
SAFENAME *item;
if ( use_xml )
{
fprintf(fp,"<data>\n");
fprintf(fp,"\t<modelname>%s</modelname>\n",global_modelname);
fprintf(fp,"\t<geographic_offsets>\n");
fprintf(fp,"\t\t<latitude>%.6f</latitude>\n",delta_latitude);
fprintf(fp,"\t\t<longitude>%.6f</longitude>\n",delta_longitude);
fprintf(fp,"\t</geographic_offsets>\n");
fprintf(fp,"\t<safename_list>\n");
for ( item=safename_list ; item!=NULL ; item=item->next )
fprintf(fp,"\t\t<name>\n\t\t\t<safe>%s</safe>\n\t\t\t<unsafe>%s</unsafe>\n\t\t</name>\n", item->name, item->old);
fprintf(fp,"\t</safename_list>\n");
fprintf(fp,"</data>\n");
}
else
{
fprintf(fp,"modelname\t= %s\n", global_modelname);
fprintf(fp,"\n[POSITIONS]\n");
fprintf(fp,"latitude\t= %.6f\n",delta_latitude);
fprintf(fp,"longitude\t= %.6f\n",delta_longitude);
fprintf(fp,"\n[NAMES]\n");
for ( item=safename_list ; item!=NULL ; item=item->next )
fprintf(fp,"%s\t= %s\n", item->name, item->old);
}
fclose(fp);
}
}
return 0;
}
int main(int argc, char** argv) {
KSOM_ERROR e = KSOM_OK;
ksom_map map;
ksom_map_options opt;
double random_number = 0.0;
double test [3];
opt.fv_size = 3;
opt.x_dim = 5;
opt.y_dim = 5;
opt.init_k_value = 0.1;
opt.num_iterations = 100;
//test random number generation functions
printf("testing random\n");
printf("initializing random... ");
e = random_init();
if (KSOM_FAIL(e)) goto end;
printf("initialized.\n");
printf("testing random (i=1000)... ");
for (int i = 0; i < 1000; ++i) {
random_number = random_uniform();
if (random_number >= 1.0 || random_number <= 0.0) {
printf("(iteration %d)error: number is %f\n", i, random_number);
e = KSOM_INTERNAL;
goto end;
}
}
puts("ok.");
puts("random ok.");
printf("initializing map... ");
e = ksom_map_init(&map, &opt);
if (KSOM_FAIL(e)) goto end;
puts("done.");
for (int i = 0; i < map.num_nodes; ++i)
printf("%10f %10f %10f\n", map.nodes[i].fv[0], map.nodes[i].fv[1], map.nodes[i].fv[2]);
for (int i = 0; i < 10; ++i) {
test[0] = random_uniform();
test[1] = random_uniform();
test[2] = random_uniform();
printf("cycle %-3d... ", map.iteration_count);
e = ksom_map_cycle(&map, test);
if (KSOM_FAIL(e)) goto end;
puts("done.");
dump_map_fvdistances(&map);
}
printf("freeing map... ");
e = ksom_map_free(&map);
if (KSOM_FAIL(e)) goto end;
puts("done.");
end:
printf("\n[!] %s\n", ksom_error_str(e));
return (EXIT_SUCCESS);
}