int hivpopulation::write_genotypes(ostream &out, int sample_size, string gt_label, int start, int length){
if (HIVPOP_VERBOSE) cerr<<"hivpopulation::write_genotypes()...";
if (HIVPOP_VERBOSE) cerr<<"start = "<<start<<"...";
if (HIVPOP_VERBOSE) cerr<<"length = "<<length<<"...";
if (out.bad()){
cerr<<"hivpopulation::write_genotypes(): BAD OUTPUT FILE!"<<endl;
return HIVPOP_BADARG;
}else{
int gti;
string temp;
if (length <= 0)
length = number_of_loci - start;
produce_random_sample(sample_size);
if (sample_size>get_population_size()){
cerr<<"hivpopulation::write_genotypes(): requested sample size exceeds population size"<<endl;
return HIVPOP_BADARG;
}else{
for (int s=0; s<sample_size; s++){
gti=random_clone();
out <<">GT-"<<gt_label<<"_"<<gti<<'\n';
for (int i =start; i<start+length; i++ ){
if (population[gti].genotype[i]) out <<'1';
else out <<'0';
}
out<<'\n';
}
}
if (HIVPOP_VERBOSE) cerr<<"...done."<<endl;
return 0;
}
}
/**
* Computing the Fisher's Exact Test (FET) and the corresponding standard deviation for a given thread.
* @param threadarg: the arguments to the thread: a thread_data struct
*/
void mycompute(void *threadarg) {
struct thread_data *my_data;
int tid;
int start, stop, regend, num_windows, wsize, wstep, alen, blen, asize, bsize, totalpos, npos, nsamples, my_task_id = 0;
double perc;
int *apos;
int *bpos;
double *avals;
double *bvals;
double *scores;
double *stddev;
/* fetch data from threadarg */
my_data = (struct thread_data *) threadarg;
tid = my_data->thread_id;
regend = my_data->regend;
num_windows = my_data->num_windows;
wsize = my_data->wsize;
wstep = my_data->wstep;
apos = my_data->apos;
bpos = my_data->bpos;
avals = my_data->avals;
bvals = my_data->bvals;
alen = my_data->alen;
blen = my_data->blen;
perc = my_data->perc;
scores = my_data->scores;
stddev = my_data->stddev;
int wcount = 0;
int wstart;
int wstop;
int *aidx;
int *bidx;
int *f;
int *tmp;
double *results;
double *fetscores = NULL;
double *samples = NULL;
double *stdsamples;
/* get size of population a and b*/
asize = get_population_size(apos);
bsize = get_population_size(bpos);
wcount = 0;
start = 0;
stop = wsize;
nsamples = 100;
/* setup idx arrays */
aidx = (int*)malloc(2*sizeof(int));
bidx = (int*)malloc(2*sizeof(int));
/* setup arrays for calc. of FET */
f = (int*)malloc(4*sizeof(int));
tmp = (int*)malloc(4*sizeof(int));
results = (double*)malloc(2*sizeof(double));
stdsamples = (double*)malloc(nsamples*sizeof(double));
int idx = 0;
/* generate seed for the PRNG */
unsigned short state[3] = {0,0,0};
unsigned short seed = time(NULL) + (unsigned short) pthread_self();
memcpy(state, &seed, sizeof(seed));
// fetch new tasks and calculate FET
while (my_task_id < num_tasks) {
// try to fetch new task
pthread_mutex_lock (&mutexTASK_ID);
my_task_id= task_id;
task_id++;
pthread_mutex_unlock(&mutexTASK_ID);
// if no more tasks, break
if (my_task_id > num_tasks)
break;
// get chromosome position of the current task
get_positions(wsize, wstep, my_task_id, &start, &stop, regend);
// to include the last (possibly smaller) window
if (stop >= regend) {
stop = regend + wstep;
}
wstart = start;
wstop = start+wsize;
aidx[0] = 0; aidx[1] = 0;
bidx[0] = 0; bidx[1] = 0;
// calculate FET for each window
while (wstart + wsize <= stop) {
slide_right(aidx, apos, wstart, wstop, alen);
slide_right(bidx, bpos, wstart, wstop, blen);
npos = (aidx[1] - aidx[0])/asize;
if (npos > 0) {
fetscores = (double*)realloc(fetscores, npos*sizeof(double));
samples = (double*)realloc(samples, npos*sizeof(double));
idx = wstart/wstep;
fisher_exact_test(results, &(avals[aidx[0]]), &(bvals[bidx[0]]), asize, bsize, npos, f, tmp, samples, stdsamples, nsamples, fetscores, state, perc);
scores[idx] = results[0];
stddev[idx] = results[1];
}
wstart += wstep;
wstop += wstep;
wcount++;
}
}
/* cleanup */
free(aidx);
free(bidx);
free(f);
free(tmp);
free(results);
free(samples);
free(stdsamples);
free(fetscores);
}
