/* Gera uma velocidade seguindo uma normal com media media e desvio
* padrao desv_pad. */
double rand_velocity(double media, double desv_pad) {
double z = gen_norm();
return (desv_pad*z + media);
}
int generate(int count)
{
FILE *ofp;
int length; /* length of seqence */
double probs[20]; /* probability mass function for AAs */
double cum_probs[20]; /* probability distribution
* (cumulative sum of probs)
*/
int s; /* counter for sequences */
int p; /* counter for position in sequence */
int a; /* counter for amino acid number */
int alo, ahi, amid; /* variables for binary search to get amino acid */
double sum; /* temporary for computing cum_probs */
double x; /* uniform random number */
gen_dirch_mix_param comp_gen; /* composition generator */
ofp = fopen("GeneratedSequences.txt", "w");
if (ofp == NULL) {
fprintf(stderr, "Can't open output file\n");
return 1;
}
srandom(getpid());
gen_dirch_mix_initialize(&comp_gen, 20, 6, mix_coeff,
(const double **)comps);
for (s=0; s<count; s++)
{
length = (int) exp(mean_log_length +stddev_log_length*gen_norm());
gen_dirch_mix(&comp_gen, probs);
sum = 0.;
/* convert probs to cumulative probabilities */
for (a=0; a<20; a++)
{ sum += probs[a];
cum_probs[a] =sum;
}
for (p=0; p< length; p++)
{ x = DRAND();
/* do binary search to determine amino acid */
alo = -1; ahi=19;
while (alo<ahi-1)
{ /* invariant:
* cum_probs[alo] < x <= cum_probs[ahi]
*/
assert (x <= cum_probs[ahi]);
amid = (alo+ahi+1)/2;
if (x > cum_probs[amid]) alo=amid;
else ahi=amid;
}
if ( fputc(AA[ahi], ofp) == EOF ) return 1;
}
if ( fputc('\n', ofp) == EOF ) return 1;
}
if ( fclose(ofp) != EOF ) return 0;
else return 1;
}
