/* Evaluate fx = rel entropy - etarget, which we want to be = 0,
* for effective sequence number <x>.
*/
static int
eweight_target_f(double Neff, void *params, double *ret_fx)
{
struct ew_param_s *p = (struct ew_param_s *) params;
p7_hmm_CopyParameters(p->hmm, p->h2);
p7_hmm_Scale(p->h2, Neff / (double) p->h2->nseq);
p7_ParameterEstimation(p->h2, p->pri);
*ret_fx = p7_MeanMatchRelativeEntropy(p->h2, p->bg) - p->etarget;
return eslOK;
}
/* set_effective_seqnumber()
*
* <hmm> comes in with weighted observed counts. It goes out with
* those observed counts rescaled to sum to the "effective sequence
* number".
*
* <msa> is needed because we may need to see the sequences in order
* to determine effective seq #. (for --eclust)
*
* <prior> is needed because we may need to parameterize test models
* looking for the right relative entropy. (for --eent, the default)
*/
static int
effective_seqnumber(P7_BUILDER *bld, const ESL_MSA *msa, P7_HMM *hmm, const P7_BG *bg)
{
int status;
if (bld->effn_strategy == p7_EFFN_NONE) hmm->eff_nseq = msa->nseq;
else if (bld->effn_strategy == p7_EFFN_SET) hmm->eff_nseq = bld->eset;
else if (bld->effn_strategy == p7_EFFN_CLUST)
{
int nclust;
status = esl_msacluster_SingleLinkage(msa, bld->eid, NULL, NULL, &nclust);
if (status == eslEMEM) ESL_XFAIL(status, bld->errbuf, "memory allocation failed");
else if (status != eslOK) ESL_XFAIL(status, bld->errbuf, "single linkage clustering algorithm (at %d%% id) failed", (int)(100 * bld->eid));
hmm->eff_nseq = (double) nclust;
}
else if (bld->effn_strategy == p7_EFFN_ENTROPY)
{
double etarget;
double eff_nseq;
etarget = (bld->esigma - eslCONST_LOG2R * log( 2.0 / ((double) hmm->M * (double) (hmm->M+1)))) / (double) hmm->M; /* xref J5/36. */
etarget = ESL_MAX(bld->re_target, etarget);
status = p7_EntropyWeight(hmm, bg, bld->prior, etarget, &eff_nseq);
if (status == eslEMEM) ESL_XFAIL(status, bld->errbuf, "memory allocation failed");
else if (status != eslOK) ESL_XFAIL(status, bld->errbuf, "internal failure in entropy weighting algorithm");
hmm->eff_nseq = eff_nseq;
}
p7_hmm_Scale(hmm, hmm->eff_nseq / (double) hmm->nseq);
return eslOK;
ERROR:
return status;
}
