int
main(int argc, char **argv)
{
ESL_GETOPTS *go = p7_CreateDefaultApp(options, 1, argc, argv, banner, usage);
char *msafile = esl_opt_GetArg(go, 1);
int fmt = eslMSAFILE_UNKNOWN;
ESL_ALPHABET *abc = NULL;
ESL_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
P7_HMM *hmm = NULL;
P7_PRIOR *prior = NULL;
P7_TRACE **trarr = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
ESL_MSA *postmsa = NULL;
int i;
int status;
/* Standard idioms for opening and reading a digital MSA. (See esl_msafile.c example). */
if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
if ((status = esl_msafile_Open(&abc, msafile, NULL, fmt, NULL, &afp)) != eslOK)
esl_msafile_OpenFailure(afp, status);
bg = p7_bg_Create(abc);
switch (abc->type) {
case eslAMINO: prior = p7_prior_CreateAmino(); break;
case eslDNA: prior = p7_prior_CreateNucleic(); break;
case eslRNA: prior = p7_prior_CreateNucleic(); break;
default: prior = p7_prior_CreateLaplace(abc); break;
}
if (prior == NULL) esl_fatal("Failed to initialize prior");
while ((status = esl_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) esl_msafile_ReadFailure(afp, status);
/* The modelmakers collect counts in an HMM structure */
status = p7_Handmodelmaker(msa, NULL, &hmm, &trarr);
if (status == eslENORESULT) esl_fatal("no consensus columns in alignment %s\n", msa->name);
else if (status != eslOK) esl_fatal("failed to build HMM from alignment %s\n", msa->name);
printf("COUNTS:\n");
p7_hmm_Dump(stdout, hmm);
/* These counts, in combination with a prior, are converted to probability parameters */
status = p7_ParameterEstimation(hmm, prior);
if (status != eslOK) esl_fatal("failed to parameterize HMM for %s", msa->name);
printf("PROBABILITIES:\n");
p7_hmm_Dump(stdout, hmm);
/* Just so we can dump a more informatively annotated trace - build a profile */
gm = p7_profile_Create(hmm->M, abc);
p7_profile_Config (gm, hmm, bg);
p7_profile_SetLength(gm, 400);
/* Dump the individual traces */
for (i = 0; i < msa->nseq; i++)
{
printf("Trace %d: %s\n", i+1, msa->sqname[i]);
p7_trace_DumpAnnotated(stdout, trarr[i], gm, msa->ax[i]);
}
/* Create an MSA from the individual traces */
status = p7_tracealign_MSA(msa, trarr, hmm->M, p7_DEFAULT, &postmsa);
if (status != eslOK) esl_fatal("failed to create new MSA from traces\n");
esl_msafile_Write(stdout, postmsa, eslMSAFILE_PFAM);
p7_profile_Destroy(gm);
p7_hmm_Destroy(hmm);
p7_trace_DestroyArray(trarr, msa->nseq);
esl_msa_Destroy(postmsa);
esl_msa_Destroy(msa);
}
esl_msafile_Close(afp);
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
/* Function: p7_builder_Create()
* Synopsis: Create a default HMM construction configuration.
*
* Purpose: Create a construction configuration for building
* HMMs in alphabet <abc>, and return a pointer to it.
*
* An application configuration <go> may optionally be
* provided. If <go> is <NULL>, default parameters are
* used. If <go> is non-<NULL>, it must include appropriate
* settings for all of the following ``standard build options'':
*
* Model construction: --fast --hand --symfrac --fragthresh
* Relative weighting: --wgsc --wblosum --wpb --wgiven --wid
* Effective seq #: --eent --eclust --enone --eset --ere --esigma --eid
* Prior scheme: --pnone --plaplace
* E-val calibration: --EmL --EmN --EvL --EvN --EfL --EfN --Eft
* run-to-run variation: --seed
*
* See <hmmbuild.c> or other big users of the build
* pipeline for an example of appropriate <ESL_GETOPTS>
* initializations of these 24 options.
*/
P7_BUILDER *
p7_builder_Create(const ESL_GETOPTS *go, const ESL_ALPHABET *abc)
{
P7_BUILDER *bld = NULL;
int seed;
int status;
ESL_ALLOC(bld, sizeof(P7_BUILDER));
bld->prior = NULL;
bld->r = NULL;
bld->S = NULL;
bld->Q = NULL;
bld->eset = -1.0; /* -1.0 = unset; must be set if effn_strategy is p7_EFFN_SET */
bld->re_target = -1.0;
if (go == NULL)
{
bld->arch_strategy = p7_ARCH_FAST;
bld->wgt_strategy = p7_WGT_PB;
bld->effn_strategy = p7_EFFN_ENTROPY;
seed = 42;
}
else
{
if (esl_opt_GetBoolean(go, "--fast")) bld->arch_strategy = p7_ARCH_FAST;
else if (esl_opt_GetBoolean(go, "--hand")) bld->arch_strategy = p7_ARCH_HAND;
if (esl_opt_GetBoolean(go, "--wpb")) bld->wgt_strategy = p7_WGT_PB;
else if (esl_opt_GetBoolean(go, "--wgsc")) bld->wgt_strategy = p7_WGT_GSC;
else if (esl_opt_GetBoolean(go, "--wblosum")) bld->wgt_strategy = p7_WGT_BLOSUM;
else if (esl_opt_GetBoolean(go, "--wnone")) bld->wgt_strategy = p7_WGT_NONE;
else if (esl_opt_GetBoolean(go, "--wgiven")) bld->wgt_strategy = p7_WGT_GIVEN;
if (esl_opt_GetBoolean(go, "--eent")) bld->effn_strategy = p7_EFFN_ENTROPY;
else if (esl_opt_GetBoolean(go, "--eclust")) bld->effn_strategy = p7_EFFN_CLUST;
else if (esl_opt_GetBoolean(go, "--enone")) bld->effn_strategy = p7_EFFN_NONE;
else if (esl_opt_IsOn (go, "--eset")) { bld->effn_strategy = p7_EFFN_SET; bld->eset = esl_opt_GetReal(go, "--eset"); }
seed = esl_opt_GetInteger(go, "--seed");
}
bld->max_insert_len = 0;
/* The default RE target is alphabet dependent. */
if (go != NULL && esl_opt_IsOn (go, "--ere"))
bld->re_target = esl_opt_GetReal(go, "--ere");
else {
switch (abc->type) {
case eslAMINO: bld->re_target = p7_ETARGET_AMINO; break;
case eslDNA: bld->re_target = p7_ETARGET_DNA; break;
case eslRNA: bld->re_target = p7_ETARGET_DNA; break;
default: bld->re_target = p7_ETARGET_OTHER; break;
}
}
bld->symfrac = (go != NULL) ? esl_opt_GetReal (go, "--symfrac") : 0.5;
bld->fragthresh = (go != NULL) ? esl_opt_GetReal (go, "--fragthresh") : 0.5;
bld->wid = (go != NULL) ? esl_opt_GetReal (go, "--wid") : 0.62;
bld->esigma = (go != NULL) ? esl_opt_GetReal (go, "--esigma") : 45.0;
bld->eid = (go != NULL) ? esl_opt_GetReal (go, "--eid") : 0.62;
bld->EmL = (go != NULL) ? esl_opt_GetInteger(go, "--EmL") : 200;
bld->EmN = (go != NULL) ? esl_opt_GetInteger(go, "--EmN") : 200;
bld->EvL = (go != NULL) ? esl_opt_GetInteger(go, "--EvL") : 200;
bld->EvN = (go != NULL) ? esl_opt_GetInteger(go, "--EvN") : 200;
bld->EfL = (go != NULL) ? esl_opt_GetInteger(go, "--EfL") : 100;
bld->EfN = (go != NULL) ? esl_opt_GetInteger(go, "--EfN") : 200;
bld->Eft = (go != NULL) ? esl_opt_GetReal (go, "--Eft") : 0.04;
/* Normally we reinitialize the RNG to original seed before calibrating each model.
* This eliminates run-to-run variation.
* As a special case, seed==0 means choose an arbitrary seed and shut off the
* reinitialization; this allows run-to-run variation.
*/
bld->r = esl_randomness_CreateFast(seed);
bld->do_reseeding = (seed == 0) ? FALSE : TRUE;
if (go && esl_opt_GetBoolean(go, "--pnone") ) bld->prior = NULL;
else if (go && esl_opt_GetBoolean(go, "--plaplace") ) bld->prior = p7_prior_CreateLaplace(abc);
else
{
switch (abc->type) {
case eslAMINO: bld->prior = p7_prior_CreateAmino(); break;
case eslDNA: bld->prior = p7_prior_CreateNucleic(); break;
case eslRNA: bld->prior = p7_prior_CreateNucleic(); break;
default: bld->prior = p7_prior_CreateLaplace(abc); break;
}
if (bld->prior == NULL) goto ERROR;
}
bld->abc = abc;
bld->errbuf[0] = '\0';
bld->popen = -1;
bld->pextend = -1;
return bld;
ERROR:
p7_builder_Destroy(bld);
return NULL;
}
