/* Function: p7_tracealign_MSA()
* Synopsis: Convert array of traces (for a previous MSA) to a new MSA.
* Incept: SRE, Mon Mar 2 18:18:22 2009 [Casa de Gatos]
*
* Purpose: Identical to <p7_tracealign_Seqs()> except that the trace
* array <tr> accompanies a digital multiple alignment <premsa>,
* rather than an array of digital sequences.
*
* This gets used in <p7_Builder()>, where we've
* constructed an array of faux traces directly from an
* input alignment, and we want to reconstruct the
* MSA that corresponds to what HMMER actually used
* to build its model (after trace doctoring to be
* compatible with Plan 7, and with <#=RF> annotation
* on assigned consensus columns).
*
* Xref: J4/102.
*/
int
p7_tracealign_MSA(const ESL_MSA *premsa, P7_TRACE **tr, int M, int optflags, ESL_MSA **ret_postmsa)
{
const ESL_ALPHABET *abc = premsa->abc;
ESL_MSA *msa = NULL; /* RETURN: new MSA */
int *inscount = NULL; /* array of max gaps between aligned columns */
int *matmap = NULL; /* matmap[k] = apos of match k matmap[1..M] = [1..alen] */
int *matuse = NULL; /* TRUE if an alignment column is associated with match state k [1..M] */
int idx; /* counter over sequences */
int alen; /* width of alignment */
int status;
if ((status = map_new_msa(tr, premsa->nseq, M, optflags, &inscount, &matuse, &matmap, &alen)) != eslOK) return status;
if (optflags & p7_DIGITIZE) { if ((status = make_digital_msa(NULL, premsa, tr, premsa->nseq, matuse, matmap, M, alen, optflags, &msa)) != eslOK) goto ERROR; }
else { if ((status = make_text_msa (NULL, premsa, tr, premsa->nseq, matuse, matmap, M, alen, optflags, &msa)) != eslOK) goto ERROR; }
if ((status = annotate_rf(msa, M, matuse, matmap)) != eslOK) goto ERROR;
if ((status = annotate_posterior_probability(msa, tr, matmap, M, optflags)) != eslOK) goto ERROR;
if (optflags & p7_DIGITIZE) rejustify_insertions_digital( msa, inscount, matmap, matuse, M);
else rejustify_insertions_text (abc, msa, inscount, matmap, matuse, M);
/* Transfer information from old MSA to new */
esl_msa_SetName (msa, premsa->name);
esl_msa_SetDesc (msa, premsa->desc);
esl_msa_SetAccession(msa, premsa->acc);
for (idx = 0; idx < premsa->nseq; idx++)
{
esl_msa_SetSeqName (msa, idx, premsa->sqname[idx]);
if (msa->sqacc) esl_msa_SetSeqAccession (msa, idx, premsa->sqacc[idx]);
if (msa->sqdesc) esl_msa_SetSeqDescription(msa, idx, premsa->sqdesc[idx]);
msa->wgt[idx] = premsa->wgt[idx];
}
if (premsa->flags & eslMSA_HASWGTS)
msa->flags |= eslMSA_HASWGTS;
free(inscount);
free(matmap);
free(matuse);
*ret_postmsa = msa;
return eslOK;
ERROR:
if (msa != NULL) esl_msa_Destroy(msa);
if (inscount != NULL) free(inscount);
if (matmap != NULL) free(matmap);
if (matuse != NULL) free(matuse);
*ret_postmsa = NULL;
return status;
}
/* set_msa_name()
* Make sure the alignment has a name; this name will
* then be transferred to the model.
*
* We can only do this for a single alignment in a file. For multi-MSA
* files, each MSA is required to have a name already.
*
* Priority is:
* 1. Use -n <name> if set, overriding any name the alignment might already have.
* 2. Use alignment's existing name, if non-NULL.
* 3. Make a name, from alignment file name without path and without filename extension
* (e.g. "/usr/foo/globins.slx" gets named "globins")
* If none of these succeeds, return <eslEINVAL>.
*
* If a multiple MSA database (e.g. Stockholm/Pfam), and we encounter
* an MSA that doesn't already have a name, return <eslEINVAL> if nali > 1.
* (We don't know we're in a multiple MSA database until we're on the second
* alignment.)
*
* If we're in MPI mode, we assume we're in a multiple MSA database,
* even on the first alignment.
*
* Because we can't tell whether we've got more than one
* alignment 'til we're on the second one, these fatal errors
* only happen after the first HMM has already been built.
* Oh well.
*/
static int
set_msa_name(struct cfg_s *cfg, char *errbuf, ESL_MSA *msa)
{
char *name = NULL;
int status;
if (cfg->do_mpi == FALSE && cfg->nali == 1) /* first (only?) HMM in file: */
{
if (cfg->hmmName != NULL)
{
if ((status = esl_msa_SetName(msa, cfg->hmmName)) != eslOK) return status;
}
else if (msa->name != NULL)
{
cfg->nnamed++;
}
else if (! cfg->afp->do_stdin)
{
if ((status = esl_FileTail(cfg->afp->fname, TRUE, &name)) != eslOK) return status; /* TRUE=nosuffix */
if ((status = esl_msa_SetName(msa, name)) != eslOK) return status;
free(name);
}
else ESL_FAIL(eslEINVAL, errbuf, "Failed to set model name: msa has no name, no msa filename, and no -n");
}
else
{
if (cfg->hmmName != NULL) ESL_FAIL(eslEINVAL, errbuf, "Oops. Wait. You can't use -n with an alignment database.");
else if (msa->name != NULL) cfg->nnamed++;
else ESL_FAIL(eslEINVAL, errbuf, "Oops. Wait. I need name annotation on each alignment in a multi MSA file; failed on #%d", cfg->nali+1);
/* special kind of failure: the *first* alignment didn't have a name, and we used the filename to
* construct one; now that we see a second alignment, we realize this was a boo-boo*/
if (cfg->nnamed != cfg->nali) ESL_FAIL(eslEINVAL, errbuf, "Oops. Wait. I need name annotation on each alignment in a multi MSA file; first MSA didn't have one");
}
return eslOK;
}
P7_HMM * constructHMM(ESL_MSA *msa, ESL_ALPHABET *abc, int ali_hmm, int frag, P7_HMM **ret_hmm, char *errbuf){
int status;
ESL_GETOPTS *go = esl_getopts_Create(options);
P7_BUILDER *bld = NULL;
P7_BG *bg = NULL;
char *args = NULL;
esl_strcat(&args, -1, "X ", -1);
status = esl_msa_SetName(msa, "Query", -1);
/* Now take this alignment and make an HMM from it */
if(status != eslOK){
ESL_XFAIL(status, errbuf, "Easel MSA SetNAME returned an error %d\n", status);
}
bg = p7_bg_Create(abc);
if(bg == NULL){
ESL_XFAIL(status, errbuf, "Error generating bg\n");
}
if (frag == 1) {
// add --fragthresh 0
esl_strcat(&args, -1, "--fragthresh 0 ", -1);
}
// if these flags are set, then we want a non standard hmm out. Used by the logo server.
if (ali_hmm == 1) {
// observed counts
// arguments "X --pnone --wnone --enone"
esl_strcat(&args, -1, "--pnone --wnone --enone --symfrac 0 ", -1);
}
else if (ali_hmm == 2) {
// weighted counts
// arguments "X --pnone");
esl_strcat(&args, -1, "--pnone --symfrac 0 ", -1);
}
else if (ali_hmm == 3) {
// Create HMM - keep all columns
esl_strcat(&args, -1, "--symfrac 0 ", -1);
}
else {
// no arguments ?
}
// pass in arguments to hmm builder
esl_opt_ProcessSpoof(go, args);
if (args != NULL) free(args);
bld = p7_builder_Create(go, abc);
if(bld == NULL){
ESL_XFAIL(eslEMEM, errbuf, "Error creating builder\n");
}
status = p7_Builder(bld, msa, bg, ret_hmm, NULL, NULL, NULL, NULL);
if (status != eslOK) {
strcpy( errbuf, bld->errbuf );
goto ERROR;
}
p7_bg_Destroy(bg);
p7_builder_Destroy(bld);
esl_getopts_Destroy(go);
return status;
ERROR:
if (bg != NULL) p7_bg_Destroy(bg);
if (bld != NULL) p7_builder_Destroy(bld);
if (go != NULL) esl_getopts_Destroy(go);
return status;
}
