/* Function: matassign2hmm()
*
* Purpose: Given an assignment of alignment columns to match vs.
* insert, finish the final part of the model construction
* calculation that is constant between model construction
* algorithms.
*
* Args: msa - multiple sequence alignment
* matassign - 1..alen bit flags for column assignments
* ret_hmm - RETURN: counts-form HMM
* opt_tr - optRETURN: array of tracebacks for aseq's
*
* Return: <eslOK> on success.
* <eslENORESULT> if no consensus columns are identified.
*
* ret_hmm and opt_tr alloc'ed here.
*/
static int
matassign2hmm(ESL_MSA *msa, int *matassign, P7_HMM **ret_hmm, P7_TRACE ***opt_tr)
{
int status; /* return status */
P7_HMM *hmm = NULL; /* RETURN: new hmm */
P7_TRACE **tr = NULL; /* RETURN: 0..nseq-1 fake traces */
int M; /* length of new model in match states */
int idx; /* counter over sequences */
int apos; /* counter for aligned columns */
#ifdef p7_DEBUGGING
char errbuf[eslERRBUFSIZE];
#endif
/* apply the model mask in the 'GC MM' row */
do_modelmask(msa);
/* How many match states in the HMM? */
for (M = 0, apos = 1; apos <= msa->alen; apos++)
if (matassign[apos]) M++;
if (M == 0) { status = eslENORESULT; goto ERROR; }
/* Make fake tracebacks for each seq */
ESL_ALLOC(tr, sizeof(P7_TRACE *) * msa->nseq);
if ((status = p7_trace_FauxFromMSA(msa, matassign, p7_MSA_COORDS, tr)) != eslOK) goto ERROR;
for (idx = 0; idx < msa->nseq; idx++)
{
if ((status = p7_trace_Doctor(tr[idx], NULL, NULL)) != eslOK) goto ERROR;
#ifdef p7_DEBUGGING
if ((status = p7_trace_Validate(tr[idx], msa->abc, msa->ax[idx], errbuf)) != eslOK)
ESL_XEXCEPTION(eslFAIL, "validation failed: %s", errbuf);
#endif
}
/* Build count model from tracebacks */
if ((hmm = p7_hmm_Create(M, msa->abc)) == NULL) { status = eslEMEM; goto ERROR; }
if ((status = p7_hmm_Zero(hmm)) != eslOK) goto ERROR;
for (idx = 0; idx < msa->nseq; idx++) {
if (tr[idx] == NULL) continue; /* skip rare examples of empty sequences */
if ((status = p7_trace_Count(hmm, msa->ax[idx], msa->wgt[idx], tr[idx])) != eslOK) goto ERROR;
}
hmm->nseq = msa->nseq;
hmm->eff_nseq = msa->nseq;
/* Transfer annotation from the MSA to the new model
*/
if ((status = annotate_model(hmm, matassign, msa)) != eslOK) goto ERROR;
/* Reset #=RF line of alignment to reflect our assignment
* of match, delete. matassign is valid from 1..alen and is off
* by one from msa->rf.
*/
if (msa->rf == NULL) ESL_ALLOC(msa->rf, sizeof(char) * (msa->alen + 1));
for (apos = 1; apos <= msa->alen; apos++)
msa->rf[apos-1] = matassign[apos] ? 'x' : '.';
msa->rf[msa->alen] = '\0';
if (opt_tr != NULL) *opt_tr = tr;
else p7_trace_DestroyArray(tr, msa->nseq);
*ret_hmm = hmm;
return eslOK;
ERROR:
if (tr != NULL) p7_trace_DestroyArray(tr, msa->nseq);
if (hmm != NULL) p7_hmm_Destroy(hmm);
if (opt_tr != NULL) *opt_tr = NULL;
*ret_hmm = NULL;
return status;
}
static int
map_alignment(const char *msafile, const P7_HMM *hmm, ESL_SQ ***ret_sq, P7_TRACE ***ret_tr, int *ret_ntot)
{
ESL_SQ **sq = NULL;
P7_TRACE **tr = NULL;
ESLX_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
ESL_ALPHABET *abc = (ESL_ALPHABET *) hmm->abc; /* removing const'ness to make compiler happy. Safe. */
int *matassign = NULL;
uint32_t chksum = 0;
int i,k;
int status;
status = eslx_msafile_Open(&abc, msafile, NULL, eslMSAFILE_UNKNOWN, NULL, &afp);
if (status != eslOK) eslx_msafile_OpenFailure(afp, status);
status = eslx_msafile_Read(afp, &msa);
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
if (! (hmm->flags & p7H_CHKSUM) ) esl_fatal("HMM has no checksum. --mapali unreliable without it.");
if (! (hmm->flags & p7H_MAP) ) esl_fatal("HMM has no map. --mapali can't work without it.");
esl_msa_Checksum(msa, &chksum);
if (hmm->checksum != chksum) esl_fatal("--mapali MSA %s isn't same as the one HMM came from (checksum mismatch)", msafile);
ESL_ALLOC(sq, sizeof(ESL_SQ *) * msa->nseq);
ESL_ALLOC(tr, sizeof(P7_TRACE *) * msa->nseq);
ESL_ALLOC(matassign, sizeof(int) * (msa->alen + 1));
esl_vec_ISet(matassign, msa->alen+1, 0);
for (k = 1; k <= hmm->M; k++) matassign[hmm->map[k]] = 1;
p7_trace_FauxFromMSA(msa, matassign, p7_DEFAULT, tr);
/* The 'faux' core traces constructed by FauxFromMSA() may contain
* D->I and I->D transitions. They may *only* now be passed to
* p7_tracealign_Seqs(), which can deal with these 'illegal'
* transitions, in order to exactly reproduce the input --mapali
* alignment.
*/
for (i = 0; i < msa->nseq; i++)
esl_sq_FetchFromMSA(msa, i, &(sq[i]));
*ret_ntot = msa->nseq;
*ret_tr = tr;
*ret_sq = sq;
eslx_msafile_Close(afp);
esl_msa_Destroy(msa);
free(matassign);
return eslOK;
ERROR:
*ret_ntot = 0;
*ret_tr = NULL;
*ret_sq = NULL;
if (afp != NULL) eslx_msafile_Close(afp);
if (msa != NULL) esl_msa_Destroy(msa);
if (matassign != NULL) free(matassign);
return status;
}
