/* Function: p7_profile_Create()
* Synopsis: Allocates a profile.
*
* Purpose: Allocates for a profile of up to <M> nodes, for digital
* alphabet <abc>.
*
* Because this function might be in the critical path (in
* hmmscan, for example), we leave much of the model
* uninitialized, including scores and length model
* probabilities. The <p7_profile_Config()> call is what
* sets these.
*
* The reference pointer <gm->abc> is set to <abc>.
*
* Returns: a pointer to the newly allocated profile.
*
* Throws: <NULL> on allocation error.
*/
P7_PROFILE *
p7_profile_Create(int allocM, const ESL_ALPHABET *abc)
{
P7_PROFILE *gm = NULL;
int x;
int status;
/* level 0 */
ESL_ALLOC(gm, sizeof(P7_PROFILE));
gm->tsc = NULL;
gm->rsc = NULL;
gm->name = NULL;
gm->acc = NULL;
gm->desc = NULL;
gm->rf = NULL;
gm->mm = NULL;
gm->cs = NULL;
gm->consensus = NULL;
/* level 1 */
ESL_ALLOC(gm->tsc, sizeof(float) * (allocM+1) * p7P_NTRANS); /* 0..M */
ESL_ALLOC(gm->rsc, sizeof(float *) * abc->Kp);
ESL_ALLOC(gm->rf, sizeof(char) * (allocM+2)); /* yes, +2: each is (0)1..M, +trailing \0 */
ESL_ALLOC(gm->mm, sizeof(char) * (allocM+2));
ESL_ALLOC(gm->cs, sizeof(char) * (allocM+2));
ESL_ALLOC(gm->consensus, sizeof(char) * (allocM+2));
gm->rsc[0] = NULL;
/* level 2 */
ESL_ALLOC(gm->rsc[0], sizeof(float) * abc->Kp * (allocM+1) * p7P_NR);
for (x = 1; x < abc->Kp; x++)
gm->rsc[x] = gm->rsc[0] + x * (allocM+1) * p7P_NR;
/* Initialization of tsc[0], including removal of I0. tsc[k-1,LM],tsc[k-1,GM] will be configured + overwritten later */
esl_vec_FSet(gm->tsc, p7P_NTRANS, -eslINFINITY);
/* tsc[M] initialized and Im removed when we know actual M : see modelconfig.c */
for (x = 0; x < abc->Kp; x++) {
P7P_MSC(gm, 0, x) = -eslINFINITY; /* no emissions from nonexistent M_0... */
P7P_ISC(gm, 0, x) = -eslINFINITY; /* nor I_0... */
/* I_M is initialized in profile config, when we know actual M, not just allocated max M */
}
x = esl_abc_XGetGap(abc); /* no emission can emit/score gap characters */
esl_vec_FSet(gm->rsc[x], (allocM+1)*p7P_NR, -eslINFINITY);
x = esl_abc_XGetMissing(abc); /* no emission can emit/score missing data characters */
esl_vec_FSet(gm->rsc[x], (allocM+1)*p7P_NR, -eslINFINITY);
/* Set remaining info */
gm->M = 0;
gm->allocM = allocM;
gm->L = -1; /* "unset" flag */
gm->nj = -1.0f; /* "unset" flag */
gm->pglocal = -1.0f; /* "unset" flag */
gm->roff = -1;
gm->eoff = -1;
gm->offs[p7_MOFFSET] = -1;
gm->offs[p7_FOFFSET] = -1;
gm->offs[p7_POFFSET] = -1;
gm->name = NULL;
gm->acc = NULL;
gm->desc = NULL;
gm->rf[0] = 0; /* RF line is optional annotation; this flags that it's not set yet */
gm->mm[0] = 0; /* likewise for MM annotation line */
gm->cs[0] = 0; /* likewise for CS annotation line */
gm->consensus[0] = 0;
for (x = 0; x < p7_NEVPARAM; x++) gm->evparam[x] = p7_EVPARAM_UNSET;
for (x = 0; x < p7_NCUTOFFS; x++) gm->cutoff[x] = p7_CUTOFF_UNSET;
for (x = 0; x < p7_MAXABET; x++) gm->compo[x] = p7_COMPO_UNSET;
gm->max_length = -1; /* "unset" */
gm->abc = abc;
return gm;
ERROR:
p7_profile_Destroy(gm);
return NULL;
}
static int
make_digital_msa(ESL_SQ **sq, const ESL_MSA *premsa, P7_TRACE **tr, int nseq, const int *matuse, const int *matmap, int M, int alen, int optflags, ESL_MSA **ret_msa)
{
const ESL_ALPHABET *abc = (sq == NULL) ? premsa->abc : sq[0]->abc;
ESL_MSA *msa = NULL;
int idx;
int apos;
int z;
int status;
if ((msa = esl_msa_CreateDigital(abc, nseq, alen)) == NULL) { status = eslEMEM; goto ERROR; }
for (idx = 0; idx < nseq; idx++)
{
msa->ax[idx][0] = eslDSQ_SENTINEL;
for (apos = 1; apos <= alen; apos++) msa->ax[idx][apos] = esl_abc_XGetGap(abc);
msa->ax[idx][alen+1] = eslDSQ_SENTINEL;
apos = 1;
for (z = 0; z < tr[idx]->N; z++)
{
switch (tr[idx]->st[z]) {
case p7T_M:
msa->ax[idx][matmap[tr[idx]->k[z]]] = get_dsq_z(sq, premsa, tr, idx, z);
apos = matmap[tr[idx]->k[z]] + 1;
break;
case p7T_D:
if (matuse[tr[idx]->k[z]]) /* bug h77: if all col is deletes, do nothing; do NOT overwrite a column */
msa->ax[idx][matmap[tr[idx]->k[z]]] = esl_abc_XGetGap(abc); /* overwrites ~ in Dk column on X->Dk */
apos = matmap[tr[idx]->k[z]] + 1;
break;
case p7T_I:
if ( !(optflags & p7_TRIM) || (tr[idx]->k[z] != 0 && tr[idx]->k[z] != M)) {
msa->ax[idx][apos] = get_dsq_z(sq, premsa, tr, idx, z);
apos++;
}
break;
case p7T_N:
case p7T_C:
if (! (optflags & p7_TRIM) && tr[idx]->i[z] > 0) {
msa->ax[idx][apos] = get_dsq_z(sq, premsa, tr, idx, z);
apos++;
}
break;
case p7T_E:
apos = matmap[M]+1; /* set position for C-terminal tail */
break;
case p7T_X:
/* Mark fragments (B->X and X->E containing core traces):
* convert flanks from gaps to ~
*/
if (tr[idx]->st[z-1] == p7T_B)
{ /* B->X leader. This is a core trace and a fragment. Convert leading gaps to ~ */
/* to set apos for an initial Ik: peek at next state for B->X->Ik; superfluous for ->{DM}k: */
for (apos = 1; apos <= matmap[tr[idx]->k[z+1]]; apos++)
msa->ax[idx][apos] = esl_abc_XGetMissing(abc);
/* tricky! apos is now exactly where it needs to be for X->Ik. all other cases except B->X->Ik set their own apos */
}
else if (tr[idx]->st[z+1] == p7T_E)
{ /* X->E trailer. This is a core trace and a fragment. Convert trailing gaps to ~ */
/* don't need to set apos for trailer. There can't be any more residues in a core trace once we hit X->E */
for (; apos <= alen; apos++)
msa->ax[idx][apos] = esl_abc_XGetMissing(abc);
}
else ESL_XEXCEPTION(eslECORRUPT, "make_digital_msa(): X state in unexpected position in trace");
break;
default:
break;
}
}
}
msa->nseq = nseq;
msa->alen = alen;
*ret_msa = msa;
return eslOK;
ERROR:
if (msa) esl_msa_Destroy(msa);
*ret_msa = NULL;
return status;
}
