/* Function: p7_bg_Create()
* Incept: SRE, Fri Jan 12 13:32:51 2007 [Janelia]
*
* Purpose: Allocate a <P7_BG> object for digital alphabet <abc>,
* initializes it to appropriate default values, and
* returns a pointer to it.
*
* For protein models, default iid background frequencies
* are set (by <p7_AminoFrequencies()>) to average
* SwissProt residue composition. For DNA, RNA and other
* alphabets, default frequencies are set to a uniform
* distribution.
*
* The model composition <bg->mcomp[]> is not initialized
* here; neither is the filter null model <bg->fhmm>. To
* use the filter null model, caller will want to
* initialize these fields by calling
* <p7_bg_SetFilterByHMM()>.
*
* Throws: <NULL> on allocation failure.
*
* Xref: STL11/125.
*/
P7_BG *
p7_bg_Create(const ESL_ALPHABET *abc)
{
P7_BG *bg = NULL;
int status;
ESL_ALLOC(bg, sizeof(P7_BG));
bg->f = NULL;
bg->fhmm = NULL;
ESL_ALLOC(bg->f, sizeof(float) * abc->K);
if ((bg->fhmm = esl_hmm_Create(abc, 2)) == NULL) goto ERROR;
if (abc->type == eslAMINO)
{
if (p7_AminoFrequencies(bg->f) != eslOK) goto ERROR;
}
else
esl_vec_FSet(bg->f, abc->K, 1. / (float) abc->K);
bg->p1 = 350./351.;
bg->omega = 1./256.;
bg->abc = abc;
return bg;
ERROR:
p7_bg_Destroy(bg);
return NULL;
}
static int
make_occasionally_dishonest_casino(ESL_HMM **ret_hmm, ESL_ALPHABET **ret_abc)
{
ESL_ALPHABET *abc = esl_alphabet_Create(eslDICE);
ESL_HMM *hmm = esl_hmm_Create(abc, 2);
int x;
/* State 0 = fair die */
hmm->pi[0] = 1.0;
hmm->pi[1] = 0.0;
hmm->pi[2] = 0.0; /* no L=0 seqs */
hmm->t[0][0] = 0.96;
hmm->t[0][1] = 0.03;
hmm->t[0][2] = 0.01; /* end from state 0; mean length 100 */
for (x = 0; x < abc->K; x++)
hmm->e[0][x] = 1.0 / (float) abc->K;
/* State 1 = loaded die */
hmm->t[1][0] = 0.05;
hmm->t[1][1] = 0.95;
hmm->t[1][2] = 0.0; /* no end from state 1 */
for (x = 0; x < abc->K-1; x++) hmm->e[1][x] = 0.5 / ((float) abc->K-1);
hmm->e[1][abc->K-1] = 0.5;
esl_hmm_Configure(hmm, NULL);
*ret_hmm = hmm;
*ret_abc = abc;
return eslOK;
}
/* Function: esl_hmm_Clone()
* Synopsis: Duplicate an HMM.
*
* Purpose: Make a newly allocated duplicate of the HMM <hmm>,
* and return a pointer to the duplicate.
*
* Throws: <NULL> on allocation failure.
*/
ESL_HMM *
esl_hmm_Clone(const ESL_HMM *hmm)
{
ESL_HMM *dup = NULL;
int k, x;
if ((dup = esl_hmm_Create(hmm->abc, hmm->M)) == NULL) return NULL;
for (k = 0; k < hmm->M; k++)
{
memcpy(dup->t[k], hmm->t[k], sizeof(float) * (hmm->M+1));
memcpy(dup->e[k], hmm->e[k], sizeof(float) * (hmm->abc->K));
}
for (x = 0; x < hmm->abc->Kp; x++)
{
memcpy(dup->eo[x], hmm->eo[x], sizeof(float) * (hmm->M));
}
memcpy(dup->pi, hmm->pi, sizeof(float) * (hmm->M+1));
return dup;
}
/* Function: p7_bg_CreateUniform()
* Synopsis: Creates background model with uniform freqs.
* Incept: SRE, Sat Jun 30 10:25:27 2007 [Janelia]
*
* Purpose: Creates a background model for alphabet <abc>
* with uniform residue frequencies.
*/
P7_BG *
p7_bg_CreateUniform(const ESL_ALPHABET *abc)
{
P7_BG *bg = NULL;
int status;
ESL_ALLOC(bg, sizeof(P7_BG));
bg->f = NULL;
bg->fhmm = NULL;
ESL_ALLOC(bg->f, sizeof(float) * abc->K);
if ((bg->fhmm = esl_hmm_Create(abc, 2)) == NULL) goto ERROR;
esl_vec_FSet(bg->f, abc->K, 1. / (float) abc->K);
bg->p1 = 350./351.;
bg->omega = 1./256.;
bg->abc = (ESL_ALPHABET *) abc; /* safe: we're just keeping a reference */
return bg;
ERROR:
p7_bg_Destroy(bg);
return NULL;
}
/* Function: p7_bg_Create()
* Synopsis: Create a <P7_BG> null model object.
*
* Purpose: Allocate a <P7_BG> object for digital alphabet <abc>,
* initializes it to appropriate default values, and
* returns a pointer to it.
*
* For protein models, default iid background frequencies
* are set (by <p7_AminoFrequencies()>) to average
* Swiss-Prot residue composition. For DNA, RNA and other
* alphabets, default frequencies are set to a uniform
* distribution.
*
* The model composition <bg->mcomp[]> is not initialized
* here; neither is the filter null model <bg->fhmm>. To
* use the filter null model, caller will want to
* initialize these fields by calling
* <p7_bg_SetFilter()>.
*
* Throws: <NULL> on allocation failure.
*
* Xref: STL11/125.
*/
P7_BG *
p7_bg_Create(const ESL_ALPHABET *abc)
{
P7_BG *bg = NULL;
int status;
ESL_ALLOC(bg, sizeof(P7_BG));
bg->f = NULL;
bg->fhmm = NULL;
// this is not hard-coded for alphabet size...
ESL_ALLOC(bg->f, sizeof(float) * abc->K);
if ((bg->fhmm = esl_hmm_Create(abc, 2)) == NULL) goto ERROR;
if (abc->type == eslAMINO)
{
if (p7_AminoFrequencies(bg->f) != eslOK) goto ERROR;
}
// adding in background probabilities for music intervals
else if (abc->type == eslMUSIC)
{
if (p7_MusicFrequencies(bg->f) != eslOK) goto ERROR;
}
else
esl_vec_FSet(bg->f, abc->K, 1. / (float) abc->K);
bg->p1 = 350./351.;
bg->omega = 1./256.;
bg->abc = abc;
return bg;
ERROR:
p7_bg_Destroy(bg);
return NULL;
}
ESL_HMM *
create_null_hmm(ESL_ALPHABET *abc)
{
ESL_HMM *hmm;
hmm = esl_hmm_Create(abc, 1);
/* state 0 = normal iid model.*/
hmm->t[0][0] = 1.0f;
hmm->t[0][1] = 1.0f; /* external length distribution */
/* SW50 iid frequencies: H3 default background */
hmm->e[0][0] = 0.0787945; /* A */
hmm->e[0][1] = 0.0151600; /* C */
hmm->e[0][2] = 0.0535222; /* D */
hmm->e[0][3] = 0.0668298; /* E */
hmm->e[0][4] = 0.0397062; /* F */
hmm->e[0][5] = 0.0695071; /* G */
hmm->e[0][6] = 0.0229198; /* H */
hmm->e[0][7] = 0.0590092; /* I */
hmm->e[0][8] = 0.0594422; /* K */
hmm->e[0][9] = 0.0963728; /* L */
hmm->e[0][10] = 0.0237718; /* M */
hmm->e[0][11] = 0.0414386; /* N */
hmm->e[0][12] = 0.0482904; /* P */
hmm->e[0][13] = 0.0395639; /* Q */
hmm->e[0][14] = 0.0540978; /* R */
hmm->e[0][15] = 0.0683364; /* S */
hmm->e[0][16] = 0.0540687; /* T */
hmm->e[0][17] = 0.0673417; /* V */
hmm->e[0][18] = 0.0114135; /* W */
hmm->e[0][19] = 0.0304133; /* Y */
hmm->pi[0] = 1.0;
esl_hmm_Configure(hmm, NULL);
return hmm;
}
ESL_HMM *
create_test_hmm(ESL_ALPHABET *abc)
{
ESL_HMM *hmm;
int L = 400;
int M = 200;
hmm = esl_hmm_Create(abc, 2);
/* state 0 = normal iid model. state 1 = biased state */
hmm->t[0][0] = (float) L / (float) (L+1);
hmm->t[0][1] = 1.0f / (float) (L+1);
hmm->t[0][2] = 1.0; /* external length distribution */
hmm->t[1][0] = (float) 2.0f / (float) (M+2);
hmm->t[1][1] = (float) M / (float) (M+2);
hmm->t[1][2] = 1.0;
/* SW50 iid frequencies: H3 default background */
hmm->e[0][0] = 0.0787945; /* A */
hmm->e[0][1] = 0.0151600; /* C */
hmm->e[0][2] = 0.0535222; /* D */
hmm->e[0][3] = 0.0668298; /* E */
hmm->e[0][4] = 0.0397062; /* F */
hmm->e[0][5] = 0.0695071; /* G */
hmm->e[0][6] = 0.0229198; /* H */
hmm->e[0][7] = 0.0590092; /* I */
hmm->e[0][8] = 0.0594422; /* K */
hmm->e[0][9] = 0.0963728; /* L */
hmm->e[0][10] = 0.0237718; /* M */
hmm->e[0][11] = 0.0414386; /* N */
hmm->e[0][12] = 0.0482904; /* P */
hmm->e[0][13] = 0.0395639; /* Q */
hmm->e[0][14] = 0.0540978; /* R */
hmm->e[0][15] = 0.0683364; /* S */
hmm->e[0][16] = 0.0540687; /* T */
hmm->e[0][17] = 0.0673417; /* V */
hmm->e[0][18] = 0.0114135; /* W */
hmm->e[0][19] = 0.0304133; /* Y */
/* average of MFS_1 core emissions */
hmm->e[1][0] = 0.1068; /* A */
hmm->e[1][1] = 0.0110; /* C */
hmm->e[1][2] = 0.0242; /* D */
hmm->e[1][3] = 0.0293; /* E */
hmm->e[1][4] = 0.0621; /* F */
hmm->e[1][5] = 0.0899; /* G */
hmm->e[1][6] = 0.0139; /* H */
hmm->e[1][7] = 0.0762; /* I */
hmm->e[1][8] = 0.0319; /* K */
hmm->e[1][9] = 0.1274; /* L */
hmm->e[1][10] = 0.0338; /* M */
hmm->e[1][11] = 0.0285; /* N */
hmm->e[1][12] = 0.0414; /* P */
hmm->e[1][13] = 0.0266; /* Q */
hmm->e[1][14] = 0.0375; /* R */
hmm->e[1][15] = 0.0747; /* S */
hmm->e[1][16] = 0.0568; /* T */
hmm->e[1][17] = 0.0815; /* V */
hmm->e[1][18] = 0.0161; /* W */
hmm->e[1][19] = 0.0303; /* Y */
hmm->pi[0] = 0.99;
hmm->pi[1] = 0.01;
esl_hmm_Configure(hmm, NULL);
return hmm;
}
