/* annotate()
* Transfer annotation information from MSA to new HMM.
* Also sets model-specific residue composition (hmm->compo).
*/
static int
annotate(P7_BUILDER *bld, const ESL_MSA *msa, P7_HMM *hmm)
{
int status;
/* Name. */
if (msa->name) p7_hmm_SetName(hmm, msa->name);
else ESL_XFAIL(eslEINVAL, bld->errbuf, "Unable to name the HMM.");
if ((status = p7_hmm_SetAccession (hmm, msa->acc)) != eslOK) ESL_XFAIL(status, bld->errbuf, "Failed to record MSA accession");
if ((status = p7_hmm_SetDescription(hmm, msa->desc)) != eslOK) ESL_XFAIL(status, bld->errbuf, "Failed to record MSA description");
// if ((status = p7_hmm_AppendComlog(hmm, go->argc, go->argv)) != eslOK) ESL_XFAIL(status, errbuf, "Failed to record command log");
if ((status = p7_hmm_SetCtime(hmm)) != eslOK) ESL_XFAIL(status, bld->errbuf, "Failed to record timestamp");
if ((status = p7_hmm_SetComposition(hmm)) != eslOK) ESL_XFAIL(status, bld->errbuf, "Failed to determine model composition");
if ((status = p7_hmm_SetConsensus(hmm, NULL)) != eslOK) ESL_XFAIL(status, bld->errbuf, "Failed to set consensus line");
if (msa->cutset[eslMSA_GA1] && msa->cutset[eslMSA_GA2]) { hmm->cutoff[p7_GA1] = msa->cutoff[eslMSA_GA1]; hmm->cutoff[p7_GA2] = msa->cutoff[eslMSA_GA2]; hmm->flags |= p7H_GA; }
if (msa->cutset[eslMSA_TC1] && msa->cutset[eslMSA_TC2]) { hmm->cutoff[p7_TC1] = msa->cutoff[eslMSA_TC1]; hmm->cutoff[p7_TC2] = msa->cutoff[eslMSA_TC2]; hmm->flags |= p7H_TC; }
if (msa->cutset[eslMSA_NC1] && msa->cutset[eslMSA_NC2]) { hmm->cutoff[p7_NC1] = msa->cutoff[eslMSA_NC1]; hmm->cutoff[p7_NC2] = msa->cutoff[eslMSA_NC2]; hmm->flags |= p7H_NC; }
return eslOK;
ERROR:
return status;
}
/* The "basic" utest is a minimal driver for making a small DNA profile and a small DNA sequence,
* then running Viterbi and Forward. It's useful for dumping DP matrices and profiles for debugging.
*/
static void
utest_basic(ESL_GETOPTS *go)
{
char *query= "# STOCKHOLM 1.0\n\nseq1 GAATTC\nseq2 GAATTC\n//\n";
int fmt = eslMSAFILE_STOCKHOLM;
char *targ = "GAATTC";
ESL_ALPHABET *abc = NULL;
ESL_MSA *msa = NULL;
P7_HMM *hmm = NULL;
P7_PROFILE *gm = NULL;
P7_BG *bg = NULL;
P7_PRIOR *pri = NULL;
ESL_DSQ *dsq = NULL;
P7_GMX *gx = NULL;
P7_TRACE *tr = NULL;
int L = strlen(targ);
float vsc, vsc2, fsc;
if ((abc = esl_alphabet_Create(eslDNA)) == NULL) esl_fatal("failed to create alphabet");
if ((pri = p7_prior_CreateNucleic()) == NULL) esl_fatal("failed to create prior");
if ((msa = esl_msa_CreateFromString(query, fmt)) == NULL) esl_fatal("failed to create MSA");
if (esl_msa_Digitize(abc, msa, NULL) != eslOK) esl_fatal("failed to digitize MSA");
if (p7_Fastmodelmaker(msa, 0.5, NULL, &hmm, NULL) != eslOK) esl_fatal("failed to create GAATTC model");
if (p7_ParameterEstimation(hmm, pri) != eslOK) esl_fatal("failed to parameterize GAATTC model");
if (p7_hmm_SetConsensus(hmm, NULL) != eslOK) esl_fatal("failed to make consensus");
if ((bg = p7_bg_Create(abc)) == NULL) esl_fatal("failed to create DNA null model");
if ((gm = p7_profile_Create(hmm->M, abc)) == NULL) esl_fatal("failed to create GAATTC profile");
if (p7_ProfileConfig(hmm, bg, gm, L, p7_UNILOCAL)!= eslOK) esl_fatal("failed to config profile");
if (p7_profile_Validate(gm, NULL, 0.0001) != eslOK) esl_fatal("whoops, profile is bad!");
if (esl_abc_CreateDsq(abc, targ, &dsq) != eslOK) esl_fatal("failed to create GAATTC digital sequence");
if ((gx = p7_gmx_Create(gm->M, L)) == NULL) esl_fatal("failed to create DP matrix");
if ((tr = p7_trace_Create()) == NULL) esl_fatal("trace creation failed");
p7_GViterbi (dsq, L, gm, gx, &vsc);
if (esl_opt_GetBoolean(go, "-v")) printf("Viterbi score: %.4f\n", vsc);
if (esl_opt_GetBoolean(go, "-v")) p7_gmx_Dump(stdout, gx, p7_DEFAULT);
p7_GTrace (dsq, L, gm, gx, tr);
p7_trace_Score(tr, dsq, gm, &vsc2);
if (esl_opt_GetBoolean(go, "-v")) p7_trace_Dump(stdout, tr, gm, dsq);
if (esl_FCompare(vsc, vsc2, 1e-5) != eslOK) esl_fatal("trace score and Viterbi score don't agree.");
p7_GForward (dsq, L, gm, gx, &fsc);
if (esl_opt_GetBoolean(go, "-v")) printf("Forward score: %.4f\n", fsc);
if (esl_opt_GetBoolean(go, "-v")) p7_gmx_Dump(stdout, gx, p7_DEFAULT);
p7_trace_Destroy(tr);
p7_gmx_Destroy(gx);
free(dsq);
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
esl_msa_Destroy(msa);
p7_prior_Destroy(pri);
esl_alphabet_Destroy(abc);
return;
}
/* Function: p7_SingleBuilder()
* Synopsis: Build a new HMM from a single sequence.
*
* Purpose: Take the sequence <sq> and a build configuration <bld>, and
* build a new HMM.
*
* The single sequence scoring system in the <bld>
* configuration must have been previously initialized by
* <p7_builder_SetScoreSystem()>.
*
* Args: bld - build configuration
* sq - query sequence
* bg - null model (needed to paramaterize insert emission probs)
* opt_hmm - optRETURN: new HMM
* opt_gm - optRETURN: profile corresponding to <hmm>
* opt_om - optRETURN: optimized profile corresponding to <gm>
*
* Returns: <eslOK> on success.
*
* Throws: <eslEMEM> on allocation error.
* <eslEINVAL> if <bld> isn't properly configured somehow.
*/
int
p7_SingleBuilder(P7_BUILDER *bld, ESL_SQ *sq, P7_BG *bg, P7_HMM **opt_hmm,
P7_TRACE **opt_tr, P7_PROFILE **opt_gm, P7_OPROFILE **opt_om)
{
P7_HMM *hmm = NULL;
P7_TRACE *tr = NULL;
int k;
int status;
bld->errbuf[0] = '\0';
if (! bld->Q) ESL_XEXCEPTION(eslEINVAL, "score system not initialized");
if ((status = p7_Seqmodel(bld->abc, sq->dsq, sq->n, sq->name, bld->Q, bg->f, bld->popen, bld->pextend, &hmm)) != eslOK) goto ERROR;
if ((status = p7_hmm_SetComposition(hmm)) != eslOK) goto ERROR;
if ((status = p7_hmm_SetConsensus(hmm, sq)) != eslOK) goto ERROR;
if ((status = calibrate(bld, hmm, bg, opt_gm, opt_om)) != eslOK) goto ERROR;
if ( bld->abc->type == eslDNA || bld->abc->type == eslRNA ) {
if (bld->w_len > 0) hmm->max_length = bld->w_len;
else if (bld->w_beta == 0.0) hmm->max_length = hmm->M *4;
else if ( (status = p7_Builder_MaxLength(hmm, bld->w_beta)) != eslOK) goto ERROR;
}
/* build a faux trace: relative to core model (B->M_1..M_L->E) */
if (opt_tr != NULL)
{
if ((tr = p7_trace_Create()) == NULL) goto ERROR;
if ((status = p7_trace_Append(tr, p7T_B, 0, 0)) != eslOK) goto ERROR;
for (k = 1; k <= sq->n; k++)
if ((status = p7_trace_Append(tr, p7T_M, k, k)) != eslOK) goto ERROR;
if ((status = p7_trace_Append(tr, p7T_E, 0, 0)) != eslOK) goto ERROR;
tr->M = sq->n;
tr->L = sq->n;
}
/* note that <opt_gm> and <opt_om> were already set by calibrate() call above. */
if (opt_hmm != NULL) *opt_hmm = hmm; else p7_hmm_Destroy(hmm);
if (opt_tr != NULL) *opt_tr = tr;
return eslOK;
ERROR:
p7_hmm_Destroy(hmm);
if (tr != NULL) p7_trace_Destroy(tr);
if (opt_gm != NULL) p7_profile_Destroy(*opt_gm);
if (opt_om != NULL) p7_oprofile_Destroy(*opt_om);
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = p7_CreateDefaultApp(options, 2, argc, argv, banner, usage);
ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO);
char *qfile = esl_opt_GetArg(go, 1);
char *tfile = esl_opt_GetArg(go, 2);
ESL_SQFILE *qfp = NULL;
ESL_SQFILE *tfp = NULL;
ESL_SQ *qsq = esl_sq_CreateDigital(abc);
ESL_SQ *tsq = esl_sq_CreateDigital(abc);
ESL_SCOREMATRIX *S = esl_scorematrix_Create(abc);
ESL_DMATRIX *Q = NULL;
P7_BG *bg = p7_bg_Create(abc);
P7_HMM *hmm = NULL;
P7_PROFILE *gm = NULL;
P7_REFMX *vit = p7_refmx_Create(200, 400); /* will grow as needed */
double *fa = malloc(sizeof(double) * abc->K);
double popen = 0.02;
double pextend = 0.4;
double lambda;
float vsc;
float nullsc;
int status;
esl_composition_BL62(fa);
esl_vec_D2F(fa, abc->K, bg->f);
esl_scorematrix_Set("BLOSUM62", S);
esl_scorematrix_ProbifyGivenBG(S, fa, fa, &lambda, &Q);
esl_scorematrix_JointToConditionalOnQuery(abc, Q);
if (esl_sqfile_OpenDigital(abc, qfile, eslSQFILE_UNKNOWN, NULL, &qfp) != eslOK) esl_fatal("failed to open %s", qfile);
if (esl_sqio_Read(qfp, qsq) != eslOK) esl_fatal("failed to read query seq");
p7_Seqmodel(abc, qsq->dsq, qsq->n, qsq->name, Q, bg->f, popen, pextend, &hmm);
p7_hmm_SetComposition(hmm);
p7_hmm_SetConsensus(hmm, qsq);
gm = p7_profile_Create(hmm->M, abc);
p7_profile_ConfigUnilocal(gm, hmm, bg, 400);
if (esl_sqfile_OpenDigital(abc, tfile, eslSQFILE_UNKNOWN, NULL, &tfp) != eslOK) esl_fatal("failed to open %s", tfile);
while ((status = esl_sqio_Read(tfp, tsq)) == eslOK)
{
p7_bg_SetLength (bg, tsq->n);
p7_profile_SetLength(gm, tsq->n);
p7_ReferenceViterbi(tsq->dsq, tsq->n, gm, vit, NULL, &vsc);
p7_bg_NullOne(bg, tsq->dsq, tsq->n, &nullsc);
printf("%.4f %-25s %-25s\n", (vsc - nullsc) / eslCONST_LOG2, tsq->name, gm->name);
esl_sq_Reuse(tsq);
p7_refmx_Reuse(vit);
}
p7_refmx_Destroy(vit);
p7_profile_Destroy(gm);
p7_hmm_Destroy(hmm);
p7_bg_Destroy(bg);
esl_dmatrix_Destroy(Q);
esl_scorematrix_Destroy(S);
free(fa);
esl_sq_Destroy(qsq);
esl_sq_Destroy(tsq);
esl_sqfile_Close(qfp);
esl_sqfile_Close(tfp);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
