static void
utest_normalization(ESL_GETOPTS *go)
{
char *msg = "seqmodel normalization utest failed";
ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO);
char *seq = "ACDEFGHIKLMNPQRSTVWYBJZOUX";
int L = strlen(seq);
ESL_DSQ *dsq = NULL;
float popen = 0.1;
float pextend = 0.4;
P7_BUILDER *bld = NULL;
P7_BG *bg = p7_bg_Create(abc);
P7_HMM *hmm = NULL;
char errbuf[eslERRBUFSIZE];
if ( esl_abc_CreateDsq(abc, seq, &dsq) != eslOK) esl_fatal(msg);
if ( (bld = p7_builder_Create(NULL, abc)) == NULL) esl_fatal(msg);
if ( p7_builder_LoadScoreSystem(bld, "BLOSUM62", popen, pextend, bg) != eslOK) esl_fatal(msg);
if ( p7_Seqmodel(abc, dsq, L, "aatest", bld->Q, bg->f, bld->popen, bld->pextend, &hmm) != eslOK) esl_fatal(msg);
if (p7_hmm_Validate(hmm, errbuf, 0.0001) != eslOK) esl_fatal("normalization utest failed\n%s\n", errbuf);
free(dsq);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
p7_builder_Destroy(bld);
esl_alphabet_Destroy(abc);
}
/* 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;
}
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 *hmmfile = esl_opt_GetArg(go, 1);
char *qfile = esl_opt_GetArg(go, 2);
ESL_SQ *qsq = esl_sq_CreateDigital(abc);
ESL_SQFILE *qfp = NULL;
FILE *hmmfp = NULL;
ESL_SCOREMATRIX *S = esl_scorematrix_Create(abc);
ESL_DMATRIX *Q = NULL;
P7_BG *bg = p7_bg_Create(abc);
P7_HMM *hmm = NULL;
double *fa = NULL;
double popen = esl_opt_GetReal (go, "-q");
double pextend = esl_opt_GetReal (go, "-r");
char *mxfile = esl_opt_GetString(go, "-m");
char errbuf[eslERRBUFSIZE];
double slambda;
int a,b;
int status;
/* Reverse engineer a scoring matrix to obtain conditional prob's
* that we'll use for the single-seq query HMM. Because score mx is
* symmetric, we can set up P[a][b] = P(b | a), so we can use the
* matrix rows as HMM match emission vectors. This means dividing
* the joint probs through by f_a.
*/
if (mxfile == NULL) {
if (esl_scorematrix_Set("BLOSUM62", S) != eslOK) esl_fatal("failed to set BLOSUM62 scores");
} else {
ESL_FILEPARSER *efp = NULL;
if ( esl_fileparser_Open(mxfile, NULL, &efp) != eslOK) esl_fatal("failed to open score file %s", mxfile);
if ( esl_scorematrix_Read(efp, abc, &S) != eslOK) esl_fatal("failed to read matrix from %s", mxfile);
esl_fileparser_Close(efp);
}
/* A wasteful conversion of the HMMER single-precision background probs to Easel double-prec */
ESL_ALLOC(fa, sizeof(double) * bg->abc->K);
esl_vec_F2D(bg->f, bg->abc->K, fa);
/* Backcalculate joint probabilities Q, given score matrix S and background frequencies fa */
status = esl_scorematrix_ProbifyGivenBG(S, fa, fa, &slambda, &Q);
if (status == eslEINVAL) esl_fatal("built-in score matrix %s has no valid solution for lambda", matrix);
else if (status == eslENOHALT) esl_fatal("failed to solve score matrix %s for lambda", matrix);
else if (status != eslOK) esl_fatal("unexpected error in solving score matrix %s for probability parameters", matrix);
esl_scorematrix_JointToConditionalOnQuery(abc, Q);
/* Open the query sequence file in FASTA format */
status = esl_sqfile_Open(qfile, eslSQFILE_FASTA, NULL, &qfp);
if (status == eslENOTFOUND) esl_fatal("No such file %s.", qfile);
else if (status == eslEFORMAT) esl_fatal("Format of %s unrecognized.", qfile);
else if (status == eslEINVAL) esl_fatal("Can't autodetect stdin or .gz.");
else if (status != eslOK) esl_fatal("Open of %s failed, code %d.", qfile, status);
/* Open the output HMM file */
if ((hmmfp = fopen(hmmfile, "w")) == NULL) esl_fatal("Failed to open output HMM file %s", hmmfile);
/* For each sequence, build a model and save it.
*/
while ((status = esl_sqio_Read(qfp, qsq)) == eslOK)
{
p7_Seqmodel(abc, qsq->dsq, qsq->n, qsq->name, Q, bg->f, popen, pextend, &hmm);
if ( p7_hmm_Validate(hmm, errbuf, 1e-5) != eslOK) esl_fatal("HMM validation failed: %s\n", errbuf);
if ( p7_hmmfile_WriteASCII(hmmfp, -1, hmm) != eslOK) esl_fatal("HMM save failed");
p7_hmm_Destroy(hmm);
}
if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s line %" PRId64 "):\n%s\n",
qfp->filename, qfp->linenumber, qfp->errbuf);
else if (status != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s",
status, qfp->filename);
esl_dmatrix_Destroy(Q);
esl_scorematrix_Destroy(S);
free(fa);
free(fb);
esl_sq_Destroy(qsq);
esl_sqfile_Close(qfp);
fclose(hmmfp);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
