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);
}
/* 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;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_RANDOMNESS *r = NULL;
char **as = NULL; /* aligned character seqs (random, iid) */
int N,L; /* # of seqs, and their aligned lengths */
int seed;
int i,j;
int status;
double p[4]; /* ACGT probabilities */
#ifdef eslAUGMENT_ALPHABET
ESL_DSQ **ax = NULL; /* digitized alignment */
ESL_ALPHABET *abc = NULL;
#endif
/* Process command line
*/
go = esl_getopts_Create(options);
esl_opt_ProcessCmdline(go, argc, argv);
esl_opt_VerifyConfig(go);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
L = esl_opt_GetInteger(go, "-L");
N = esl_opt_GetInteger(go, "-N");
seed = esl_opt_GetInteger(go, "--seed");
if (esl_opt_ArgNumber(go) != 0) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
esl_getopts_Destroy(go);
/* Create a random DNA alignment;
* force it to obey the conventions of the unit tests:
* 0,1 are identical
* 0,2 are completely dissimilar
*/
r = esl_randomness_Create(seed);
for (i = 0; i < 4; i++) p[i] = 0.25;
ESL_ALLOC(as, sizeof(char *) * N);
for (i = 0; i < N; i++)
ESL_ALLOC(as[i], sizeof(char) * (L+1));
esl_rsq_IID(r, "ACGT", p, 4, L, as[0]);
strcpy(as[1], as[0]);
esl_rsq_IID(r, "ACGT", p, 4, L, as[2]);
for (j = 0; j < L; j++)
while (as[2][j] == as[0][j])
as[2][j] = "ACGT"[esl_rnd_Roll(r, 4)];
for (i = 3; i < N; i++)
esl_rsq_IID(r, "ACGT", p, 4, L, as[i]);
#ifdef eslAUGMENT_ALPHABET
abc = esl_alphabet_Create(eslDNA);
ESL_ALLOC(ax, sizeof(ESL_DSQ *) * N);
for (i = 0; i < N; i++)
esl_abc_CreateDsq(abc, as[i], &(ax[i]));
#endif /*eslAUGMENT_ALPHABET*/
/* Unit tests
*/
if (utest_CPairId(as, N) != eslOK) return eslFAIL;
if (utest_CJukesCantor(4, as, N) != eslOK) return eslFAIL;
#ifdef eslAUGMENT_ALPHABET
if (utest_XPairId(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XJukesCantor(abc, as, ax, N) != eslOK) return eslFAIL;
#endif /*eslAUGMENT_ALPHABET*/
#ifdef eslAUGMENT_DMATRIX
if (utest_CPairIdMx(as, N) != eslOK) return eslFAIL;
if (utest_CDiffMx(as, N) != eslOK) return eslFAIL;
if (utest_CJukesCantorMx(4, as, N) != eslOK) return eslFAIL;
#endif /* eslAUGMENT_DMATRIX*/
#if defined (eslAUGMENT_ALPHABET) && defined (eslAUGMENT_DMATRIX)
if (utest_XPairIdMx(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XDiffMx(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XJukesCantorMx(abc, as, ax, N) != eslOK) return eslFAIL;
#endif
esl_randomness_Destroy(r);
esl_Free2D((void **) as, N);
#ifdef eslAUGMENT_ALPHABET
esl_alphabet_Destroy(abc);
esl_Free2D((void **) ax, N);
#endif
return eslOK;
ERROR:
return eslFAIL;
}
