/* 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_ALPHABET *aa_abc = NULL,
*nt_abc = NULL;
ESL_MSA *msa1 = NULL,
*msa2 = NULL,
*msa3 = NULL,
*msa4 = NULL,
*msa5 = NULL;
double uniform[5] = { 1.0, 1.0, 1.0, 1.0, 1.0 };
double wgt2[5] = { 0.833333, 0.833333, 0.833333, 0.833333, 1.66667 }; /* GSC, PB give same answer */
double gsc3[4] = { 1.125000, 0.875000, 0.875000, 1.125000 };
double pb3[4] = { 1.066667, 1.066667, 0.800000, 1.066667 };
double blosum3[4] = { 1.333333, 0.666667, 0.666667, 1.333333 };
double gsc4[4] = { 0.760870, 0.760870, 1.086957, 1.391304 };
double pb4[4] = { 0.800000, 0.800000, 1.000000, 1.400000 };
double blosum4[4] = { 0.666667, 0.666667, 1.333333, 1.333333 };
if ((aa_abc = esl_alphabet_Create(eslAMINO)) == NULL) esl_fatal("failed to create amino alphabet");
if ((nt_abc = esl_alphabet_Create(eslDNA)) == NULL) esl_fatal("failed to create DNA alphabet");
/* msa1: all sequences identical. Any weighting method should assign uniform weights.
* msa2: "contrived" example of [Henikoff94b]. "Correct" solution is 1==2, 3==4, and 5==2x other weights.
* msa3: the "nitrogenase segments" example of [Henikoff94b].
* msa4: alignment that makes the same distances as Figure 4 from [Gerstein94]
* msa5: gap pathology. no information here, so weighting methods should resort to uniform weights.
*/
if ((msa1 = esl_msa_CreateFromString("# STOCKHOLM 1.0\n\nseq1 AAAAA\nseq2 AAAAA\nseq3 AAAAA\nseq4 AAAAA\nseq5 AAAAA\n//\n",
eslMSAFILE_STOCKHOLM)) == NULL) esl_fatal("msa 1 creation failed");
if ((msa2 = esl_msa_CreateFromString("# STOCKHOLM 1.0\n\nseq1 AAAAA\nseq2 AAAAA\nseq3 CCCCC\nseq4 CCCCC\nseq5 TTTTT\n//\n",
eslMSAFILE_STOCKHOLM)) == NULL) esl_fatal("msa 2 creation failed");
if ((msa3 = esl_msa_CreateFromString("# STOCKHOLM 1.0\n\nNIFE_CLOPA GYVGS\nNIFD_AZOVI GFDGF\nNIFD_BRAJA GYDGF\nNIFK_ANASP GYQGG\n//\n",
eslMSAFILE_STOCKHOLM)) == NULL) esl_fatal("msa 3 creation failed");
if ((msa4 = esl_msa_CreateFromString("# STOCKHOLM 1.0\n\nA AAAAAAAAAA\nB TTAAAAAAAA\nC ATAAAACCCC\nD GGGAAGGGGG\n//\n",
eslMSAFILE_STOCKHOLM)) == NULL) esl_fatal("msa 4 creation failed");
if ((msa5 = esl_msa_CreateFromString("# STOCKHOLM 1.0\n\nA A----\nB -C---\nC --G--\nD ---T-\nE ----T\n//\n",
eslMSAFILE_STOCKHOLM)) == NULL) esl_fatal("msa 5 creation failed");
utest_GSC(aa_abc, msa1, uniform);
utest_GSC(nt_abc, msa1, uniform);
utest_GSC(aa_abc, msa2, wgt2);
utest_GSC(nt_abc, msa2, wgt2);
utest_GSC(aa_abc, msa3, gsc3);
/* no nt test on msa3: it's protein-only */
utest_GSC(aa_abc, msa4, gsc4);
utest_GSC(nt_abc, msa4, gsc4);
utest_GSC(aa_abc, msa5, uniform);
utest_GSC(aa_abc, msa5, uniform);
utest_PB(aa_abc, msa1, uniform);
utest_PB(nt_abc, msa1, uniform);
utest_PB(aa_abc, msa2, wgt2);
utest_PB(nt_abc, msa2, wgt2);
utest_PB(aa_abc, msa3, pb3);
/* no nt test on msa3: it's protein-only */
utest_PB(aa_abc, msa4, pb4);
utest_PB(nt_abc, msa4, pb4);
utest_PB(aa_abc, msa5, uniform);
utest_PB(nt_abc, msa5, uniform);
utest_BLOSUM(aa_abc, msa1, 0.62, uniform);
utest_BLOSUM(nt_abc, msa1, 0.62, uniform);
utest_BLOSUM(aa_abc, msa2, 0.62, wgt2);
utest_BLOSUM(nt_abc, msa2, 0.62, wgt2);
utest_BLOSUM(aa_abc, msa3, 0.62, blosum3);
/* no nt test on msa3: it's protein-only */
utest_BLOSUM(aa_abc, msa4, 0.62, blosum4);
utest_BLOSUM(nt_abc, msa4, 0.62, blosum4);
utest_BLOSUM(aa_abc, msa5, 0.62, uniform);
utest_BLOSUM(nt_abc, msa5, 0.62, uniform);
/* BLOSUM weights have the peculiar property of going flat at maxid=0.0 (everyone
* clusters) or maxid=1.0 (nobody clusters).
*/
utest_BLOSUM(aa_abc, msa4, 0.0, uniform);
utest_BLOSUM(aa_abc, msa4, 1.0, uniform);
esl_msa_Destroy(msa1);
esl_msa_Destroy(msa2);
esl_msa_Destroy(msa3);
esl_msa_Destroy(msa4);
esl_msa_Destroy(msa5);
esl_alphabet_Destroy(aa_abc);
esl_alphabet_Destroy(nt_abc);
exit(0);
}
