/* 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 = esl_getopts_CreateDefaultApp(options, 2, argc, argv, banner, usage);
char *hmmfile = esl_opt_GetArg(go, 1);
char *seqfile = esl_opt_GetArg(go, 2);
ESL_ALPHABET *abc = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
P7_GMX *fwd = NULL;
ESL_SQ *sq = NULL;
ESL_SQFILE *sqfp = NULL;
P7_TRACE *tr = NULL;
int format = eslSQFILE_UNKNOWN;
char errbuf[eslERRBUFSIZE];
float sc;
int d;
int status;
/* Read in one HMM */
if (p7_hmmfile_Open(hmmfile, NULL, &hfp) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile);
if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM");
p7_hmmfile_Close(hfp);
/* Read in one sequence */
sq = esl_sq_CreateDigital(abc);
status = esl_sqfile_Open(seqfile, format, NULL, &sqfp);
if (status == eslENOTFOUND) p7_Fail("No such file.");
else if (status == eslEFORMAT) p7_Fail("Format unrecognized.");
else if (status == eslEINVAL) p7_Fail("Can't autodetect stdin or .gz.");
else if (status != eslOK) p7_Fail("Open failed, code %d.", status);
if (esl_sqio_Read(sqfp, sq) != eslOK) p7_Fail("Failed to read sequence");
esl_sqfile_Close(sqfp);
/* Configure a profile from the HMM */
bg = p7_bg_Create(abc);
p7_bg_SetLength(bg, sq->n);
gm = p7_profile_Create(hmm->M, abc);
p7_ProfileConfig(hmm, bg, gm, sq->n, p7_LOCAL);
/* Allocate matrix and a trace */
fwd = p7_gmx_Create(gm->M, sq->n);
tr = p7_trace_Create();
/* Run Viterbi; do traceback */
p7_GViterbi (sq->dsq, sq->n, gm, fwd, &sc);
p7_GTrace (sq->dsq, sq->n, gm, fwd, tr);
/* Dump and validate the trace. */
p7_trace_Dump(stdout, tr, gm, sq->dsq);
if (p7_trace_Validate(tr, abc, sq->dsq, errbuf) != eslOK) p7_Die("trace fails validation:\n%s\n", errbuf);
/* Domain info in the trace. */
p7_trace_Index(tr);
printf("# Viterbi: %d domains : ", tr->ndom);
for (d = 0; d < tr->ndom; d++) printf("%6d %6d %6d %6d ", tr->sqfrom[d], tr->sqto[d], tr->hmmfrom[d], tr->hmmto[d]);
printf("\n");
/* Cleanup */
p7_trace_Destroy(tr);
p7_gmx_Destroy(fwd);
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
esl_sq_Destroy(sq);
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);
char *hmmfile = esl_opt_GetArg(go, 1);
char *seqfile = esl_opt_GetArg(go, 2);
ESL_ALPHABET *abc = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
P7_GMX *gx1 = NULL;
P7_GMX *gx2 = NULL;
ESL_SQ *sq = NULL;
ESL_SQFILE *sqfp = NULL;
P7_TRACE *tr = NULL;
int format = eslSQFILE_UNKNOWN;
char errbuf[eslERRBUFSIZE];
float fsc, bsc, vsc;
float accscore;
int status;
/* Read in one HMM */
if (p7_hmmfile_OpenE(hmmfile, NULL, &hfp, NULL) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile);
if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM");
p7_hmmfile_Close(hfp);
/* Read in one sequence */
sq = esl_sq_CreateDigital(abc);
status = esl_sqfile_OpenDigital(abc, seqfile, format, NULL, &sqfp);
if (status == eslENOTFOUND) p7_Fail("No such file.");
else if (status == eslEFORMAT) p7_Fail("Format unrecognized.");
else if (status == eslEINVAL) p7_Fail("Can't autodetect stdin or .gz.");
else if (status != eslOK) p7_Fail("Open failed, code %d.", status);
if (esl_sqio_Read(sqfp, sq) != eslOK) p7_Fail("Failed to read sequence");
esl_sqfile_Close(sqfp);
/* Configure a profile from the HMM */
bg = p7_bg_Create(abc);
p7_bg_SetLength(bg, sq->n);
gm = p7_profile_Create(hmm->M, abc);
p7_ProfileConfig(hmm, bg, gm, sq->n, p7_LOCAL); /* multihit local: H3 default */
/* Allocations */
gx1 = p7_gmx_Create(gm->M, sq->n);
gx2 = p7_gmx_Create(gm->M, sq->n);
tr = p7_trace_CreateWithPP();
p7_FLogsumInit();
/* Run Forward, Backward; do OA fill and trace */
p7_GForward (sq->dsq, sq->n, gm, gx1, &fsc);
p7_GBackward(sq->dsq, sq->n, gm, gx2, &bsc);
p7_GDecoding(gm, gx1, gx2, gx2); /* <gx2> is now the posterior decoding matrix */
p7_GOptimalAccuracy(gm, gx2, gx1, &accscore); /* <gx1> is now the OA matrix */
p7_GOATrace(gm, gx2, gx1, tr);
if (esl_opt_GetBoolean(go, "-d")) p7_gmx_Dump(stdout, gx2, p7_DEFAULT);
if (esl_opt_GetBoolean(go, "-m")) p7_gmx_Dump(stdout, gx1, p7_DEFAULT);
p7_trace_Dump(stdout, tr, gm, sq->dsq);
if (p7_trace_Validate(tr, abc, sq->dsq, errbuf) != eslOK) p7_Die("trace fails validation:\n%s\n", errbuf);
printf("fwd = %.4f nats\n", fsc);
printf("bck = %.4f nats\n", bsc);
printf("acc = %.4f (%.2f%%)\n", accscore, accscore * 100. / (float) sq->n);
p7_trace_Reuse(tr);
p7_GViterbi(sq->dsq, sq->n, gm, gx1, &vsc);
p7_GTrace (sq->dsq, sq->n, gm, gx1, tr);
p7_trace_SetPP(tr, gx2);
p7_trace_Dump(stdout, tr, gm, sq->dsq);
printf("vit = %.4f nats\n", vsc);
printf("acc = %.4f\n", p7_trace_GetExpectedAccuracy(tr));
/* Cleanup */
esl_sq_Destroy(sq);
p7_trace_Destroy(tr);
p7_gmx_Destroy(gx1);
p7_gmx_Destroy(gx2);
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
/*
* 1. Compare accscore to GOptimalAccuracy().
* 2. Compare trace to GOATrace().
*
* Note: This test is subject to some expected noise and can fail
* for entirely innocent reasons. Generic Forward/Backward calculations with
* p7_GForward(), p7_GBackward() use coarse-grain table lookups to sum
* log probabilities, and sufficient roundoff error can accumulate to
* change the optimal accuracy traceback, causing this test to fail.
* So, if optacc_utest fails, before you go looking for bugs, first
* go to ../logsum.c, change the #ifdef to activate the slow/accurate
* version, recompile and rerun optacc_utest. If the failure goes away,
* you can ignore it. - SRE, Wed Dec 17 09:45:31 2008
*/
static void
utest_optacc(ESL_GETOPTS *go, ESL_RANDOMNESS *r, ESL_ALPHABET *abc, P7_BG *bg, int M, int L, int N)
{
char *msg = "optimal accuracy unit test failed";
P7_HMM *hmm = NULL;
P7_PROFILE *gm = NULL;
P7_OPROFILE *om = NULL;
ESL_SQ *sq = esl_sq_CreateDigital(abc);
P7_OMX *ox1 = p7_omx_Create(M, L, L);
P7_OMX *ox2 = p7_omx_Create(M, L, L);
P7_GMX *gx1 = p7_gmx_Create(M, L);
P7_GMX *gx2 = p7_gmx_Create(M, L);
P7_TRACE *tr = p7_trace_CreateWithPP();
P7_TRACE *trg = p7_trace_CreateWithPP();
P7_TRACE *tro = p7_trace_CreateWithPP();
float accscore_o;
float fsc, bsc, accscore;
float fsc_g, bsc_g, accscore_g, accscore_g2;
float pptol = 0.01;
float sctol = 0.001;
float gtol;
p7_FLogsumInit();
gtol = ( (p7_FLogsumError(-0.4, -0.5) > 0.0001) ? 0.1 : 0.001);
if (p7_oprofile_Sample(r, abc, bg, M, L, &hmm, &gm, &om)!= eslOK) esl_fatal(msg);
while (N--)
{
if (p7_ProfileEmit(r, hmm, gm, bg, sq, tro) != eslOK) esl_fatal(msg);
if (p7_omx_GrowTo(ox1, M, sq->n, sq->n) != eslOK) esl_fatal(msg);
if (p7_omx_GrowTo(ox2, M, sq->n, sq->n) != eslOK) esl_fatal(msg);
if (p7_gmx_GrowTo(gx1, M, sq->n) != eslOK) esl_fatal(msg);
if (p7_gmx_GrowTo(gx2, M, sq->n) != eslOK) esl_fatal(msg);
if (p7_Forward (sq->dsq, sq->n, om, ox1, &fsc) != eslOK) esl_fatal(msg);
if (p7_Backward(sq->dsq, sq->n, om, ox1, ox2, &bsc) != eslOK) esl_fatal(msg);
if (p7_Decoding(om, ox1, ox2, ox2) != eslOK) esl_fatal(msg);
if (p7_OptimalAccuracy(om, ox2, ox1, &accscore) != eslOK) esl_fatal(msg);
#if 0
p7_omx_FDeconvert(ox1, gx1);
p7_gmx_Dump(stdout, gx1, p7_DEFAULT);
p7_omx_FDeconvert(ox2, gx1);
p7_gmx_Dump(stdout, gx1, p7_DEFAULT);
#endif
if (p7_OATrace(om, ox2, ox1, tr) != eslOK) esl_fatal(msg);
if (p7_GForward (sq->dsq, sq->n, gm, gx1, &fsc_g) != eslOK) esl_fatal(msg);
if (p7_GBackward(sq->dsq, sq->n, gm, gx2, &bsc_g) != eslOK) esl_fatal(msg);
#if 0
p7_gmx_Dump(stdout, gx1, p7_DEFAULT); /* fwd */
p7_gmx_Dump(stdout, gx2, p7_DEFAULT); /* bck */
#endif
if (p7_GDecoding(gm, gx1, gx2, gx2) != eslOK) esl_fatal(msg);
if (p7_GOptimalAccuracy(gm, gx2, gx1, &accscore_g) != eslOK) esl_fatal(msg);
#if 0
p7_gmx_Dump(stdout, gx1, p7_DEFAULT); /* oa */
p7_gmx_Dump(stdout, gx2, p7_DEFAULT); /* pp */
#endif
if (p7_GOATrace(gm, gx2, gx1, trg) != eslOK) esl_fatal(msg);
if (p7_trace_SetPP(tro, gx2) != eslOK) esl_fatal(msg);
if (esl_opt_GetBoolean(go, "--traces"))
{
p7_trace_Dump(stdout, tro, gm, sq->dsq);
p7_trace_Dump(stdout, tr, gm, sq->dsq);
p7_trace_Dump(stdout, trg, gm, sq->dsq);
}
if (p7_trace_Validate(tr, abc, sq->dsq, NULL) != eslOK) esl_fatal(msg);
if (p7_trace_Validate(trg, abc, sq->dsq, NULL) != eslOK) esl_fatal(msg);
if (p7_trace_Compare(tr, trg, pptol) != eslOK) esl_fatal(msg);
accscore_o = p7_trace_GetExpectedAccuracy(tro); /* according to gx2; see p7_trace_SetPP() call above */
accscore_g2 = p7_trace_GetExpectedAccuracy(trg);
#if 0
printf("%f %f %f %f\n", accscore, accscore_g, accscore_g2, accscore_o);
#endif
if (esl_FCompare(fsc, bsc, sctol) != eslOK) esl_fatal(msg);
if (esl_FCompare(fsc_g, bsc_g, gtol) != eslOK) esl_fatal(msg);
if (esl_FCompare(fsc, fsc_g, gtol) != eslOK) esl_fatal(msg);
if (esl_FCompare(accscore, accscore_g, gtol) != eslOK) esl_fatal(msg);
if (esl_FCompare(accscore_g, accscore_g2, gtol) != eslOK) esl_fatal(msg);
if (accscore_g2 < accscore_o) esl_fatal(msg);
/* the above deserves explanation:
* - accscore_o is the accuracy of the originally emitted trace, according
* to the generic posterior decoding matrix <gx2>. This is a lower bound
* on the expected # of accurately aligned residues found by a DP
* optimization.
* - accscore is the accuracy found by the fast (vector) code DP implementation.
* - accscore_g is the accuracy found by the generic DP implementation.
* accscore and accscore_g should be nearly identical,
* within tolerance of roundoff error accumulation and
* the imprecision of Logsum() tables.
* - accscore_g2 is the accuracy of the traceback identified by the generic
* DP implementation. It should be identical (within order-of-evaluation
* roundoff error) to accscore_g.
*
* the "accscore_g2 < accscore_o" test is carefully contrived.
* accscore_o is a theoretical lower bound but because of fp error,
* accscore and (much more rarely) even accscore_g can exceed accscore_o.
* accscore_g2, however, is calculated with identical order of evaluation
* as accscore_o if the optimal trace does turn out to be identical to
* the originally emitted trace. It should be extremely unlikely (though
* not impossible) for accscore_o to exceed accscore_g2. (The DP algorithm
* would have to identify a trace that was different than the original trace,
* which the DP algorithm, by order-of-evaluation, assigned higher accuracy,
* but order-of-evaluation in traceback dependent code assigned lower accuracy.
* [xref J5/29]
*/
esl_sq_Reuse(sq);
p7_trace_Reuse(tr);
p7_trace_Reuse(trg);
p7_trace_Reuse(tro);
}
p7_trace_Destroy(tro);
p7_trace_Destroy(trg);
p7_trace_Destroy(tr);
p7_gmx_Destroy(gx2);
p7_gmx_Destroy(gx1);
p7_omx_Destroy(ox2);
p7_omx_Destroy(ox1);
esl_sq_Destroy(sq);
p7_oprofile_Destroy(om);
p7_profile_Destroy(gm);
p7_hmm_Destroy(hmm);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 2, argc, argv, banner, usage);
char *hmmfile = esl_opt_GetArg(go, 1);
char *seqfile = esl_opt_GetArg(go, 2);
ESL_ALPHABET *abc = NULL;
ESL_RANDOMNESS *rng = esl_randomness_CreateFast(0);
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
P7_OPROFILE *om = NULL;
P7_GMX *gx = NULL;
P7_OMX *fwd = NULL;
P7_TRACE *tr = NULL;
ESL_SQ *sq = NULL;
ESL_SQFILE *sqfp = NULL;
int format = eslSQFILE_UNKNOWN;
int N = esl_opt_GetInteger(go, "-N");
int i;
float vsc, fsc, tsc;
char errbuf[eslERRBUFSIZE];
int status;
/* Read in one HMM */
if (p7_hmmfile_Open(hmmfile, NULL, &hfp) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile);
if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM");
/* Read in one sequence */
sq = esl_sq_CreateDigital(abc);
status = esl_sqfile_Open(seqfile, format, NULL, &sqfp);
if (status == eslENOTFOUND) p7_Fail("No such file.");
else if (status == eslEFORMAT) p7_Fail("Format unrecognized.");
else if (status == eslEINVAL) p7_Fail("Can't autodetect stdin or .gz.");
else if (status != eslOK) p7_Fail("Open failed, code %d.", status);
if (esl_sqio_Read(sqfp, sq) != eslOK) p7_Fail("Failed to read sequence");
/* create default null model, then create and optimize profile */
bg = p7_bg_Create(abc); p7_bg_SetLength(bg, sq->n);
gm = p7_profile_Create(hmm->M, abc); p7_ProfileConfig(hmm, bg, gm, sq->n, p7_LOCAL);
om = p7_oprofile_Create(gm->M, abc); p7_oprofile_Convert(gm, om);
if (esl_opt_GetBoolean(go, "-p")) p7_oprofile_Dump(stdout, om);
fwd = p7_omx_Create(gm->M, sq->n, sq->n);
gx = p7_gmx_Create(gm->M, sq->n);
tr = p7_trace_Create();
if (esl_opt_GetBoolean(go, "-m") == TRUE) p7_omx_SetDumpMode(stdout, fwd, TRUE);
p7_GViterbi(sq->dsq, sq->n, gm, gx, &vsc);
p7_Forward (sq->dsq, sq->n, om, fwd, &fsc);
for (i = 0; i < N; i++)
{
p7_StochasticTrace(rng, sq->dsq, sq->n, om, fwd, tr);
p7_trace_Score(tr, sq->dsq, gm, &tsc);
if (esl_opt_GetBoolean(go, "-t") == TRUE) p7_trace_Dump(stdout, tr, gm, sq->dsq);
if (p7_trace_Validate(tr, abc, sq->dsq, errbuf) != eslOK) p7_Die("trace %d fails validation:\n%s\n", i, errbuf);
printf("Sampled trace: %.4f nats\n", tsc);
p7_trace_Reuse(tr);
}
printf("Forward score: %.4f nats\n", fsc);
printf("Viterbi score: %.4f nats\n", vsc);
/* cleanup */
esl_sq_Destroy(sq);
esl_sqfile_Close(sqfp);
p7_trace_Destroy(tr);
p7_omx_Destroy(fwd);
p7_gmx_Destroy(gx);
p7_oprofile_Destroy(om);
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
p7_hmmfile_Close(hfp);
esl_randomness_Destroy(rng);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
