/*
* compare to GForward() scores.
*/
static void
utest_fwdback(ESL_RANDOMNESS *r, ESL_ALPHABET *abc, P7_BG *bg, int M, int L, int N)
{
char *msg = "forward/backward unit test failed";
P7_HMM *hmm = NULL;
P7_PROFILE *gm = NULL;
P7_OPROFILE *om = NULL;
ESL_DSQ *dsq = malloc(sizeof(ESL_DSQ) * (L+2));
P7_OMX *fwd = p7_omx_Create(M, 0, L);
P7_OMX *bck = p7_omx_Create(M, 0, L);
P7_OMX *oxf = p7_omx_Create(M, L, L);
P7_OMX *oxb = p7_omx_Create(M, L, L);
P7_GMX *gx = p7_gmx_Create(M, L);
float tolerance;
float fsc1, fsc2;
float bsc1, bsc2;
float generic_sc;
p7_FLogsumInit();
if (p7_FLogsumError(-0.4, -0.5) > 0.0001) tolerance = 1.0; /* weaker test against GForward() */
else tolerance = 0.0001; /* stronger test: FLogsum() is in slow exact mode. */
p7_oprofile_Sample(r, abc, bg, M, L, &hmm, &gm, &om);
while (N--)
{
esl_rsq_xfIID(r, bg->f, abc->K, L, dsq);
p7_Forward (dsq, L, om, oxf, &fsc1);
p7_Backward (dsq, L, om, oxf, oxb, &bsc1);
p7_ForwardParser (dsq, L, om, fwd, &fsc2);
p7_BackwardParser(dsq, L, om, fwd, bck, &bsc2);
p7_GForward (dsq, L, gm, gx, &generic_sc);
/* Forward and Backward scores should agree with high tolerance */
if (fabs(fsc1-bsc1) > 0.0001) esl_fatal(msg);
if (fabs(fsc2-bsc2) > 0.0001) esl_fatal(msg);
if (fabs(fsc1-fsc2) > 0.0001) esl_fatal(msg);
/* GForward scores should approximate Forward scores,
* with tolerance that depends on how logsum.c was compiled
*/
if (fabs(fsc1-generic_sc) > tolerance) esl_fatal(msg);
}
free(dsq);
p7_hmm_Destroy(hmm);
p7_omx_Destroy(oxb);
p7_omx_Destroy(oxf);
p7_omx_Destroy(bck);
p7_omx_Destroy(fwd);
p7_gmx_Destroy(gx);
p7_profile_Destroy(gm);
p7_oprofile_Destroy(om);
}
/* Function: p7_Tau()
* Synopsis: Determine Forward tau by brief simulation.
* Incept: SRE, Thu Aug 9 15:08:39 2007 [Janelia]
*
* Purpose: Determine the <tau> parameter for an exponential tail fit
* to the Forward score distribution for model <om>, on
* random sequences with the composition of the background
* model <bg>. This <tau> parameter is for an exponential
* distribution anchored from $P=1.0$, so it's not really a
* tail per se; but it's only an accurate fit in the tail
* of the Forward score distribution, from about $P=0.001$
* or so.
*
* The determination of <tau> is done by a brief simulation
* in which we fit a Gumbel distribution to a small number
* of Forward scores of random sequences, and use that to
* predict the location of the tail at probability <tailp>.
*
* The Gumbel is of course inaccurate, but we can use it
* here solely as an empirical distribution to determine
* the location of a reasonable <tau> more accurately on a
* smaller number of samples than we could do with raw
* order statistics.
*
* Typical choices are L=100, N=200, tailp=0.04, which
* typically yield estimates $\hat{\mu}$ with a precision
* (standard deviation) of $\pm$ 0.2 bits, corresponding to
* a $\pm$ 15\% error in E-values. See [J1/135].
*
* The use of Gumbel fitting to a small number of $N$
* samples and the extrapolation of $\hat{\mu}$ from the
* estimated location of the 0.04 tail mass are both
* empirical and carefully optimized against several
* tradeoffs. Most importantly, around this choice of tail
* probability, a systematic error introduced by the use of
* the Gumbel fit is being cancelled by systematic error
* introduced by the use of a higher tail probability than
* the regime in which the exponential tail is a valid
* approximation. See [J1/135] for discussion.
*
* This function changes the length configuration of both
* <om> and <bg>. The caller must remember to reconfigure
* both of their length models appropriately for any
* subsequent alignments.
*
* Args: r : source of randomness
* om : configured profile to sample sequences from
* bg : null model (for background residue frequencies)
* L : mean length model for seq emission from profile
* N : number of sequences to generate
* lambda : expected slope of the exponential tail (from p7_Lambda())
* tailp : tail mass from which we will extrapolate mu
* ret_mu : RETURN: estimate for the Forward mu (base of exponential tail)
*
* Returns: <eslOK> on success, and <*ret_fv> is the score difference
* in bits.
*
* Throws: <eslEMEM> on allocation error, and <*ret_fv> is 0.
*/
int
p7_Tau(ESL_RANDOMNESS *r, P7_OPROFILE *om, P7_BG *bg, int L, int N, double lambda, double tailp, double *ret_tau)
{
P7_OMX *ox = p7_omx_Create(om->M, 0, L); /* DP matrix: for ForwardParser, L rows */
ESL_DSQ *dsq = NULL;
double *xv = NULL;
float fsc, nullsc;
double gmu, glam;
int status;
int i;
ESL_ALLOC(xv, sizeof(double) * N);
ESL_ALLOC(dsq, sizeof(ESL_DSQ) * (L+2));
if (ox == NULL) { status = eslEMEM; goto ERROR; }
p7_oprofile_ReconfigLength(om, L);
p7_bg_SetLength(bg, L);
for (i = 0; i < N; i++)
{
if ((status = esl_rsq_xfIID(r, bg->f, om->abc->K, L, dsq)) != eslOK) goto ERROR;
if ((status = p7_ForwardParser(dsq, L, om, ox, &fsc)) != eslOK) goto ERROR;
if ((status = p7_bg_NullOne(bg, dsq, L, &nullsc)) != eslOK) goto ERROR;
xv[i] = (fsc - nullsc) / eslCONST_LOG2;
}
if ((status = esl_gumbel_FitComplete(xv, N, &gmu, &glam)) != eslOK) goto ERROR;
/* Explanation of the eqn below: first find the x at which the Gumbel tail
* mass is predicted to be equal to tailp. Then back up from that x
* by log(tailp)/lambda to set the origin of the exponential tail to 1.0
* instead of tailp.
*/
*ret_tau = esl_gumbel_invcdf(1.0-tailp, gmu, glam) + (log(tailp) / lambda);
free(xv);
free(dsq);
p7_omx_Destroy(ox);
return eslOK;
ERROR:
*ret_tau = 0.;
if (xv != NULL) free(xv);
if (dsq != NULL) free(dsq);
if (ox != NULL) p7_omx_Destroy(ox);
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 1, argc, argv, banner, usage);
char *hmmfile = esl_opt_GetArg(go, 1);
ESL_STOPWATCH *w = esl_stopwatch_Create();
ESL_RANDOMNESS *r = esl_randomness_CreateFast(esl_opt_GetInteger(go, "-s"));
ESL_ALPHABET *abc = NULL;
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_OMX *bck = NULL;
int L = esl_opt_GetInteger(go, "-L");
int N = esl_opt_GetInteger(go, "-N");
ESL_DSQ *dsq = malloc(sizeof(ESL_DSQ) * (L+2));
int i;
float fsc, bsc;
float fsc2, bsc2;
double base_time, bench_time, Mcs;
p7_FLogsumInit();
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");
bg = p7_bg_Create(abc);
p7_bg_SetLength(bg, L);
gm = p7_profile_Create(hmm->M, abc);
p7_ProfileConfig(hmm, bg, gm, L, p7_LOCAL);
om = p7_oprofile_Create(gm->M, abc);
p7_oprofile_Convert(gm, om);
p7_oprofile_ReconfigLength(om, L);
if (esl_opt_GetBoolean(go, "-x") && p7_FLogsumError(-0.4, -0.5) > 0.0001)
p7_Fail("-x here requires p7_Logsum() recompiled in slow exact mode");
if (esl_opt_GetBoolean(go, "-P")) {
fwd = p7_omx_Create(gm->M, 0, L);
bck = p7_omx_Create(gm->M, 0, L);
} else {
fwd = p7_omx_Create(gm->M, L, L);
bck = p7_omx_Create(gm->M, L, L);
}
gx = p7_gmx_Create(gm->M, L);
/* Get a baseline time: how long it takes just to generate the sequences */
esl_stopwatch_Start(w);
for (i = 0; i < N; i++) esl_rsq_xfIID(r, bg->f, abc->K, L, dsq);
esl_stopwatch_Stop(w);
base_time = w->user;
esl_stopwatch_Start(w);
for (i = 0; i < N; i++)
{
esl_rsq_xfIID(r, bg->f, abc->K, L, dsq);
if (esl_opt_GetBoolean(go, "-P")) {
if (! esl_opt_GetBoolean(go, "-B")) p7_ForwardParser (dsq, L, om, fwd, &fsc);
if (! esl_opt_GetBoolean(go, "-F")) p7_BackwardParser(dsq, L, om, fwd, bck, &bsc);
} else {
if (! esl_opt_GetBoolean(go, "-B")) p7_Forward (dsq, L, om, fwd, &fsc);
if (! esl_opt_GetBoolean(go, "-F")) p7_Backward(dsq, L, om, fwd, bck, &bsc);
}
if (esl_opt_GetBoolean(go, "-c") || esl_opt_GetBoolean(go, "-x"))
{
p7_GForward (dsq, L, gm, gx, &fsc2);
p7_GBackward(dsq, L, gm, gx, &bsc2);
printf("%.4f %.4f %.4f %.4f\n", fsc, bsc, fsc2, bsc2);
}
}
esl_stopwatch_Stop(w);
bench_time = w->user - base_time;
Mcs = (double) N * (double) L * (double) gm->M * 1e-6 / (double) bench_time;
esl_stopwatch_Display(stdout, w, "# CPU time: ");
printf("# M = %d\n", gm->M);
printf("# %.1f Mc/s\n", Mcs);
free(dsq);
p7_omx_Destroy(bck);
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_alphabet_Destroy(abc);
esl_stopwatch_Destroy(w);
esl_randomness_Destroy(r);
esl_getopts_Destroy(go);
return 0;
}
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_OPROFILE *om = NULL;
P7_GMX *gx = NULL;
P7_OMX *fwd = NULL;
P7_OMX *bck = NULL;
ESL_SQ *sq = NULL;
ESL_SQFILE *sqfp = NULL;
int format = eslSQFILE_UNKNOWN;
float fraw, braw, nullsc, fsc;
float gfraw, gbraw, gfsc;
double P, gP;
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");
/* Open sequence file for reading */
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);
/* 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_UNILOCAL);
om = p7_oprofile_Create(gm->M, abc);
p7_oprofile_Convert(gm, om);
/* p7_oprofile_Dump(stdout, om); */
/* allocate DP matrices for O(M+L) parsers */
fwd = p7_omx_Create(gm->M, 0, sq->n);
bck = p7_omx_Create(gm->M, 0, sq->n);
gx = p7_gmx_Create(gm->M, sq->n);
/* allocate DP matrices for O(ML) fills */
/* fwd = p7_omx_Create(gm->M, sq->n, sq->n); */
/* bck = p7_omx_Create(gm->M, sq->n, sq->n); */
/* p7_omx_SetDumpMode(stdout, fwd, TRUE); */ /* makes the fast DP algorithms dump their matrices */
/* p7_omx_SetDumpMode(stdout, bck, TRUE); */
while ((status = esl_sqio_Read(sqfp, sq)) == eslOK)
{
p7_oprofile_ReconfigLength(om, sq->n);
p7_ReconfigLength(gm, sq->n);
p7_bg_SetLength(bg, sq->n);
p7_omx_GrowTo(fwd, om->M, 0, sq->n);
p7_omx_GrowTo(bck, om->M, 0, sq->n);
p7_gmx_GrowTo(gx, gm->M, sq->n);
p7_bg_NullOne (bg, sq->dsq, sq->n, &nullsc);
p7_ForwardParser (sq->dsq, sq->n, om, fwd, &fraw);
p7_BackwardParser(sq->dsq, sq->n, om, fwd, bck, &braw);
/* p7_Forward (sq->dsq, sq->n, om, fwd, &fsc); printf("forward: %.2f nats\n", fsc); */
/* p7_Backward(sq->dsq, sq->n, om, fwd, bck, &bsc); printf("backward: %.2f nats\n", bsc); */
/* Comparison to other F/B implementations */
p7_GForward (sq->dsq, sq->n, gm, gx, &gfraw);
p7_GBackward (sq->dsq, sq->n, gm, gx, &gbraw);
/* p7_gmx_Dump(stdout, gx); */
fsc = (fraw-nullsc) / eslCONST_LOG2;
gfsc = (gfraw-nullsc) / eslCONST_LOG2;
P = esl_exp_surv(fsc, om->evparam[p7_FTAU], om->evparam[p7_FLAMBDA]);
gP = esl_exp_surv(gfsc, gm->evparam[p7_FTAU], gm->evparam[p7_FLAMBDA]);
if (esl_opt_GetBoolean(go, "-1"))
{
printf("%-30s\t%-20s\t%9.2g\t%6.1f\t%9.2g\t%6.1f\n", sq->name, hmm->name, P, fsc, gP, gfsc);
}
else if (esl_opt_GetBoolean(go, "-P"))
{ /* output suitable for direct use in profmark benchmark postprocessors: */
printf("%g\t%.2f\t%s\t%s\n", P, fsc, sq->name, hmm->name);
}
else
{
printf("target sequence: %s\n", sq->name);
printf("fwd filter raw score: %.2f nats\n", fraw);
printf("bck filter raw score: %.2f nats\n", braw);
printf("null score: %.2f nats\n", nullsc);
printf("per-seq score: %.2f bits\n", fsc);
printf("P-value: %g\n", P);
printf("GForward raw score: %.2f nats\n", gfraw);
printf("GBackward raw score: %.2f nats\n", gbraw);
printf("GForward seq score: %.2f bits\n", gfsc);
printf("GForward P-value: %g\n", gP);
}
esl_sq_Reuse(sq);
}
/* cleanup */
esl_sq_Destroy(sq);
esl_sqfile_Close(sqfp);
p7_omx_Destroy(bck);
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_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
