static int
serial_loop(WORKER_INFO *info, struct cfg_s *cfg)
{
P7_BUILDER *bld = NULL;
ESL_MSA *msa = NULL;
ESL_MSA *postmsa = NULL;
ESL_MSA **postmsa_ptr = (cfg->postmsafile != NULL) ? &postmsa : NULL;
P7_HMM *hmm = NULL;
char errmsg[eslERRBUFSIZE];
int status;
double entropy;
cfg->nali = 0;
while ((status = esl_msa_Read(cfg->afp, &msa)) == eslOK)
{
cfg->nali++;
if ((status = set_msa_name(cfg, errmsg, msa)) != eslOK) p7_Fail("%s\n", errmsg); /* cfg->nnamed gets incremented in this call */
/* bg new-HMM trarr gm om */
if ((status = p7_Builder(info->bld, msa, info->bg, &hmm, NULL, NULL, NULL, postmsa_ptr)) != eslOK) p7_Fail("build failed: %s", bld->errbuf);
entropy = p7_MeanMatchRelativeEntropy(hmm, info->bg);
if ((status = output_result(cfg, errmsg, cfg->nali, msa, hmm, postmsa, entropy)) != eslOK) p7_Fail(errmsg);
p7_hmm_Destroy(hmm);
esl_msa_Destroy(msa);
esl_msa_Destroy(postmsa);
}
return status;
}
/* Function: p7_Lambda()
* Synopsis: Determines length-corrected local lambda parameter.
* Incept: SRE, Wed Aug 8 17:54:55 2007 [Janelia]
*
* Purpose: Determine the effective scale parameter $\hat{\lambda}$ to
* use for model <hmm>. This will be applied both to
* Viterbi Gumbel distributions and Forward exponential
* tails.
*
* The 'true' $\lambda$ is always $\log 2 = 0.693$. The effective
* lambda is corrected for edge effect, using the equation
*
* \[
* \hat{\lambda} = \lambda + \frac{1.44}{MH}
* \]
*
* where $M$ is the model length and $H$ is the model
* relative entropy. The model relative entropy is
* approximated by the average relative entropy of match
* emission distributions. The 1.44 is an empirically
* determined fudge factor [J1/125]. This edge-effect
* correction is based largely on \citep{Altschul01},
* except for the fudge factor, which we don't understand
* and can't theoretically justify.
*
* Args: hmm : model to calculate corrected lambda for
* bg : null model (source of background frequencies)
* ret_lambda : RETURN: edge-corrected lambda
*
* Returns: <eslOK> on success, and <*ret_lambda> is the result.
*
* Throws: (no abnormal error conditions)
*/
int
p7_Lambda(P7_HMM *hmm, P7_BG *bg, double *ret_lambda)
{
double H = p7_MeanMatchRelativeEntropy(hmm, bg);
*ret_lambda = eslCONST_LOG2 + 1.44 / ((double) hmm->M * H);
return eslOK;
}
/* Evaluate fx = rel entropy - etarget, which we want to be = 0,
* for effective sequence number <x>.
*/
static int
eweight_target_f(double Neff, void *params, double *ret_fx)
{
struct ew_param_s *p = (struct ew_param_s *) params;
p7_hmm_CopyParameters(p->hmm, p->h2);
p7_hmm_Scale(p->h2, Neff / (double) p->h2->nseq);
p7_ParameterEstimation(p->h2, p->pri);
*ret_fx = p7_MeanMatchRelativeEntropy(p->h2, p->bg) - p->etarget;
return eslOK;
}
static void
pipeline_thread(void *arg)
{
int workeridx;
int status;
WORK_ITEM *item;
void *newItem;
WORKER_INFO *info;
ESL_THREADS *obj;
obj = (ESL_THREADS *) arg;
esl_threads_Started(obj, &workeridx);
info = (WORKER_INFO *) esl_threads_GetData(obj, workeridx);
status = esl_workqueue_WorkerUpdate(info->queue, NULL, &newItem);
if (status != eslOK) esl_fatal("Work queue worker failed");
/* loop until all blocks have been processed */
item = (WORK_ITEM *) newItem;
while (item->msa != NULL)
{
status = p7_Builder(info->bld, item->msa, info->bg, &item->hmm, NULL, NULL, NULL, &item->postmsa);
if (status != eslOK) p7_Fail("build failed: %s", info->bld->errbuf);
item->entropy = p7_MeanMatchRelativeEntropy(item->hmm, info->bg);
item->processed = TRUE;
status = esl_workqueue_WorkerUpdate(info->queue, item, &newItem);
if (status != eslOK) esl_fatal("Work queue worker failed");
item = (WORK_ITEM *) newItem;
}
status = esl_workqueue_WorkerUpdate(info->queue, item, NULL);
if (status != eslOK) esl_fatal("Work queue worker failed");
esl_threads_Finished(obj, workeridx);
return;
}
/* mpi_master()
* The MPI version of hmmbuild.
* Follows standard pattern for a master/worker load-balanced MPI program (J1/78-79).
*
* A master can only return if it's successful.
* Errors in an MPI master come in two classes: recoverable and nonrecoverable.
*
* Recoverable errors include all worker-side errors, and any
* master-side error that do not affect MPI communication. Error
* messages from recoverable messages are delayed until we've cleanly
* shut down the workers.
*
* Unrecoverable errors are master-side errors that may affect MPI
* communication, meaning we cannot count on being able to reach the
* workers and shut them down. Unrecoverable errors result in immediate
* p7_Fail()'s, which will cause MPI to shut down the worker processes
* uncleanly.
*/
static void
mpi_master(const ESL_GETOPTS *go, struct cfg_s *cfg)
{
int xstatus = eslOK; /* changes from OK on recoverable error */
int status;
int have_work = TRUE; /* TRUE while alignments remain */
int nproc_working = 0; /* number of worker processes working, up to nproc-1 */
int wi; /* rank of next worker to get an alignment to work on */
char *buf = NULL; /* input/output buffer, for packed MPI messages */
int bn = 0;
ESL_MSA *msa = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
ESL_MSA **msalist = NULL;
ESL_MSA *postmsa = NULL;
int *msaidx = NULL;
char errmsg[eslERRBUFSIZE];
MPI_Status mpistatus;
int n;
int pos;
double entropy;
/* Master initialization: including, figure out the alphabet type.
* If any failure occurs, delay printing error message until we've shut down workers.
*/
if (xstatus == eslOK) { if ((status = init_master_cfg(go, cfg, errmsg)) != eslOK) xstatus = status; }
if (xstatus == eslOK) { bn = 4096; if ((buf = malloc(sizeof(char) * bn)) == NULL) { sprintf(errmsg, "allocation failed"); xstatus = eslEMEM; } }
if (xstatus == eslOK) { if ((msalist = malloc(sizeof(ESL_MSA *) * cfg->nproc)) == NULL) { sprintf(errmsg, "allocation failed"); xstatus = eslEMEM; } }
if (xstatus == eslOK) { if ((msaidx = malloc(sizeof(int) * cfg->nproc)) == NULL) { sprintf(errmsg, "allocation failed"); xstatus = eslEMEM; } }
MPI_Bcast(&xstatus, 1, MPI_INT, 0, MPI_COMM_WORLD);
if (xstatus != eslOK) { MPI_Finalize(); p7_Fail(errmsg); }
ESL_DPRINTF1(("MPI master is initialized\n"));
bg = p7_bg_Create(cfg->abc);
for (wi = 0; wi < cfg->nproc; wi++) { msalist[wi] = NULL; msaidx[wi] = 0; }
/* Worker initialization:
* Because we've already successfully initialized the master before we start
* initializing the workers, we don't expect worker initialization to fail;
* so we just receive a quick OK/error code reply from each worker to be sure,
* and don't worry about an informative message.
*/
MPI_Bcast(&(cfg->abc->type), 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Reduce(&xstatus, &status, 1, MPI_INT, MPI_MAX, 0, MPI_COMM_WORLD);
if (status != eslOK) { MPI_Finalize(); p7_Fail("One or more MPI worker processes failed to initialize."); }
ESL_DPRINTF1(("%d workers are initialized\n", cfg->nproc-1));
/* Main loop: combining load workers, send/receive, clear workers loops;
* also, catch error states and die later, after clean shutdown of workers.
*
* When a recoverable error occurs, have_work = FALSE, xstatus !=
* eslOK, and errmsg is set to an informative message. No more
* errmsg's can be received after the first one. We wait for all the
* workers to clear their work units, then send them shutdown signals,
* then finally print our errmsg and exit.
*
* Unrecoverable errors just crash us out with p7_Fail().
*/
wi = 1;
while (have_work || nproc_working)
{
if (have_work)
{
if ((status = esl_msa_Read(cfg->afp, &msa)) == eslOK)
{
cfg->nali++;
ESL_DPRINTF1(("MPI master read MSA %s\n", msa->name == NULL? "" : msa->name));
}
else
{
have_work = FALSE;
if (status == eslEFORMAT) { xstatus = eslEFORMAT; snprintf(errmsg, eslERRBUFSIZE, "Alignment file parse error:\n%s\n", cfg->afp->errbuf); }
else if (status == eslEINVAL) { xstatus = eslEFORMAT; snprintf(errmsg, eslERRBUFSIZE, "Alignment file parse error:\n%s\n", cfg->afp->errbuf); }
else if (status != eslEOF) { xstatus = status; snprintf(errmsg, eslERRBUFSIZE, "Alignment file read unexpectedly failed with code %d\n", status); }
ESL_DPRINTF1(("MPI master has run out of MSAs (having read %d)\n", cfg->nali));
}
}
if ((have_work && nproc_working == cfg->nproc-1) || (!have_work && nproc_working > 0))
{
if (MPI_Probe(MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &mpistatus) != 0) { MPI_Finalize(); p7_Fail("mpi probe failed"); }
if (MPI_Get_count(&mpistatus, MPI_PACKED, &n) != 0) { MPI_Finalize(); p7_Fail("mpi get count failed"); }
wi = mpistatus.MPI_SOURCE;
ESL_DPRINTF1(("MPI master sees a result of %d bytes from worker %d\n", n, wi));
if (n > bn) {
if ((buf = realloc(buf, sizeof(char) * n)) == NULL) p7_Fail("reallocation failed");
bn = n;
}
if (MPI_Recv(buf, bn, MPI_PACKED, wi, 0, MPI_COMM_WORLD, &mpistatus) != 0) { MPI_Finalize(); p7_Fail("mpi recv failed"); }
ESL_DPRINTF1(("MPI master has received the buffer\n"));
/* If we're in a recoverable error state, we're only clearing worker results;
* just receive them, don't unpack them or print them.
* But if our xstatus is OK, go ahead and process the result buffer.
*/
if (xstatus == eslOK)
{
pos = 0;
if (MPI_Unpack(buf, bn, &pos, &xstatus, 1, MPI_INT, MPI_COMM_WORLD) != 0) { MPI_Finalize(); p7_Fail("mpi unpack failed");}
if (xstatus == eslOK) /* worker reported success. Get the HMM. */
{
ESL_DPRINTF1(("MPI master sees that the result buffer contains an HMM\n"));
if (p7_hmm_MPIUnpack(buf, bn, &pos, MPI_COMM_WORLD, &(cfg->abc), &hmm) != eslOK) { MPI_Finalize(); p7_Fail("HMM unpack failed"); }
ESL_DPRINTF1(("MPI master has unpacked the HMM\n"));
if (cfg->postmsafile != NULL) {
if (esl_msa_MPIUnpack(cfg->abc, buf, bn, &pos, MPI_COMM_WORLD, &postmsa) != eslOK) { MPI_Finalize(); p7_Fail("postmsa unpack failed");}
}
entropy = p7_MeanMatchRelativeEntropy(hmm, bg);
if ((status = output_result(cfg, errmsg, msaidx[wi], msalist[wi], hmm, postmsa, entropy)) != eslOK) xstatus = status;
esl_msa_Destroy(postmsa); postmsa = NULL;
p7_hmm_Destroy(hmm); hmm = NULL;
}
else /* worker reported an error. Get the errmsg. */
{
if (MPI_Unpack(buf, bn, &pos, errmsg, eslERRBUFSIZE, MPI_CHAR, MPI_COMM_WORLD) != 0) { MPI_Finalize(); p7_Fail("mpi unpack of errmsg failed"); }
ESL_DPRINTF1(("MPI master sees that the result buffer contains an error message\n"));
}
}
esl_msa_Destroy(msalist[wi]);
msalist[wi] = NULL;
msaidx[wi] = 0;
nproc_working--;
}
if (have_work)
{
ESL_DPRINTF1(("MPI master is sending MSA %s to worker %d\n", msa->name == NULL ? "":msa->name, wi));
if (esl_msa_MPISend(msa, wi, 0, MPI_COMM_WORLD, &buf, &bn) != eslOK) p7_Fail("MPI msa send failed");
msalist[wi] = msa;
msaidx[wi] = cfg->nali; /* 1..N for N alignments in the MSA database */
msa = NULL;
wi++;
nproc_working++;
}
}
/* On success or recoverable errors:
* Shut down workers cleanly.
*/
ESL_DPRINTF1(("MPI master is done. Shutting down all the workers cleanly\n"));
for (wi = 1; wi < cfg->nproc; wi++)
if (esl_msa_MPISend(NULL, wi, 0, MPI_COMM_WORLD, &buf, &bn) != eslOK) p7_Fail("MPI msa send failed");
free(buf);
free(msaidx);
free(msalist);
p7_bg_Destroy(bg);
if (xstatus != eslOK) { MPI_Finalize(); p7_Fail(errmsg); }
else return;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL;
char *hmmfile = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
int nhmm;
double x;
float KL;
int status;
char errbuf[eslERRBUFSIZE];
float nseq;
int do_eval2score = 0;
int do_score2eval = 0;
int z_val;
float e_val;
float s_val;
/* Process the command line options.
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=docgroup, 2 = indentation; 80=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL)
{
puts("Failed to read <hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
output_header(stdout, go);
if ( esl_opt_IsOn(go, "--eval2score") ) {
do_eval2score = TRUE;
e_val = esl_opt_GetReal(go, "-E");
} else if ( esl_opt_IsOn(go, "--score2eval") ) {
do_score2eval = TRUE;
s_val = esl_opt_GetReal(go, "-S");
} else if ( esl_opt_IsUsed(go, "--baseZ") || esl_opt_IsUsed(go, "--baseZ1") || esl_opt_IsUsed(go, "-Z") ) {
puts("The flags -Z, --baseZ, and --baseZ1 are for use with --eval2score and --score2eval.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_IsUsed(go, "--baseZ") ) {
z_val = 1000000 * 2 * (long)(esl_opt_GetInteger(go, "--baseZ"));
} else if (esl_opt_IsUsed(go, "--baseZ1") ) {
z_val = 1000000 * (long)(esl_opt_GetInteger(go, "--baseZ1"));
} else {
z_val = esl_opt_GetInteger(go, "-Z");
}
/* Initializations: open the HMM file
*/
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
/* Main body: read HMMs one at a time, print one line of stats
*/
printf("#\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s", "idx", "name", "accession", "nseq", "eff_nseq", "M", "relent", "info", "p relE", "compKL");
if (do_eval2score)
printf (" %6s %6.2g", "sc for", e_val);
if (do_score2eval)
printf (" %6s %6.2f", "E-val for", s_val);
printf("\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s", "----", "--------------------", "------------", "--------", "--------", "------", "------", "------", "------", "------");
if (do_eval2score)
printf (" %13s", "-------------");
if (do_score2eval)
printf (" %13s", "-------------");
printf("\n");
nhmm = 0;
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
nhmm++;
if ( esl_opt_IsOn(go, "--eval2score") || esl_opt_IsOn(go, "--score2eval") ) {
if (esl_opt_IsUsed(go, "--baseZ") || esl_opt_IsUsed(go, "--baseZ1" ) ) {
if ( hmm->abc->type != eslRNA && hmm->abc->type != eslDNA) {
puts("The flags --baseZ and --baseZ1 can't be used with non-nucleotide models.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
} else if ( hmm->abc->type != eslAMINO && hmm->abc->type != eslRNA && hmm->abc->type != eslDNA) {
puts("The flags --eval2score and --score2eval can't be used with non-sequence models.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
}
if (esl_opt_IsUsed(go, "--baseZ") ) {
nseq = (float)z_val / (float)(hmm->max_length);
} else if (esl_opt_IsUsed(go, "--baseZ1") ) {
nseq = (float)z_val / (float)(hmm->max_length);
} else {
nseq = z_val;
}
if (bg == NULL) bg = p7_bg_Create(abc);
p7_MeanPositionRelativeEntropy(hmm, bg, &x);
p7_hmm_CompositionKLDist(hmm, bg, &KL, NULL);
printf("%-6d %-20s %-12s %8d %8.2f %6d %6.2f %6.2f %6.2f %6.2f",
nhmm,
hmm->name,
hmm->acc == NULL ? "-" : hmm->acc,
hmm->nseq,
hmm->eff_nseq,
hmm->M,
p7_MeanMatchRelativeEntropy(hmm, bg),
p7_MeanMatchInfo(hmm, bg),
x,
KL);
if ( do_eval2score )
{
float sc;
sc = esl_exp_invsurv( e_val / nseq , hmm->evparam[p7_FTAU], hmm->evparam[p7_FLAMBDA]);
printf("%13.2f", sc);
}
else if ( do_score2eval)
{
float e;
e = nseq * esl_exp_surv( s_val , hmm->evparam[p7_FTAU], hmm->evparam[p7_FLAMBDA]);
printf("%13.2g", e);
}
printf("\n");
/* p7_MeanForwardScore(hmm, bg)); */
p7_hmm_Destroy(hmm);
}
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
p7_hmmfile_Close(hfp);
esl_getopts_Destroy(go);
exit(0);
}
/**
* int main(int argc, char **argv)
* Main driver
*/
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL;
char *hmmfile = NULL;
char *outhmmfile = NULL;
P7_HMMFILE *hfp = NULL;
FILE *outhmmfp; /* HMM output file handle */
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
int nhmm;
double x;
float KL;
int status;
char errbuf[eslERRBUFSIZE];
float average_internal_transitions[ p7H_NTRANSITIONS ];
int k;
char errmsg[eslERRBUFSIZE];
/* Process the command line options.
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
profillic_p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=docgroup, 2 = indentation; 80=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 2)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL)
{
puts("Failed to read <input hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((outhmmfile = esl_opt_GetArg(go, 2)) == NULL)
{
puts("Failed to read <output hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
profillic_p7_banner(stdout, argv[0], banner);
/* Initializations: open the input HMM file for reading
*/
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
/* Initializations: open the output HMM file for writing
*/
if ((outhmmfp = fopen(outhmmfile, "w")) == NULL) ESL_FAIL(status, errmsg, "Failed to open HMM file %s for writing", outhmmfile);
/* Main body: read HMMs one at a time, print one line of stats
*/
printf("#\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "idx", "name", "accession", "nseq", "eff_nseq", "M", "relent", "info", "p relE", "compKL");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "----", "--------------------", "------------", "--------", "--------", "------", "------", "------", "------", "------");
nhmm = 0;
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
nhmm++;
if (bg == NULL) bg = p7_bg_Create(abc);
esl_vec_FSet(average_internal_transitions, p7H_NTRANSITIONS, 0.);
for( k = 1; k < hmm->M; k++ ) {
esl_vec_FAdd(average_internal_transitions, hmm->t[k], p7H_NTRANSITIONS);
}
// Match transitions
esl_vec_FNorm(average_internal_transitions, 3);
// Insert transitions
esl_vec_FNorm(average_internal_transitions + 3, 2);
// Delete transitions
esl_vec_FNorm(average_internal_transitions + 5, 2);
// Ok now set them.
for( k = 1; k < hmm->M; k++ ) {
esl_vec_FCopy( average_internal_transitions, p7H_NTRANSITIONS, hmm->t[k] );
}
if ((status = p7_hmm_Validate(hmm, errmsg, 0.0001)) != eslOK) return status;
if ((status = p7_hmmfile_WriteASCII(outhmmfp, -1, hmm)) != eslOK) ESL_FAIL(status, errmsg, "HMM save failed");
p7_MeanPositionRelativeEntropy(hmm, bg, &x);
p7_hmm_CompositionKLDist(hmm, bg, &KL, NULL);
printf("%-6d %-20s %-12s %8d %8.2f %6d %6.2f %6.2f %6.2f %6.2f\n",
nhmm,
hmm->name,
hmm->acc == NULL ? "-" : hmm->acc,
hmm->nseq,
hmm->eff_nseq,
hmm->M,
p7_MeanMatchRelativeEntropy(hmm, bg),
p7_MeanMatchInfo(hmm, bg),
x,
KL);
/* p7_MeanForwardScore(hmm, bg)); */
p7_hmm_Destroy(hmm);
}
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
p7_hmmfile_Close(hfp);
if (outhmmfp != NULL) fclose(outhmmfp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
int status;
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL;
char *hmmfile = NULL;
char *outhmmfile = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
int nhmm;
double x;
float KL;
char errmsg[eslERRBUFSIZE];
/* Process the command line options.
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
profillic_p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nwhere options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=docgroup, 2 = indentation; 80=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 2)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL)
{
puts("Failed to read <input hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((outhmmfile = esl_opt_GetArg(go, 2)) == NULL)
{
puts("Failed to read <output hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
profillic_p7_banner(stdout, argv[0], banner);
/* Initializations: open the input HMM file for reading
*/
status = p7_hmmfile_Open(hmmfile, NULL, &hfp);
if (status == eslENOTFOUND) p7_Fail("Failed to open HMM file %s for reading.\n", hmmfile);
else if (status == eslEFORMAT) p7_Fail("File %s does not appear to be in a recognized HMM format.\n", hmmfile);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n", status, hmmfile);
/* Main body: read HMMs one at a time, print one line of stats
*/
printf("#\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "idx", "name", "accession", "nseq", "eff_nseq", "M", "relent", "info", "p relE", "compKL");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "----", "--------------------", "------------", "--------", "--------", "------", "------", "------", "------", "------");
nhmm = 0;
if ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
nhmm++;
if (bg == NULL) bg = p7_bg_Create(abc);
if( abc->type == eslDNA ) {
galosh::ProfileTreeRoot<seqan::Dna, floatrealspace> profile;
if( (status = convert_to_galosh_profile( hmm, profile )) != eslOK ) esl_fatal("Unexpected error in converting HMM from file %s to a dna galosh profile", hmmfile);
std::ofstream fs ( outhmmfile );
if( !fs.is_open() ) {
esl_fatal("Unexpected error in opening the file %s for writing", outhmmfile);
} else {
fs << profile;
fs.close();
}
} else if( abc->type == eslAMINO ) {
galosh::ProfileTreeRoot<seqan::AminoAcid20, floatrealspace> profile;
if( (status = convert_to_galosh_profile( hmm, profile )) != eslOK ) esl_fatal("Unexpected error in converting HMM from file %s to an amino galosh profile", hmmfile);
std::ofstream fs ( outhmmfile );
if( !fs.is_open() ) {
esl_fatal("Unexpected error in opening the file %s for writing", outhmmfile);
} else {
fs << profile;
fs.close();
}
} else {
ESL_EXCEPTION(eslEUNIMPLEMENTED, "Sorry, at present the profillic-hmmtoprofile software can only handle amino and dna.");
}
p7_MeanPositionRelativeEntropy(hmm, bg, &x);
p7_hmm_CompositionKLDist(hmm, bg, &KL, NULL);
printf("%-6d %-20s %-12s %8d %8.2f %6d %6.2f %6.2f %6.2f %6.2f\n",
nhmm,
hmm->name,
hmm->acc == NULL ? "-" : hmm->acc,
hmm->nseq,
hmm->eff_nseq,
hmm->M,
p7_MeanMatchRelativeEntropy(hmm, bg),
p7_MeanMatchInfo(hmm, bg),
x,
KL);
/* p7_MeanForwardScore(hmm, bg)); */
p7_hmm_Destroy(hmm);
}
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
p7_hmmfile_Close(hfp);
esl_getopts_Destroy(go);
exit(0);
}
