/* Function: p7_CreateDefaultApp()
* Synopsis: Initialize a small/simple/standard HMMER application
* Incept: SRE, Thu Oct 28 15:03:21 2010 [Janelia]
*
* Purpose: Identical to <esl_getopts_CreateDefaultApp()>, but
* specialized for HMMER. See documentation in
* <easel/esl_getopts.c>.
*
* Args: options - array of <ESL_OPTIONS> structures for getopts
* nargs - number of cmd line arguments expected (excl. of cmdname)
* argc - <argc> from main()
* argv - <argv> from main()
* banner - optional one-line description of program (or NULL)
* usage - optional one-line usage hint (or NULL)
*
* Returns: ptr to new <ESL_GETOPTS> object.
*
* On command line errors, this routine prints an error
* message to <stderr> then calls <exit(1)> to halt
* execution with abnormal (1) status.
*
* If the standard <-h> option is seen, the routine prints
* the help page (using the data in the <options> structure),
* then calls <exit(0)> to exit with normal (0) status.
*
* Xref: J7/3
*
* Note: The only difference between this and esl_getopts_CreateDefaultApp()
* is to call p7_banner() instead of esl_banner(), to get HMMER
* versioning info into the header. There ought to be a better way
* (perhaps using PACKAGE_* define's instead of HMMER_* vs. EASEL_*
* define's in esl_banner(), thus removing the need for p7_banner).
*/
ESL_GETOPTS *
p7_CreateDefaultApp(ESL_OPTIONS *options, int nargs, int argc, char **argv, char *banner, char *usage)
{
ESL_GETOPTS *go = NULL;
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);
if (usage != NULL) 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)
{
if (banner != NULL) p7_banner(stdout, argv[0], banner);
if (usage != NULL) esl_usage (stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != nargs)
{
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);
}
return go;
}
static int
output_header(FILE *ofp, const ESL_GETOPTS *go)
{
p7_banner(ofp, go->argv[0], banner);
if (esl_opt_IsUsed(go, "--eval2score")) {
if ( fprintf(ofp, "# show score required to reach E-value: %.2g\n", esl_opt_GetReal(go, "-E")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (esl_opt_IsUsed(go, "--score2eval")) {
if ( fprintf(ofp, "# show E-value corresponding to score: %.2g\n", esl_opt_GetReal(go, "-S")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (esl_opt_IsUsed(go, "--eval2score") || esl_opt_IsUsed(go, "--score2eval")) {
if (esl_opt_IsUsed(go, "--baseZ") ) {
if ( fprintf(ofp, "# using base count (search both strands): %d Mb\n", esl_opt_GetInteger(go, "--baseZ")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
} else if (esl_opt_IsUsed(go, "--baseZ1") ) {
if ( fprintf(ofp, "# using base count (search single strand): %d Mb\n", esl_opt_GetInteger(go, "--baseZ1")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
} else {
if ( fprintf(ofp, "# using database sequence count: %d\n", esl_opt_GetInteger(go, "-Z")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
}
if (fprintf(ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
return eslOK;
}
static int
output_header(const ESL_GETOPTS *go, const struct cfg_s *cfg)
{
if (cfg->my_rank > 0) return eslOK;
/* if (! cfg->be_verbose) return eslOK; */
p7_banner(cfg->ofp, go->argv[0], banner);
fprintf(cfg->ofp, "# input alignment file: %s\n", cfg->alifile);
fprintf(cfg->ofp, "# output HMM file: %s\n", cfg->hmmfile);
if (esl_opt_IsUsed(go, "-n")) fprintf(cfg->ofp, "# name (the single) HMM: %s\n", esl_opt_GetString(go, "-n"));
if (esl_opt_IsUsed(go, "-o")) fprintf(cfg->ofp, "# output directed to file: %s\n", esl_opt_GetString(go, "-o"));
if (esl_opt_IsUsed(go, "-O")) fprintf(cfg->ofp, "# processed alignment resaved to: %s\n", esl_opt_GetString(go, "-O"));
if (esl_opt_IsUsed(go, "--amino")) fprintf(cfg->ofp, "# input alignment is asserted as: protein\n");
if (esl_opt_IsUsed(go, "--dna")) fprintf(cfg->ofp, "# input alignment is asserted as: DNA\n");
if (esl_opt_IsUsed(go, "--rna")) fprintf(cfg->ofp, "# input alignment is asserted as: RNA\n");
if (esl_opt_IsUsed(go, "--fast")) fprintf(cfg->ofp, "# model architecture construction: fast/heuristic\n");
if (esl_opt_IsUsed(go, "--hand")) fprintf(cfg->ofp, "# model architecture construction: hand-specified by RF annotation\n");
if (esl_opt_IsUsed(go, "--symfrac")) fprintf(cfg->ofp, "# sym fraction for model structure: %.3f\n", esl_opt_GetReal(go, "--symfrac"));
if (esl_opt_IsUsed(go, "--fragthresh"))fprintf(cfg->ofp, "# seq called fragment if < xL : %.3f\n", esl_opt_GetReal(go, "--fragthresh"));
if (esl_opt_IsUsed(go, "--wpb")) fprintf(cfg->ofp, "# relative weighting scheme: Henikoff PB\n");
if (esl_opt_IsUsed(go, "--wgsc")) fprintf(cfg->ofp, "# relative weighting scheme: G/S/C\n");
if (esl_opt_IsUsed(go, "--wblosum")) fprintf(cfg->ofp, "# relative weighting scheme: BLOSUM filter\n");
if (esl_opt_IsUsed(go, "--wnone")) fprintf(cfg->ofp, "# relative weighting scheme: none\n");
if (esl_opt_IsUsed(go, "--wid")) fprintf(cfg->ofp, "# frac id cutoff for BLOSUM wgts: %f\n", esl_opt_GetReal(go, "--wid"));
if (esl_opt_IsUsed(go, "--eent")) fprintf(cfg->ofp, "# effective seq number scheme: entropy weighting\n");
if (esl_opt_IsUsed(go, "--eclust")) fprintf(cfg->ofp, "# effective seq number scheme: single linkage clusters\n");
if (esl_opt_IsUsed(go, "--enone")) fprintf(cfg->ofp, "# effective seq number scheme: none\n");
if (esl_opt_IsUsed(go, "--eset")) fprintf(cfg->ofp, "# effective seq number: set to %f\n", esl_opt_GetReal(go, "--eset"));
if (esl_opt_IsUsed(go, "--ere") ) fprintf(cfg->ofp, "# minimum rel entropy target: %f bits\n", esl_opt_GetReal(go, "--ere"));
if (esl_opt_IsUsed(go, "--esigma") ) fprintf(cfg->ofp, "# entropy target sigma parameter: %f bits\n", esl_opt_GetReal(go, "--esigma"));
if (esl_opt_IsUsed(go, "--eid") ) fprintf(cfg->ofp, "# frac id cutoff for --eclust: %f\n", esl_opt_GetReal(go, "--eid"));
if (esl_opt_IsUsed(go, "--EmL") ) fprintf(cfg->ofp, "# seq length for MSV Gumbel mu fit: %d\n", esl_opt_GetInteger(go, "--EmL"));
if (esl_opt_IsUsed(go, "--EmN") ) fprintf(cfg->ofp, "# seq number for MSV Gumbel mu fit: %d\n", esl_opt_GetInteger(go, "--EmN"));
if (esl_opt_IsUsed(go, "--EvL") ) fprintf(cfg->ofp, "# seq length for Vit Gumbel mu fit: %d\n", esl_opt_GetInteger(go, "--EvL"));
if (esl_opt_IsUsed(go, "--EvN") ) fprintf(cfg->ofp, "# seq number for Vit Gumbel mu fit: %d\n", esl_opt_GetInteger(go, "--EvN"));
if (esl_opt_IsUsed(go, "--EfL") ) fprintf(cfg->ofp, "# seq length for Fwd exp tau fit: %d\n", esl_opt_GetInteger(go, "--EfL"));
if (esl_opt_IsUsed(go, "--EfN") ) fprintf(cfg->ofp, "# seq number for Fwd exp tau fit: %d\n", esl_opt_GetInteger(go, "--EfN"));
if (esl_opt_IsUsed(go, "--Eft") ) fprintf(cfg->ofp, "# tail mass for Fwd exp tau fit: %f\n", esl_opt_GetReal(go, "--Eft"));
#ifdef HMMER_THREADS
if (esl_opt_IsUsed(go, "--cpu")) fprintf(cfg->ofp, "# number of worker threads: %d\n", esl_opt_GetInteger(go, "--cpu"));
#endif
#ifdef HAVE_MPI
if (esl_opt_IsUsed(go, "--mpi") ) fprintf(cfg->ofp, "# parallelization mode: MPI\n");
#endif
if (esl_opt_IsUsed(go, "--seed")) {
if (esl_opt_GetInteger(go, "--seed") == 0) fprintf(cfg->ofp,"# random number seed: one-time arbitrary\n");
else fprintf(cfg->ofp,"# random number seed set to: %d\n", esl_opt_GetInteger(go, "--seed"));
}
if (esl_opt_IsUsed(go, "--laplace") ) fprintf(cfg->ofp, "# prior: Laplace +1\n");
fprintf(cfg->ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n");
return eslOK;
}
static int
output_header(FILE *ofp, ESL_GETOPTS *go, char *hmmfile, char *seqfile)
{
p7_banner(ofp, go->argv[0], banner);
if (fprintf(ofp, "# query sequence file: %s\n", seqfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# target HMM database: %s\n", hmmfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-o") && fprintf(ofp, "# output directed to file: %s\n", esl_opt_GetString(go, "-o")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--tblout") && fprintf(ofp, "# per-seq hits tabular output: %s\n", esl_opt_GetString(go, "--tblout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--dfamtblout") && fprintf(ofp, "# hits output in Dfam format: %s\n", esl_opt_GetString(go, "--dfamtblout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--aliscoresout") && fprintf(ofp, "# alignment scores output: %s\n", esl_opt_GetString(go, "--aliscoresout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--acc") && fprintf(ofp, "# prefer accessions over names: yes\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--noali") && fprintf(ofp, "# show alignments in output: no\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--notextw") && fprintf(ofp, "# max ASCII text line length: unlimited\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--textw") && fprintf(ofp, "# max ASCII text line length: %d\n", esl_opt_GetInteger(go, "--textw")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-E") && fprintf(ofp, "# profile reporting threshold: E-value <= %g\n", esl_opt_GetReal(go, "-E")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-T") && fprintf(ofp, "# profile reporting threshold: score >= %g\n", esl_opt_GetReal(go, "-T")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incE") && fprintf(ofp, "# profile inclusion threshold: E-value <= %g\n", esl_opt_GetReal(go, "--incE")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incT") && fprintf(ofp, "# profile inclusion threshold: score >= %g\n", esl_opt_GetReal(go, "--incT")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_ga") && fprintf(ofp, "# model-specific thresholding: GA cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_nc") && fprintf(ofp, "# model-specific thresholding: NC cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_tc") && fprintf(ofp, "# model-specific thresholding: TC cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--max") && fprintf(ofp, "# Max sensitivity mode: on [all heuristic filters off]\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F1") && fprintf(ofp, "# MSV filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F1")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F2") && fprintf(ofp, "# Vit filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F2")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F3") && fprintf(ofp, "# Fwd filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F3")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--nobias") && fprintf(ofp, "# biased composition HMM filter: off\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--B1") && fprintf(ofp, "# biased comp MSV window len: %d\n", esl_opt_GetInteger(go, "--B1")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--B2") && fprintf(ofp, "# biased comp Viterbi window len: %d\n", esl_opt_GetInteger(go, "--B2")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--B3") && fprintf(ofp, "# biased comp Forward window len: %d\n", esl_opt_GetInteger(go, "--B3")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--nonull2") && fprintf(ofp, "# null2 bias corrections: off\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--toponly") && fprintf(ofp, "# search only top strand: on\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--bottomonly") && fprintf(ofp, "# search only bottom strand: on\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-Z") && fprintf(ofp, "# sequence search space set to: %.0f\n", esl_opt_GetReal(go, "-Z")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--seed")) {
if (esl_opt_GetInteger(go, "--seed")==0 && fprintf(ofp, "# random number seed: one-time arbitrary\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
else if ( fprintf(ofp, "# random number seed set to: %d\n", esl_opt_GetInteger(go, "--seed")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (esl_opt_IsUsed(go, "--qformat") && fprintf(ofp, "# input seqfile format asserted: %s\n", esl_opt_GetString(go, "--qformat")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--w_beta") && fprintf(ofp, "# window length beta value: %g\n", esl_opt_GetReal(go, "--w_beta")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--w_length") && fprintf(ofp, "# window length : %d\n", esl_opt_GetInteger(go, "--w_length")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
// if (esl_opt_IsUsed(go, "--daemon") && fprintf(ofp, "run as a daemon process\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
#ifdef HMMER_THREADS
if (esl_opt_IsUsed(go, "--cpu") && fprintf(ofp, "# number of worker threads: %d\n", esl_opt_GetInteger(go, "--cpu")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
#endif
if (fprintf(ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
return eslOK;
}
static void
cmdline_help(char *argv0, ESL_GETOPTS *go)
{
p7_banner (stdout, argv0, banner);
esl_usage (stdout, argv0, usage);
puts("\nBasic options:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
puts("\nLess common options:");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nSequence input formats include: FASTA, EMBL, GenBank, UniProt");
puts("Alignment output formats include: Stockholm, Pfam, A2M, PSIBLAST\n");
exit(0);
}
/* Function: output_header()
* Synopsis: Print details of FM-index construction
*/
static int
output_header(FILE *ofp, const ESL_GETOPTS *go, char *seqfile, char *fmfile)
{
p7_banner(ofp, go->argv[0], banner);
if (fprintf(ofp, "# input sequence file: %s\n", seqfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# output binary-formatted HMMER database: %s\n", fmfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# alphabet : %s\n", esl_opt_GetString(go, "--alph")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# bin_length : %d\n", esl_opt_GetInteger(go, "--bin_length")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# suffix array sample rate: %d\n", esl_opt_GetInteger(go, "--sa_freq")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
return eslOK;
}
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_seqfile, char **ret_fmfile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 120=textwidth*/
if (puts("\nSpecial options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80); /* 2= group; 2 = indentation; 120=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_seqfile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <seqfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_fmfile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <fmfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* Validate any attempted use of stdin streams */
if (esl_strcmp(*ret_seqfile, "-") == 0 && esl_strcmp(*ret_fmfile, "-") == 0)
{ if (puts("Either <seqfile> or <fmfile> may be '-' (to read from stdin), but not both.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
static int
output_header(FILE *ofp, const ESL_GETOPTS *go, char *hmmfile, char *seqfile)
{
p7_banner(ofp, go->argv[0], banner);
if (fprintf(ofp, "# query HMM file: %s\n", hmmfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# target sequence database: %s\n", seqfile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-o") && fprintf(ofp, "# output directed to file: %s\n", esl_opt_GetString(go, "-o")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-A") && fprintf(ofp, "# MSA of all hits saved to file: %s\n", esl_opt_GetString(go, "-A")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--tblout") && fprintf(ofp, "# per-seq hits tabular output: %s\n", esl_opt_GetString(go, "--tblout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--domtblout") && fprintf(ofp, "# per-dom hits tabular output: %s\n", esl_opt_GetString(go, "--domtblout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--pfamtblout") && fprintf(ofp, "# pfam-style tabular hit output: %s\n", esl_opt_GetString(go, "--pfamtblout")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--acc") && fprintf(ofp, "# prefer accessions over names: yes\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--noali") && fprintf(ofp, "# show alignments in output: no\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--notextw") && fprintf(ofp, "# max ASCII text line length: unlimited\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--textw") && fprintf(ofp, "# max ASCII text line length: %d\n", esl_opt_GetInteger(go, "--textw")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-E") && fprintf(ofp, "# sequence reporting threshold: E-value <= %g\n", esl_opt_GetReal(go, "-E")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-T") && fprintf(ofp, "# sequence reporting threshold: score >= %g\n", esl_opt_GetReal(go, "-T")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--domE") && fprintf(ofp, "# domain reporting threshold: E-value <= %g\n", esl_opt_GetReal(go, "--domE")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--domT") && fprintf(ofp, "# domain reporting threshold: score >= %g\n", esl_opt_GetReal(go, "--domT")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incE") && fprintf(ofp, "# sequence inclusion threshold: E-value <= %g\n", esl_opt_GetReal(go, "--incE")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incT") && fprintf(ofp, "# sequence inclusion threshold: score >= %g\n", esl_opt_GetReal(go, "--incT")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incdomE") && fprintf(ofp, "# domain inclusion threshold: E-value <= %g\n", esl_opt_GetReal(go, "--incdomE")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--incdomT") && fprintf(ofp, "# domain inclusion threshold: score >= %g\n", esl_opt_GetReal(go, "--incdomT")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_ga") && fprintf(ofp, "# model-specific thresholding: GA cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_nc") && fprintf(ofp, "# model-specific thresholding: NC cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--cut_tc") && fprintf(ofp, "# model-specific thresholding: TC cutoffs\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--max") && fprintf(ofp, "# Max sensitivity mode: on [all heuristic filters off]\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F1") && fprintf(ofp, "# MSV filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F1")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F2") && fprintf(ofp, "# Vit filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F2")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--F3") && fprintf(ofp, "# Fwd filter P threshold: <= %g\n", esl_opt_GetReal(go, "--F3")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--nobias") && fprintf(ofp, "# biased composition HMM filter: off\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--restrictdb_stkey") && fprintf(ofp, "# Restrict db to start at seq key: %s\n", esl_opt_GetString(go, "--restrictdb_stkey")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--restrictdb_n") && fprintf(ofp, "# Restrict db to # target seqs: %d\n", esl_opt_GetInteger(go, "--restrictdb_n")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--ssifile") && fprintf(ofp, "# Override ssi file to: %s\n", esl_opt_GetString(go, "--ssifile")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--nonull2") && fprintf(ofp, "# null2 bias corrections: off\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-Z") && fprintf(ofp, "# sequence search space set to: %.0f\n", esl_opt_GetReal(go, "-Z")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--domZ") && fprintf(ofp, "# domain search space set to: %.0f\n", esl_opt_GetReal(go, "--domZ")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--seed")) {
if (esl_opt_GetInteger(go, "--seed") == 0 && fprintf(ofp, "# random number seed: one-time arbitrary\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
else if ( fprintf(ofp, "# random number seed set to: %d\n", esl_opt_GetInteger(go, "--seed")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (esl_opt_IsUsed(go, "--tformat") && fprintf(ofp, "# targ <seqfile> format asserted: %s\n", esl_opt_GetString(go, "--tformat")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (fprintf(ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
return eslOK;
}
static void
cmdline_help(char *argv0, ESL_GETOPTS *go)
{
p7_banner (stdout, argv0, banner);
esl_usage (stdout, argv0, usage);
puts("\nCommon options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
puts("\nOptions controlling what to emit:");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nOptions controlling emission from profiles with -p:");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
puts("\nOptions controlling fancy consensus emission with -C:");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
puts("\nOther options::");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
exit(0);
}
static void
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_hmmfile, char **ret_alifile)
{
ESL_GETOPTS *go = NULL;
if ((go = esl_getopts_Create(options)) == NULL) p7_Die("problem with options structure");
if (esl_opt_ProcessEnvironment(go) != eslOK) { printf("Failed to process environment: %s\n", go->errbuf); goto ERROR; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { printf("Failed to parse command line: %s\n", go->errbuf); goto ERROR; }
if (esl_opt_VerifyConfig(go) != eslOK) { printf("Failed to parse command line: %s\n", go->errbuf); goto ERROR; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
puts("\nOptions for selecting alphabet rather than guessing it:");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nAlternative model construction strategies:");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
puts("\nAlternative relative sequence weighting strategies:");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
puts("\nAlternate effective sequence weighting strategies:");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
puts("\nControl of E-value calibration:");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
puts("\nOther options:");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { puts("Incorrect number of command line arguments."); goto ERROR; }
if ((*ret_hmmfile = esl_opt_GetArg(go, 1)) == NULL) { puts("Failed to get <hmmfile> argument on command line"); goto ERROR; }
if ((*ret_alifile = esl_opt_GetArg(go, 2)) == NULL) { puts("Failed to get <alifile> argument on command line"); goto ERROR; }
*ret_go = go;
return;
ERROR: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
printf("\nTo see more help on other available options, do %s -h\n\n", argv[0]);
exit(1);
}
static int
output_header(const ESL_GETOPTS *go, FILE *ofp, char *alifile, char *postmsafile)
{
p7_banner(ofp, go->argv[0], banner);
if (fprintf(ofp, "# input alignment file: %s\n", alifile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
if (fprintf(ofp, "# output alignment file: %s\n", postmsafile) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (esl_opt_IsUsed(go, "--alirange") && fprintf(ofp, "# alignment range: %s\n", esl_opt_GetString(go, "--alirange")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--modelrange") && fprintf(ofp, "# model range: %s\n", esl_opt_GetString(go, "--modelrange"))< 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--apendmask") && fprintf(ofp, "# add to existing mask: [on]\n" )< 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--model2ali") && fprintf(ofp, "# ali ranges for model range: %s\n", esl_opt_GetString(go, "--model2ali")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--ali2model") && fprintf(ofp, "# model ranges for ali range: %s\n", esl_opt_GetString(go, "--ali2model")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "-o") && fprintf(ofp, "# output directed to file: %s\n", esl_opt_GetString(go, "-o")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--amino") && fprintf(ofp, "# input alignment is asserted as: protein\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--dna") && fprintf(ofp, "# input alignment is asserted as: DNA\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--rna") && fprintf(ofp, "# input alignment is asserted as: RNA\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--fast") && fprintf(ofp, "# model architecture construction: fast/heuristic\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--hand") && fprintf(ofp, "# model architecture construction: hand-specified by RF annotation\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--symfrac") && fprintf(ofp, "# sym fraction for model structure: %.3f\n", esl_opt_GetReal(go, "--symfrac")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--fragthresh") && fprintf(ofp, "# seq called frag if L <= x*alen: %.3f\n", esl_opt_GetReal(go, "--fragthresh")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--wpb") && fprintf(ofp, "# relative weighting scheme: Henikoff PB\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--wgsc") && fprintf(ofp, "# relative weighting scheme: G/S/C\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--wblosum") && fprintf(ofp, "# relative weighting scheme: BLOSUM filter\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--wnone") && fprintf(ofp, "# relative weighting scheme: none\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--wid") && fprintf(ofp, "# frac id cutoff for BLOSUM wgts: %f\n", esl_opt_GetReal(go, "--wid")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
if (esl_opt_IsUsed(go, "--seed")) {
if (esl_opt_GetInteger(go, "--seed") == 0 && fprintf(ofp,"# random number seed: one-time arbitrary\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
else if ( fprintf(ofp,"# random number seed set to: %d\n", esl_opt_GetInteger(go, "--seed")) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
}
if (fprintf(ofp, "# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed");
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
int i, j;
int status = eslOK;
P7_HMMFILE *hfp = NULL; /* open input HMM file */
P7_HMM *hmm = NULL; /* one HMM query */
ESL_ALPHABET *abc = NULL; /* digital alphabet */
P7_BG *bg = NULL;
char errbuf[eslERRBUFSIZE];
char* hmmfile;
float *rel_ents = NULL;
float **heights = NULL;
float **probs = NULL;
float *ins_P = NULL;
float *ins_expL = NULL;
float *occupancy = NULL;
int mode = HMMLOGO_RELENT_ALL; //default
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) {
p7_banner (stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 100);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1) esl_fatal(argv[0], "Incorrect number of command line arguments.\n");
hmmfile = esl_opt_GetArg(go, 1);
if (esl_opt_IsOn(go, "--height_relent_all"))
mode = HMMLOGO_RELENT_ALL;
else if (esl_opt_IsOn(go, "--height_relent_abovebg"))
mode = HMMLOGO_RELENT_ABOVEBG;
else if (esl_opt_IsOn(go, "--height_score"))
mode = HMMLOGO_SCORE;
else
mode = HMMLOGO_RELENT_ALL; //default
/* Open the query profile 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);
status = p7_hmmfile_Read(hfp, &abc, &hmm);
bg = p7_bg_Create(abc);
ESL_ALLOC(rel_ents, (hmm->M+1) * sizeof(float));
ESL_ALLOC(heights, (hmm->M+1) * sizeof(float*));
ESL_ALLOC(probs, (hmm->M+1) * sizeof(float*));
for (i = 1; i <= hmm->M; i++) {
ESL_ALLOC(heights[i], abc->K * sizeof(float));
ESL_ALLOC(probs[i], abc->K * sizeof(float));
}
/* residue heights */
if (mode == HMMLOGO_RELENT_ALL) {
printf ("max expected height = %.2f\n", hmmlogo_maxHeight(bg) );
hmmlogo_RelativeEntropy_all(hmm, bg, rel_ents, probs, heights);
} else if (mode == HMMLOGO_RELENT_ABOVEBG) {
printf ("max expected height = %.2f\n", hmmlogo_maxHeight(bg) );
hmmlogo_RelativeEntropy_above_bg(hmm, bg, rel_ents, probs, heights);
} else if (mode == HMMLOGO_SCORE) {
hmmlogo_ScoreHeights(hmm, bg, heights );
}
printf ("Residue heights\n");
for (i = 1; i <= hmm->M; i++) {
printf("%d: ", i);
for (j=0; j<abc->K; j++)
printf("%6.3f ", heights[i][j] );
if (mode != HMMLOGO_SCORE)
printf(" (%6.3f)", rel_ents[i]);
printf("\n");
}
if (rel_ents != NULL) free(rel_ents);
if (heights != NULL) {
for (i = 1; i <= hmm->M; i++)
if (heights[i] != NULL) free(heights[i]);
free(heights);
}
/* indel values */
if (! esl_opt_IsOn(go, "--no_indel")) {
ESL_ALLOC(ins_P, (hmm->M+1) * sizeof(float));
ESL_ALLOC(ins_expL, (hmm->M+1) * sizeof(float));
ESL_ALLOC(occupancy, (hmm->M+1) * sizeof(float));
hmmlogo_IndelValues(hmm, ins_P, ins_expL, occupancy);
printf ("Indel values\n");
for (i = 1; i <= hmm->M; i++)
printf("%d: %6.3f %6.3f %6.3f\n", i, ins_P[i], ins_expL[i], occupancy[i] );
free(ins_P);
free(ins_expL);
free(occupancy);
}
p7_hmmfile_Close(hfp);
esl_alphabet_Destroy(abc);
p7_bg_Destroy(bg);
exit(0);
ERROR:
if (rel_ents != NULL) free(rel_ents);
if (heights != NULL) {
for (i = 1; i <= hmm->M; i++)
if (heights[i] != NULL) free(heights[i]);
free(heights);
}
if (hfp != NULL) p7_hmmfile_Close(hfp);
if (abc != NULL) esl_alphabet_Destroy(abc);
if (ins_P != NULL) free(ins_P);
if (ins_expL != NULL) free(ins_expL);
if (occupancy != NULL) free(occupancy);
}
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);
}
/* process_commandline()
*
* Processes the commandline, filling in fields in <cfg> and creating and returning
* an <ESL_GETOPTS> options structure. The help page (hmmsearch -h) is formatted
* here.
*/
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_hmmfile, char **ret_seqfile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
if (puts("\nOptions controlling output:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
if (puts("\nOptions controlling reporting thresholds:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
if (puts("\nOptions controlling inclusion (significance) thresholds:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
if (puts("\nOptions for model-specific thresholding:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
if (puts("\nOptions controlling acceleration heuristics:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 7, 2, 80);
if (puts("\nOther expert options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 12, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_hmmfile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <hmmdb> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_seqfile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <seqfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* Validate any attempted use of stdin streams */
if (strcmp(*ret_hmmfile, "-") == 0)
{ if (puts("nhmmscan cannot read <hmm database> from stdin stream, because it must have hmmpress'ed auxfiles") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
if (puts("\nwhere most common options are:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
if (printf("\nTo see more help on available options, do %s -h\n\n", argv[0]) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_alifile, char **ret_postalifile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
if (puts("\nMask range options (format: --xxx 10-20,30-40 ) :") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
if (puts("\nOptions for selecting alphabet rather than guessing it:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
if (puts("\nAlternative model construction strategies:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
if (puts("\nAlternative relative sequence weighting strategies:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
if (puts("\nOther options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) > 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_alifile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <msafile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
if ((*ret_postalifile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <postmsafile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
}
if (strcmp(*ret_alifile, "-") == 0 && ! esl_opt_IsOn(go, "--informat"))
{ if (puts("Must specify --informat to read <alifile> from stdin ('-')") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
printf("\nTo see more help on other available options, do:\n %s -h\n\n", argv[0]);
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_STOPWATCH *w = esl_stopwatch_Create();
struct cfg_s cfg;
p7_Init();
/* Process 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("\nCommon options:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1=docgroup, 2 = indentation; 80=textwidth*/
puts("\nChoice of score type :");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nChoice of alignment mode :");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
puts("\nChoice of multi vs. unihit config :");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
puts("\nChoice of generic vs. vector DP implementation :");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
puts("\nOutput options (use only in serial mode, for single HMM input):");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
puts("\nControlling range of fitted tail masses :");
esl_opt_DisplayHelp(stdout, go, 7, 2, 80);
puts("\nRecalibrating E-values, and replacing HMM's existing parameters :");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
puts("\nDebugging :");
esl_opt_DisplayHelp(stdout, go, 9, 2, 80);
puts("\nExperiments :");
esl_opt_DisplayHelp(stdout, go, 10, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
puts("\nwhere general options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1=docgroup, 2 = indentation; 80=textwidth*/
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
/* Validate combinations of score/config/implementation (4x3x2x2, score x mode x hit x DPimpl: 20 of 48 possible combos valid */
if (esl_opt_GetBoolean(go, "--vector") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("SIMD vector implementations only work for local alignment."); /* -16/48 */
if (esl_opt_GetBoolean(go, "--msv") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("MSV scoring is local by definition: use --local."); /* -4/48 */
if (esl_opt_GetBoolean(go, "--vit") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("no p7_GViterbiDual for new dual-mode profile implemented yet"); /* -4/48 */
/* Initialize configuration shared across all kinds of masters
* and workers in this .c file.
*/
cfg.hmmfile = esl_opt_GetArg(go, 1);
cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
// cfg.abc = esl_alphabet_Create(eslAMINO);
cfg.my_rank = 0; /* MPI init will change this soon, if --mpi was set */
cfg.nproc = 0; /* MPI init will change this soon, if --mpi was set */
cfg.do_mpi = FALSE; /* --mpi will change this soon (below) if necessary */
cfg.do_stall = esl_opt_GetBoolean(go, "--stall");
cfg.N = esl_opt_GetInteger(go, "-N");
cfg.L = esl_opt_GetInteger(go, "-L");
cfg.hfp = NULL;
cfg.ofp = NULL;
cfg.survfp = NULL;
cfg.efp = NULL;
cfg.ffp = NULL;
cfg.xfp = NULL;
cfg.alfp = NULL;
cfg.bg = NULL;
/* This is our stall point, if we need to wait until we get a
* debugger attached to this process for debugging (especially
* useful for MPI):
*/
while (cfg.do_stall);
/* Start timing. */
esl_stopwatch_Start(w);
/* Main body:
* Handed off to serial version or MPI masters and workers as appropriate.
*/
#ifdef HAVE_MPI
if (esl_opt_GetBoolean(go, "--mpi"))
{
/* Initialize MPI, figure out who we are, and whether we're running
* this show (proc 0) or working in it (procs >0).
*/
cfg.do_mpi = TRUE;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &(cfg.my_rank));
MPI_Comm_size(MPI_COMM_WORLD, &(cfg.nproc));
if (cfg.my_rank == 0 && cfg.nproc < 2) p7_Fail("Need at least 2 MPI processes to run --mpi mode.");
if (cfg.my_rank > 0) mpi_worker(go, &cfg);
else mpi_master(go, &cfg);
esl_stopwatch_Stop(w);
esl_stopwatch_MPIReduce(w, 0, MPI_COMM_WORLD);
MPI_Finalize(); /* both workers and masters reach this line */
}
else
#endif /*HAVE_MPI*/
{
/* No MPI? Then we're just the serial master. */
serial_master(go, &cfg);
esl_stopwatch_Stop(w);
}
/* Stop timing. */
if (cfg.my_rank == 0) esl_stopwatch_Display(stdout, w, "# CPU time: ");
/* Clean up and exit. */
if (cfg.my_rank == 0) {
if (cfg.hfp != NULL) p7_hmmfile_Close(cfg.hfp);
if (esl_opt_IsOn(go, "-o")) fclose(cfg.ofp);
if (cfg.survfp != NULL) fclose(cfg.survfp);
if (cfg.efp != NULL) fclose(cfg.efp);
if (cfg.ffp != NULL) fclose(cfg.ffp);
if (cfg.xfp != NULL) fclose(cfg.xfp);
if (cfg.alfp != NULL) fclose(cfg.alfp);
}
p7_bg_Destroy(cfg.bg);
esl_alphabet_Destroy(cfg.abc);
esl_randomness_Destroy(cfg.r);
esl_getopts_Destroy(go);
esl_stopwatch_Destroy(w);
return eslOK;
}
