int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_RANDOMNESS *r = NULL;
int be_verbose;
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
esl_usage(stdout, argv[0], usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
exit(0);
}
if (esl_opt_ArgNumber(go) != 0) {
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
exit(1);
}
be_verbose = esl_opt_GetBoolean(go, "-v");
if (esl_opt_GetBoolean(go, "-r")) {
r = esl_randomness_CreateTimeseeded();
if (be_verbose) printf("seed = %ld\n", esl_randomness_GetSeed(r));
} else r = esl_randomness_Create(esl_opt_GetInteger(go, "-s"));
utest_LinearRegression(r, TRUE, be_verbose);
utest_LinearRegression(r, FALSE, be_verbose);
esl_getopts_Destroy(go);
esl_randomness_Destroy(r);
exit(0);
}
/* Function: esl_dst_XAverageId()
* Synopsis: Calculate avg identity for digital MSA
* Incept: SRE, Fri May 18 15:19:14 2007 [Janelia]
*
* Purpose: Calculates the average pairwise fractional identity in
* a digital multiple sequence alignment <ax>, consisting of <N>
* aligned digital sequences of identical length.
*
* If an exhaustive calculation would require more than
* <max_comparisons> pairwise comparisons, then instead of
* looking at all pairs, calculate the average over a
* stochastic sample of <max_comparisons> random pairs.
* This allows the routine to work efficiently even on very
* deep MSAs.
*
* Each fractional pairwise identity (range $[0..$ pid $..1]$
* is calculated using <esl_dsq_XPairId()>.
*
* Returns: <eslOK> on success, and <*ret_id> contains the average
* fractional identity.
*
* Throws: <eslEMEM> on allocation failure.
* <eslEINVAL> if any of the aligned sequence pairs aren't
* of the same length.
* In either case, <*ret_id> is set to 0.
*/
int
esl_dst_XAverageId(const ESL_ALPHABET *abc, ESL_DSQ **ax, int N, int max_comparisons, double *ret_id)
{
int status;
double id;
double sum;
int i,j,n;
if (N <= 1) { *ret_id = 1.; return eslOK; }
*ret_id = 0.;
/* Is N small enough that we can average over all pairwise comparisons?
watch out for numerical overflow in this: Pfam N's easily overflow when squared
*/
if (N <= max_comparisons &&
N <= sqrt(2. * max_comparisons) &&
(N * (N-1) / 2) <= max_comparisons)
{
for (i = 0; i < N; i++)
for (j = i+1; j < N; j++)
{
if ((status = esl_dst_XPairId(abc, ax[i], ax[j], &id, NULL, NULL)) != eslOK) return status;
sum += id;
}
sum /= (double) (N * (N-1) / 2);
}
/* If nseq is large, calculate average over a stochastic sample. */
else
{
ESL_RANDOMNESS *r = esl_randomness_CreateTimeseeded();
for (n = 0; n < max_comparisons; n++)
{
do { i = esl_rnd_Roll(r, N); j = esl_rnd_Roll(r, N); } while (j == i); /* make sure j != i */
if ((status = esl_dst_XPairId(abc, ax[i], ax[j], &id, NULL, NULL)) != eslOK) return status;
sum += id;
}
sum /= (double) max_comparisons;
esl_randomness_Destroy(r);
}
*ret_id = sum;
return eslOK;
}
/* Function: esl_dst_CAverageId()
* Synopsis: Calculate avg identity for multiple alignment
* Incept: SRE, Fri May 18 15:02:38 2007 [Janelia]
*
* Purpose: Calculates the average pairwise fractional identity in
* a multiple sequence alignment <as>, consisting of <N>
* aligned character sequences of identical length.
*
* If an exhaustive calculation would require more than
* <max_comparisons> pairwise comparisons, then instead of
* looking at all pairs, calculate the average over a
* stochastic sample of <max_comparisons> random pairs.
* This allows the routine to work efficiently even on very
* deep MSAs.
*
* Each fractional pairwise identity (range $[0..$ pid $..1]$
* is calculated using <esl_dsq_CPairId()>.
*
* Returns: <eslOK> on success, and <*ret_id> contains the average
* fractional identity.
*
* Throws: <eslEMEM> on allocation failure.
* <eslEINVAL> if any of the aligned sequence pairs aren't
* of the same length.
* In either case, <*ret_id> is set to 0.
*/
int
esl_dst_CAverageId(char **as, int N, int max_comparisons, double *ret_id)
{
int status;
double id;
double sum;
int i,j,n;
if (N <= 1) { *ret_id = 1.; return eslOK; }
*ret_id = 0.;
/* Is nseq small enough that we can average over all pairwise comparisons? */
if ((N * (N-1) / 2) <= max_comparisons)
{
for (i = 0; i < N; i++)
for (j = i+1; j < N; j++)
{
if ((status = esl_dst_CPairId(as[i], as[j], &id, NULL, NULL)) != eslOK) return status;
sum += id;
}
id /= (double) (N * (N-1) / 2);
}
/* If nseq is large, calculate average over a stochastic sample. */
else
{
ESL_RANDOMNESS *r = esl_randomness_CreateTimeseeded();
for (n = 0; n < max_comparisons; n++)
{
do { i = esl_rnd_Roll(r, N); j = esl_rnd_Roll(r, N); } while (j == i); /* make sure j != i */
if ((status = esl_dst_CPairId(as[i], as[j], &id, NULL, NULL)) != eslOK) return status;
sum += id;
}
id /= (double) max_comparisons;
esl_randomness_Destroy(r);
}
*ret_id = id;
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 0, argc, argv, banner, usage);
ESL_RANDOMNESS *r = NULL;
char *bitfile = esl_opt_GetString (go, "--bitfile");
int nbins = esl_opt_GetInteger(go, "-b");
int n = esl_opt_GetInteger(go, "-n");
int be_verbose = esl_opt_GetBoolean(go, "-v");
int seed = esl_opt_GetInteger(go, "-s");
if (esl_opt_GetBoolean(go, "-r")) r = esl_randomness_CreateTimeseeded();
else r = esl_randomness_Create(seed);
utest_random(seed, n, nbins, be_verbose);
utest_choose(r, n, nbins, be_verbose);
if (bitfile != NULL) save_bitfile(bitfile, r, n);
esl_randomness_Destroy(r);
esl_getopts_Destroy(go);
return 0;
}
int main(void)
{
ESL_RANDOMNESS *r = esl_randomness_CreateTimeseeded();
double a = -3.;
double b = 1.;
double xori = -20.;
double xstep = 1.0;
double setsigma = 1.0; /* sigma on all points */
int n = 100;
double *x = malloc(sizeof(double) * n);
double *y = malloc(sizeof(double) * n);
double *sigma = malloc(sizeof(double) * n);
int i;
double ae, be, siga, sigb, cov_ab, cc, Q;
/* Simulate some linear data, with Gaussian noise added to y_i */
for (i = 0; i < n; i++) {
sigma[i] = setsigma;
x[i] = xori + i*xstep;
y[i] = esl_rnd_Gaussian(r, a + b*x[i], sigma[i]);
}
if (esl_stats_LinearRegression(x, y, sigma, n, &ae, &be, &siga, &sigb, &cov_ab, &cc, &Q) != eslOK)
esl_fatal("linear regression failed");
printf("estimated intercept a = %8.4f [true = %8.4f]\n", ae, a);
printf("estimated slope b = %8.4f [true = %8.4f]\n", be, b);
printf("estimated sigma on a = %8.4f\n", siga);
printf("estimated sigma on b = %8.4f\n", sigb);
printf("estimated cov(a,b) = %8.4f\n", cov_ab);
printf("correlation coeff = %8.4f\n", cc);
printf("P-value = %8.4f\n", Q);
free(x); free(y); free(sigma);
esl_randomness_Destroy(r);
exit(0);
}
/* The esl_random() unit test:
* a binned frequency test.
*/
static void
utest_random(long seed, int n, int nbins, int be_verbose)
{
ESL_RANDOMNESS *r = NULL;
int *counts = NULL;
double X2p = 0.;
int i;
double X2, exp, diff;
if ((counts = malloc(sizeof(int) * nbins)) == NULL) esl_fatal("malloc failed");
esl_vec_ISet(counts, nbins, 0);
/* This contrived call sequence exercises CreateTimeseeded() and
* Init(), while leaving us a reproducible chain. Because it's
* reproducible, we know this test succeeds, despite being
* statistical in nature.
*/
if ((r = esl_randomness_CreateTimeseeded()) == NULL) esl_fatal("randomness create failed");
if (esl_randomness_Init(r, seed) != eslOK) esl_fatal("randomness init failed");
for (i = 0; i < n; i++)
counts[esl_rnd_Roll(r, nbins)]++;
/* X^2 value: \sum (o_i - e_i)^2 / e_i */
for (X2 = 0., i = 0; i < nbins; i++) {
exp = (double) n / (double) nbins;
diff = (double) counts[i] - exp;
X2 += diff*diff/exp;
}
if (esl_stats_ChiSquaredTest(nbins, X2, &X2p) != eslOK) esl_fatal("chi squared eval failed");
if (be_verbose) printf("random(): \t%g\n", X2p);
if (X2p < 0.01) esl_fatal("chi squared test failed");
esl_randomness_Destroy(r);
free(counts);
return;
}
int
main(int argc, char **argv)
{
int status;
ESL_RANDOMNESS *r = esl_randomness_CreateTimeseeded();
ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO);
ESL_SCOREMATRIX *S = NULL;
ESL_DSQ *x = NULL; /* iid query */
ESL_DSQ *y = NULL; /* iid target */
double lambda;
double bg[20]; /* iid background probabilities */
int L; /* query length */
int M; /* target length */
int nseq; /* number of target seqs to simulate */
int i;
int gop;
int gex;
char *mxfile = "PMX";
int raw_sc;
/* Configuration
*/
L = 400; /* query length */
M = 400; /* target length */
nseq = 50000;
gop = -11;
gex = -1;
lambda = 0.3207;
ESL_ALLOC(x, sizeof(ESL_DSQ) * (L+2));
ESL_ALLOC(y, sizeof(ESL_DSQ) * (M+2));
/* Input an amino acid score matrix from a file. */
if (mxfile != NULL) {
ESL_FILEPARSER *efp = NULL;
if ( esl_fileparser_Open(mxfile, &efp) != eslOK) esl_fatal("failed to open score file %s", mxfile);
if ( esl_sco_Read(efp, abc, &S) != eslOK) esl_fatal("failed to read matrix from %s", mxfile);
esl_fileparser_Close(efp);
} else { /* default = BLOSUM62 */
S = esl_scorematrix_Create(abc);
esl_scorematrix_SetBLOSUM62(S);
}
esl_composition_BL62(bg);
esl_rsq_xIID(r, bg, 20, L, x);
for (i = 0; i < nseq; i++)
{
esl_rsq_xIID(r, bg, 20, M, y);
esl_swat_Score(x, L, y, M, S, gop, gex, &raw_sc);
printf("%d\n", raw_sc);
}
free(x);
free(y);
esl_scorematrix_Destroy(S);
esl_alphabet_Destroy(abc);
esl_randomness_Destroy(r);
exit(0);
ERROR:
exit(status);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line configuration */
struct cfg_s cfg; /* application configuration */
char *basename= NULL; /* base of the output file names */
char *alifile = NULL; /* alignment file name */
char *dbfile = NULL; /* name of seq db file */
char outfile[256]; /* name of an output file */
int alifmt; /* format code for alifile */
int dbfmt; /* format code for dbfile */
ESL_MSAFILE *afp = NULL; /* open alignment file */
ESL_MSA *origmsa = NULL; /* one multiple sequence alignment */
ESL_MSA *msa = NULL; /* MSA after frags are removed */
ESL_MSA *trainmsa= NULL; /* training set, aligned */
ESL_STACK *teststack=NULL; /* test set: stack of ESL_SQ ptrs */
int status; /* easel return code */
int nfrags; /* # of fragments removed */
int ntestdom; /* # of test domains */
int ntest; /* # of test sequences created */
int nali; /* number of alignments read */
double avgid;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h")) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 3) cmdline_failure(argv[0], "Incorrect number of command line arguments\n");
basename = esl_opt_GetArg(go, 1);
alifile = esl_opt_GetArg(go, 2);
dbfile = esl_opt_GetArg(go, 3);
alifmt = eslMSAFILE_STOCKHOLM;
dbfmt = eslSQFILE_FASTA;
/* Set up the configuration structure shared amongst functions here */
if (esl_opt_IsDefault(go, "--seed")) cfg.r = esl_randomness_CreateTimeseeded();
else cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
cfg.abc = NULL; /* until we open the MSA file, below */
cfg.fragfrac = esl_opt_GetReal(go, "-F");
cfg.idthresh1 = esl_opt_GetReal(go, "-1");
cfg.idthresh2 = esl_opt_GetReal(go, "-2");
cfg.test_lens = NULL;
cfg.ntest = 0;
/* Open the output files */
if (snprintf(outfile, 256, "%s.msa", basename) >= 256) esl_fatal("Failed to construct output MSA file name");
if ((cfg.out_msafp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open MSA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.fa", basename) >= 256) esl_fatal("Failed to construct output FASTA file name");
if ((cfg.out_seqfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open FASTA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.pos", basename) >= 256) esl_fatal("Failed to construct pos test set summary file name");
if ((cfg.possummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open pos test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.neg", basename) >= 256) esl_fatal("Failed to construct neg test set summary file name");
if ((cfg.negsummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open neg test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.tbl", basename) >= 256) esl_fatal("Failed to construct benchmark table file name");
if ((cfg.tblfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open benchmark table file %s\n", outfile);
/* Open the MSA file; determine alphabet */
status = esl_msafile_Open(alifile, alifmt, NULL, &afp);
if (status == eslENOTFOUND) esl_fatal("Alignment file %s doesn't exist or is not readable\n", alifile);
else if (status == eslEFORMAT) esl_fatal("Couldn't determine format of alignment %s\n", alifile);
else if (status != eslOK) esl_fatal("Alignment file open failed with error %d\n", status);
if (esl_opt_GetBoolean(go, "--amino")) cfg.abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) cfg.abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) cfg.abc = esl_alphabet_Create(eslRNA);
else {
int type;
status = esl_msafile_GuessAlphabet(afp, &type);
if (status == eslEAMBIGUOUS) esl_fatal("Failed to guess the bio alphabet used in %s.\nUse --dna, --rna, or --amino option to specify it.", alifile);
else if (status == eslEFORMAT) esl_fatal("Alignment file parse failed: %s\n", afp->errbuf);
else if (status == eslENODATA) esl_fatal("Alignment file %s is empty\n", alifile);
else if (status != eslOK) esl_fatal("Failed to read alignment file %s\n", alifile);
cfg.abc = esl_alphabet_Create(type);
}
esl_msafile_SetDigital(afp, cfg.abc);
if (cfg.abc->type == eslAMINO) esl_composition_SW34(cfg.fq);
else esl_vec_DSet(cfg.fq, cfg.abc->K, 1.0 / (double) cfg.abc->K);
/* Open and process the dbfile; make sure it's in the same alphabet */
process_dbfile(&cfg, dbfile, dbfmt);
/* Read and process MSAs one at a time */
nali = 0;
while ((status = esl_msa_Read(afp, &origmsa)) == eslOK)
{
remove_fragments(&cfg, origmsa, &msa, &nfrags);
separate_sets (&cfg, msa, &trainmsa, &teststack);
ntestdom = esl_stack_ObjectCount(teststack);
if (ntestdom >= 2)
{
esl_stack_Shuffle(cfg.r, teststack);
synthesize_positives(go, &cfg, msa->name, teststack, &ntest);
esl_msa_MinimGaps(trainmsa, NULL, NULL);
esl_msa_Write(cfg.out_msafp, trainmsa, eslMSAFILE_STOCKHOLM);
esl_dst_XAverageId(cfg.abc, trainmsa->ax, trainmsa->nseq, 10000, &avgid); /* 10000 is max_comparisons, before sampling kicks in */
fprintf(cfg.tblfp, "%-20s %3.0f%% %6d %6d %6d %6d %6d %6d\n", msa->name, 100.*avgid, (int) trainmsa->alen, msa->nseq, nfrags, trainmsa->nseq, ntestdom, ntest);
nali++;
}
esl_msa_Destroy(trainmsa);
esl_msa_Destroy(origmsa);
esl_msa_Destroy(msa);
}
if (status == eslEFORMAT) esl_fatal("Alignment file parse error, line %d of file %s:\n%s\nOffending line is:\n%s\n",
afp->linenumber, afp->fname, afp->errbuf, afp->buf);
else if (status != eslEOF) esl_fatal("Alignment file read failed with error code %d\n", status);
else if (nali == 0) esl_fatal("No alignments found in file %s\n", alifile);
if (nali > 0)
synthesize_negatives(go, &cfg, esl_opt_GetInteger(go, "-N"));
fclose(cfg.out_msafp);
fclose(cfg.out_seqfp);
fclose(cfg.possummfp);
fclose(cfg.negsummfp);
fclose(cfg.tblfp);
esl_randomness_Destroy(cfg.r);
esl_alphabet_Destroy(cfg.abc);
esl_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line configuration */
struct cfg_s cfg; /* application configuration */
char *basename= NULL; /* base of the output file names */
char *alifile = NULL; /* alignment file name */
char *dbfile = NULL; /* name of seq db file */
char outfile[256]; /* name of an output file */
int alifmt; /* format code for alifile */
int dbfmt; /* format code for dbfile */
ESLX_MSAFILE *afp = NULL; /* open alignment file */
ESL_MSA *origmsa = NULL; /* one multiple sequence alignment */
ESL_MSA *msa = NULL; /* MSA after frags are removed */
ESL_MSA *trainmsa= NULL; /* training set, aligned */
ESL_STACK *teststack=NULL; /* test set: stack of ESL_SQ ptrs */
int status; /* easel return code */
int nfrags; /* # of fragments removed */
int ntestdom; /* # of test domains */
int ntest; /* # of test sequences created */
int nali; /* number of alignments read */
double avgid;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h")) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 3) cmdline_failure(argv[0], "Incorrect number of command line arguments\n");
basename = esl_opt_GetArg(go, 1);
alifile = esl_opt_GetArg(go, 2);
dbfile = esl_opt_GetArg(go, 3);
alifmt = eslMSAFILE_STOCKHOLM;
dbfmt = eslSQFILE_FASTA;
/* Set up the configuration structure shared amongst functions here */
if (esl_opt_IsDefault(go, "--seed")) cfg.r = esl_randomness_CreateTimeseeded();
else cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
cfg.abc = NULL; /* until we open the MSA file, below */
cfg.fragfrac = esl_opt_GetReal(go, "-F");
cfg.idthresh1 = esl_opt_GetReal(go, "-1");
cfg.idthresh2 = esl_opt_GetReal(go, "-2");
cfg.test_lens = NULL;
cfg.ntest = 0;
cfg.max_ntest = (esl_opt_IsOn(go, "--maxtest") ? esl_opt_GetInteger(go, "--maxtest") : 0);
cfg.max_ntrain = (esl_opt_IsOn(go, "--maxtrain") ? esl_opt_GetInteger(go, "--maxtrain") : 0);
/* Open the output files */
if (snprintf(outfile, 256, "%s.msa", basename) >= 256) esl_fatal("Failed to construct output MSA file name");
if ((cfg.out_msafp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open MSA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.fa", basename) >= 256) esl_fatal("Failed to construct output FASTA file name");
if ((cfg.out_seqfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open FASTA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.pos", basename) >= 256) esl_fatal("Failed to construct pos test set summary file name");
if ((cfg.possummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open pos test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.neg", basename) >= 256) esl_fatal("Failed to construct neg test set summary file name");
if ((cfg.negsummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open neg test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.tbl", basename) >= 256) esl_fatal("Failed to construct benchmark table file name");
if ((cfg.tblfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open benchmark table file %s\n", outfile);
if (esl_opt_GetBoolean(go, "--pid")) {
if (snprintf(outfile, 256, "%s.pid", basename) >= 256) esl_fatal("Failed to construct %%id table file name");
if ((cfg.pidfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open %%id table file %s\n", outfile);
} else cfg.pidfp = NULL;
/* Open the MSA file, digital mode; determine alphabet */
if (esl_opt_GetBoolean(go, "--amino")) cfg.abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) cfg.abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) cfg.abc = esl_alphabet_Create(eslRNA);
status = eslx_msafile_Open(&(cfg.abc), alifile, NULL, alifmt, NULL, &afp);
if (status != eslOK) eslx_msafile_OpenFailure(afp, status);
if (cfg.abc->type == eslAMINO) esl_composition_SW34(cfg.fq);
else esl_vec_DSet(cfg.fq, cfg.abc->K, 1.0 / (double) cfg.abc->K);
/* Open and process the dbfile; make sure it's in the same alphabet */
process_dbfile(&cfg, dbfile, dbfmt);
/* Read and process MSAs one at a time */
nali = 0;
while ((status = eslx_msafile_Read(afp, &origmsa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
esl_msa_ConvertDegen2X(origmsa);
esl_msa_Hash(origmsa);
remove_fragments(&cfg, origmsa, &msa, &nfrags);
separate_sets (&cfg, msa, &trainmsa, &teststack);
if ( esl_stack_ObjectCount(teststack) >= 2)
{
/* randomize test domain order, and apply size limit if any */
esl_stack_Shuffle(cfg.r, teststack);
if (cfg.max_ntest) pstack_select_topn(&teststack, cfg.max_ntest);
ntestdom = esl_stack_ObjectCount(teststack);
/* randomize training set alignment order, and apply size limit if any */
esl_msashuffle_PermuteSequenceOrder(cfg.r, trainmsa);
if (cfg.max_ntrain) msa_select_topn(&trainmsa, cfg.max_ntrain);
esl_msa_MinimGaps(trainmsa, NULL, NULL, FALSE);
if (esl_opt_GetBoolean(go, "--pid")) write_pids(cfg.pidfp, origmsa, trainmsa, teststack);
synthesize_positives(go, &cfg, msa->name, teststack, &ntest);
eslx_msafile_Write(cfg.out_msafp, trainmsa, eslMSAFILE_STOCKHOLM);
esl_dst_XAverageId(cfg.abc, trainmsa->ax, trainmsa->nseq, 10000, &avgid); /* 10000 is max_comparisons, before sampling kicks in */
fprintf(cfg.tblfp, "%-20s %3.0f%% %6d %6d %6d %6d %6d %6d\n", msa->name, 100.*avgid, (int) trainmsa->alen, msa->nseq, nfrags, trainmsa->nseq, ntestdom, ntest);
nali++;
}
esl_msa_Destroy(trainmsa);
esl_msa_Destroy(origmsa);
esl_msa_Destroy(msa);
}
if (nali == 0) esl_fatal("No alignments found in file %s\n", alifile);
synthesize_negatives(go, &cfg, esl_opt_GetInteger(go, "-N"));
fclose(cfg.out_msafp);
fclose(cfg.out_seqfp);
fclose(cfg.possummfp);
fclose(cfg.negsummfp);
fclose(cfg.tblfp);
if (cfg.pidfp) fclose(cfg.pidfp);
esl_randomness_Destroy(cfg.r);
esl_alphabet_Destroy(cfg.abc);
eslx_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
}