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;
}
int main(int argc, char **argv)
{
ESL_MSAFILE *afp;
ESL_MSA *msa;
ESL_DMATRIX *P;
int status;
int i,j;
double min, avg, max;
esl_msafile_Open(argv[1], eslMSAFILE_UNKNOWN, NULL, &afp);
esl_msa_Read(afp, &msa);
esl_dst_CPairIdMx(msa->aseq, msa->nseq, &P);
min = 1.0;
max = 0.0;
avg = 0.0;
for (i = 0; i < msa->nseq; i++)
for (j = i+1; j < msa->nseq; j++)
{
avg += P->mx[i][j];
if (P->mx[i][j] < min) min = P->mx[i][j];
if (P->mx[i][j] > max) max = P->mx[i][j];
}
avg /= (double) (msa->nseq * (msa->nseq-1) / 2);
printf("Average pairwise %% id: %.1f%%\n", avg * 100.);
printf("Minimum pairwise %% id: %.1f%%\n", min * 100.);
printf("Maximum pairwise %% id: %.1f%%\n", max * 100.);
esl_dmatrix_Destroy(P);
esl_msa_Destroy(msa);
esl_msafile_Close(afp);
return 0;
}
/* msa_shuffling()
* SRE, Tue Jan 22 08:39:51 2008 [Market Street Cafe, Leesburg]
*
* Shuffling multiple sequence alignments
*/
static int
msa_shuffling(ESL_GETOPTS *go, ESL_RANDOMNESS *r, FILE *ofp, int outfmt)
{
char *msafile = esl_opt_GetArg(go, 1);
int infmt = eslMSAFILE_UNKNOWN;
ESL_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
ESL_MSA *shuf = NULL;
int N = esl_opt_GetInteger(go, "-N");
int i;
int status, mstatus;
status = esl_msafile_Open(msafile, infmt, NULL, &afp);
if (status == eslENOTFOUND) esl_fatal("Alignment file %s isn't readable\n", msafile);
else if (status == eslEFORMAT) esl_fatal("Couldn't determine format of %s\n", msafile);
else if (status != eslOK) esl_fatal("Alignment file open failed (error %d)\n", status);
while ((mstatus = esl_msa_Read(afp, &msa)) != eslEOF)
{
if (status == eslEFORMAT) esl_fatal("Alignment file parse error:\n%s\n", afp->errbuf);
else if (status == eslEINVAL) esl_fatal("Alignment file parse error:\n%s\n", afp->errbuf);
else if (status != eslOK) esl_fatal("Alignment file read failed with error code %d\n", status);
shuf = esl_msa_Clone(msa);
for (i = 0; i < N; i++)
{
if (esl_opt_GetBoolean(go, "--boot")) esl_msashuffle_Bootstrap(r, msa, shuf);
else esl_msashuffle_Shuffle (r, msa, shuf);
/* Set the name of the shuffled alignment */
if (msa->name != NULL) {
if (esl_opt_GetBoolean(go, "--boot")) {
if (N > 1) esl_msa_FormatName(shuf, "%s-sample-%d", msa->name, i);
else esl_msa_FormatName(shuf, "%s-sample", msa->name);
} else {
if (N > 1) esl_msa_FormatName(shuf, "%s-shuffle-%d", msa->name, i);
else esl_msa_FormatName(shuf, "%s-shuffle", msa->name);
}
} else {
if (esl_opt_GetBoolean(go, "--boot")) {
if (N > 1) esl_msa_FormatName(shuf, "sample-%d", i);
else esl_msa_FormatName(shuf, "sample");
} else {
if (N > 1) esl_msa_FormatName(shuf, "shuffle-%d", i);
else esl_msa_FormatName(shuf, "shuffle");
}
}
esl_msa_Write(ofp, shuf, outfmt);
}
esl_msa_Destroy(shuf);
esl_msa_Destroy(msa);
}
return eslOK;
}
int
main(int argc, char **argv)
{
char *filename = argv[1];
int fmt = eslMSAFILE_UNKNOWN;
int type = eslUNKNOWN;
ESL_ALPHABET *abc = NULL;
ESL_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
double maxid = 0.62; /* cluster at 62% identity: the BLOSUM62 rule */
int *assignment = NULL;
int *nin = NULL;
int nclusters;
int c, i;
int status;
/* Open; guess alphabet; set to digital mode */
status = esl_msafile_Open(filename, fmt, NULL, &afp);
if (status == eslENOTFOUND) esl_fatal("Alignment file %s isn't readable", filename);
else if (status == eslEFORMAT) esl_fatal("Couldn't determine format of %s", filename);
else if (status != eslOK) esl_fatal("Alignment file open failed (error code %d)", status);
status = esl_msafile_GuessAlphabet(afp, &type);
if (status == eslEAMBIGUOUS) esl_fatal("Couldn't guess alphabet from first alignment in %s", filename);
else if (status == eslEFORMAT) esl_fatal("Alignment file parse error, line %d of file %s:\n%s\nBad line is: %s\n",
afp->linenumber, afp->fname, afp->errbuf, afp->buf);
else if (status == eslENODATA) esl_fatal("Alignment file %s contains no data?", filename);
else if (status != eslOK) esl_fatal("Failed to guess alphabet (error code %d)\n", status);
abc = esl_alphabet_Create(type);
esl_msafile_SetDigital(afp, abc);
/* read one alignment */
status = esl_msa_Read(afp, &msa);
if (status == eslEFORMAT) esl_fatal("alignment file %s: %s\n", afp->fname, afp->errbuf);
else if (status != eslOK) esl_fatal("Alignment file read failed with error code %d\n", status);
/* do the clustering */
esl_msacluster_SingleLinkage(msa, maxid, &assignment, &nin, &nclusters);
printf("%d clusters at threshold of %f fractional identity\n", nclusters, maxid);
for (c = 0; c < nclusters; c++) {
printf("cluster %d:\n", c);
for (i = 0; i < msa->nseq; i++) if (assignment[i] == c) printf(" %s\n", msa->sqname[i]);
printf("(%d sequences)\n\n", nin[c]);
}
esl_msa_Destroy(msa);
esl_msafile_Close(afp);
free(assignment);
free(nin);
return 0;
}
int main(int argc, char **argv)
{
ESL_MSAFILE *afp;
ESL_MSA *msa;
int i;
esl_msafile_Open(argv[1], eslMSAFILE_UNKNOWN, NULL, &afp);
esl_msa_Read(afp, &msa);
esl_msafile_Close(afp);
esl_msaweight_GSC(msa);
for (i = 0; i < msa->nseq; i++)
printf("%20s %f\n", msa->sqname[i], msa->wgt[i]);
return 0;
}
static int
thread_loop(ESL_THREADS *obj, ESL_WORK_QUEUE *queue, struct cfg_s *cfg)
{
int status = eslOK;
int sstatus = eslOK;
int processed = 0;
WORK_ITEM *item;
void *newItem;
int next = 1;
PENDING_ITEM *top = NULL;
PENDING_ITEM *empty = NULL;
PENDING_ITEM *tmp = NULL;
char errmsg[eslERRBUFSIZE];
esl_workqueue_Reset(queue);
esl_threads_WaitForStart(obj);
status = esl_workqueue_ReaderUpdate(queue, NULL, &newItem);
if (status != eslOK) esl_fatal("Work queue reader failed");
/* Main loop: */
item = (WORK_ITEM *) newItem;
while (sstatus == eslOK) {
sstatus = esl_msa_Read(cfg->afp, &item->msa);
if (sstatus == eslOK) {
item->nali = ++cfg->nali;
if (set_msa_name(cfg, errmsg, item->msa) != eslOK) p7_Fail("%s\n", errmsg);
}
if (sstatus == eslEOF && processed < cfg->nali) sstatus = eslOK;
if (sstatus == eslOK) {
status = esl_workqueue_ReaderUpdate(queue, item, &newItem);
if (status != eslOK) esl_fatal("Work queue reader failed");
/* process any results */
item = (WORK_ITEM *) newItem;
if (item->processed == TRUE) {
++processed;
/* try to keep the input output order the same */
if (item->nali == next) {
sstatus = output_result(cfg, errmsg, item->nali, item->msa, item->hmm, item->postmsa, item->entropy);
if (sstatus != eslOK) p7_Fail(errmsg);
p7_hmm_Destroy(item->hmm);
esl_msa_Destroy(item->msa);
esl_msa_Destroy(item->postmsa);
++next;
/* output any pending msa as long as the order
* remains the same as read in.
*/
while (top != NULL && top->nali == next) {
sstatus = output_result(cfg, errmsg, top->nali, top->msa, top->hmm, top->postmsa, top->entropy);
if (sstatus != eslOK) p7_Fail(errmsg);
p7_hmm_Destroy(top->hmm);
esl_msa_Destroy(top->msa);
esl_msa_Destroy(top->postmsa);
tmp = top;
top = tmp->next;
tmp->next = empty;
empty = tmp;
++next;
}
} else {
/* queue up the msa so the sequence order is the same in
* the .sto and .hmm
*/
if (empty != NULL) {
tmp = empty;
empty = tmp->next;
} else {
ESL_ALLOC(tmp, sizeof(PENDING_ITEM));
}
tmp->nali = item->nali;
tmp->hmm = item->hmm;
tmp->msa = item->msa;
tmp->postmsa = item->postmsa;
tmp->entropy = item->entropy;
/* add the msa to the pending list */
if (top == NULL || tmp->nali < top->nali) {
tmp->next = top;
top = tmp;
} else {
PENDING_ITEM *ptr = top;
while (ptr->next != NULL && tmp->nali > ptr->next->nali) {
ptr = ptr->next;
}
tmp->next = ptr->next;
ptr->next = tmp;
}
}
item->nali = 0;
item->processed = FALSE;
item->hmm = NULL;
item->msa = NULL;
item->postmsa = NULL;
item->entropy = 0.0;
}
}
}
if (top != NULL) esl_fatal("Top is not empty\n");
while (empty != NULL) {
tmp = empty;
empty = tmp->next;
free(tmp);
}
status = esl_workqueue_ReaderUpdate(queue, item, NULL);
if (status != eslOK) esl_fatal("Work queue reader failed");
if (sstatus == eslEOF)
{
/* wait for all the threads to complete */
esl_threads_WaitForFinish(obj);
esl_workqueue_Complete(queue);
}
return sstatus;
ERROR:
return eslEMEM;
}
/* 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 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;
char *msafile;
ESL_MSAFILE *afp;
ESL_MSA *msa;
int do_gsc;
int do_pb;
int do_blosum;
int maxN;
double maxid;
int nsmall, nbig;
int i;
/* Process command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE){
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
if ((msafile = esl_opt_GetArg(go, 1)) == NULL) esl_fatal("%s", go->errbuf);
esl_getopts_Destroy(go);
/* Weight one or more alignments from input file
*/
esl_msafile_Open(msafile, eslMSAFILE_UNKNOWN, NULL, &afp);
while (esl_msa_Read(afp, &msa) == eslOK)
{
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
if (do_gsc) esl_msaweight_GSC(msa);
else if (do_pb) esl_msaweight_PB(msa);
else if (do_blosum) esl_msaweight_BLOSUM(msa, maxid);
for (nsmall = 0, nbig = 0, i = 0; i < msa->nseq; i++) {
if (msa->wgt[i] < 0.2) nsmall++;
if (msa->wgt[i] > 5.0) nbig++;
}
printf("%-20s %5d %5d %8.4f %8.4f %5d %5d\n",
msa->name,
msa->nseq,
msa->alen,
esl_vec_DMin(msa->wgt, msa->nseq),
esl_vec_DMax(msa->wgt, msa->nseq),
nsmall,
nbig);
esl_msa_Destroy(msa);
}
esl_msafile_Close(afp);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_STOPWATCH *w;
ESL_GETOPTS *go;
char *msafile;
ESL_MSAFILE *afp;
ESL_MSA *msa;
int do_gsc;
int do_pb;
int do_blosum;
int maxN;
double maxid;
double cpu;
/* Process command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("failed to parse cmd line: %s", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("failed to parse cmd line: %s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
if ((msafile = esl_opt_GetArg(go, 1)) == NULL) esl_fatal("failed to parse cmd line: %s", go->errbuf);
esl_getopts_Destroy(go);
w = esl_stopwatch_Create();
/* Weight one or more alignments from input file
*/
esl_msafile_Open(msafile, eslMSAFILE_UNKNOWN, NULL, &afp);
while (esl_msa_Read(afp, &msa) == eslOK)
{
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
esl_stopwatch_Start(w);
if (do_gsc) esl_msaweight_GSC(msa);
else if (do_pb) esl_msaweight_PB(msa);
else if (do_blosum) esl_msaweight_BLOSUM(msa, maxid);
esl_stopwatch_Stop(w);
cpu = w->user;
printf("%-20s %6d %6d %.3f\n", msa->name, msa->alen, msa->nseq, cpu);
esl_msa_Destroy(msa);
}
esl_msafile_Close(afp);
esl_stopwatch_Destroy(w);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go;
char *msafile;
ESL_MSAFILE *afp;
ESL_MSA *msa;
float *sqd;
int status;
int nbad;
int nali = 0;
int nbadali = 0;
int nwgt = 0;
int nbadwgt = 0;
int i;
int be_quiet;
int do_gsc;
int do_pb;
int do_blosum;
double maxid;
double tol;
int maxN;
/* Process command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("failed to parse cmd line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("failed to parse cmd line: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
be_quiet = esl_opt_GetBoolean(go, "-q");
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
tol = esl_opt_GetReal (go, "--tol");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
msafile = esl_opt_GetArg(go, 1);
esl_getopts_Destroy(go);
/* Weight one or more alignments from input file
*/
esl_msafile_Open(msafile, eslMSAFILE_UNKNOWN, NULL, &afp);
while (esl_msa_Read(afp, &msa) == eslOK)
{
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
nali++;
nwgt += msa->nseq;
ESL_ALLOC(sqd, sizeof(float) * msa->nseq);
if (do_gsc) {
esl_msaweight_GSC(msa);
GSCWeights(msa->aseq, msa->nseq, msa->alen, sqd);
} else if (do_pb) {
esl_msaweight_PB(msa);
PositionBasedWeights(msa->aseq, msa->nseq, msa->alen, sqd);
} else if (do_blosum) {
esl_msaweight_BLOSUM(msa, maxid);
BlosumWeights(msa->aseq, msa->nseq, msa->alen, maxid, sqd);
/* workaround SQUID bug: BLOSUM weights weren't renormalized to sum to nseq. */
esl_vec_FNorm (sqd, msa->nseq);
esl_vec_FScale(sqd, msa->nseq, (float) msa->nseq);
}
if (! be_quiet) {
for (i = 0; i < msa->nseq; i++)
fprintf(stdout, "%-20s %.3f %.3f\n",
msa->sqname[i], msa->wgt[i], sqd[i]);
}
nbad = 0;
for (i = 0; i < msa->nseq; i++)
if (esl_DCompare((double) sqd[i], msa->wgt[i], tol) != eslOK)
nbad++;
if (nbad > 0) nbadali++;
nbadwgt += nbad;
if (nbad > 0) printf("%-20s :: alignment shows %d weights that differ (out of %d) \n",
msa->name, nbad, msa->nseq);
esl_msa_Destroy(msa);
free(sqd);
}
esl_msafile_Close(afp);
if (nbadali == 0)
printf("OK: all weights identical between squid and Easel in %d alignment(s)\n", nali);
else {
printf("%d of %d weights mismatched at (> %f fractional difference)\n",
nbadwgt, nwgt, tol);
printf("involving %d of %d total alignments\n", nbadali, nali);
}
return eslOK;
ERROR:
return status;
}
