/* Create an SSI index file for open MSA file <afp>.
* Both name and accession of MSAs are stored as keys.
*/
static void
create_ssi_index(ESL_GETOPTS *go, ESLX_MSAFILE *afp)
{
ESL_NEWSSI *ns = NULL;
ESL_MSA *msa = NULL;
int nali = 0;
char *ssifile = NULL;
uint16_t fh;
int status;
if (afp->bf->mode_is != eslBUFFER_FILE &&
afp->bf->mode_is != eslBUFFER_ALLFILE &&
afp->bf->mode_is != eslBUFFER_MMAP)
esl_fatal("<msafile> must be a regular file to be SSI indexed");
esl_sprintf(&ssifile, "%s.ssi", afp->bf->filename);
status = esl_newssi_Open(ssifile, FALSE, &ns);
if (status == eslENOTFOUND) esl_fatal("failed to open SSI index %s", ssifile);
else if (status == eslEOVERWRITE) esl_fatal("SSI index %s already exists; delete or rename it", ssifile);
else if (status != eslOK) esl_fatal("failed to create a new SSI index");
if (esl_newssi_AddFile(ns, afp->bf->filename, afp->format, &fh) != eslOK)
esl_fatal("Failed to add MSA file %s to new SSI index\n", afp->bf->filename);
printf("Working... ");
fflush(stdout);
while ((status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK)
eslx_msafile_ReadFailure(afp, status);
nali++;
if (! msa->name)
esl_fatal("Every alignment in file must have a name to be indexed. Failed to find name of alignment #%d\n", nali);
if (esl_newssi_AddKey(ns, msa->name, fh, msa->offset, 0, 0) != eslOK)
esl_fatal("Failed to add key %s to SSI index", msa->name);
if (msa->acc && esl_newssi_AddAlias(ns, msa->acc, msa->name) != eslOK)
esl_fatal("Failed to add secondary key %s to SSI index", msa->acc);
esl_msa_Destroy(msa);
}
if (esl_newssi_Write(ns) != eslOK)
esl_fatal("Failed to write keys to ssi file %s\n", ssifile);
printf("done.\n");
if (ns->nsecondary) printf("Indexed %d alignments (%ld names and %ld accessions).\n", nali, (long) ns->nprimary, (long) ns->nsecondary);
else printf("Indexed %d alignments (%ld names).\n", nali, (long) ns->nprimary);
printf("SSI index written to file %s\n", ssifile);
free(ssifile);
esl_newssi_Close(ns);
return;
}
/* Create an SSI index file for open HMM file <hfp>.
* Both name and accession of HMMs are stored as keys.
*/
static void
create_ssi_index(ESL_GETOPTS *go, P7_HMMFILE *hfp)
{
ESL_NEWSSI *ns = NULL;
ESL_ALPHABET *abc = NULL;
P7_HMM *hmm = NULL;
int nhmm = 0;
char *ssifile = NULL;
uint16_t fh;
int status;
if (esl_sprintf(&ssifile, "%s.ssi", hfp->fname) != eslOK) p7_Die("esl_sprintf() failed");
status = esl_newssi_Open(ssifile, FALSE, &ns);
if (status == eslENOTFOUND) esl_fatal("failed to open SSI index %s", ssifile);
else if (status == eslEOVERWRITE) esl_fatal("SSI index %s already exists; delete or rename it", ssifile);
else if (status != eslOK) esl_fatal("failed to create a new SSI index");
if (esl_newssi_AddFile(ns, hfp->fname, 0, &fh) != eslOK) /* 0 = format code (HMMs don't have any yet) */
esl_fatal("Failed to add HMM file %s to new SSI index\n", hfp->fname);
printf("Working... ");
fflush(stdout);
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) p7_Fail("read failed, HMM file %s may be truncated?", hfp->fname);
else if (status == eslEFORMAT) p7_Fail("bad file format in HMM file %s", hfp->fname);
else if (status == eslEINCOMPAT) p7_Fail("HMM file %s contains different alphabets", hfp->fname);
else if (status != eslOK) p7_Fail("Unexpected error in reading HMMs from %s", hfp->fname);
nhmm++;
if (hmm->name == NULL) p7_Fail("Every HMM must have a name to be indexed. Failed to find name of HMM #%d\n", nhmm);
if (esl_newssi_AddKey(ns, hmm->name, fh, hmm->offset, 0, 0) != eslOK)
p7_Fail("Failed to add key %s to SSI index", hmm->name);
if (hmm->acc) {
if (esl_newssi_AddAlias(ns, hmm->acc, hmm->name) != eslOK)
p7_Fail("Failed to add secondary key %s to SSI index", hmm->acc);
}
p7_hmm_Destroy(hmm);
}
if (esl_newssi_Write(ns) != eslOK)
p7_Fail("Failed to write keys to ssi file %s\n", ssifile);
printf("done.\n");
if (ns->nsecondary > 0)
printf("Indexed %d HMMs (%ld names and %ld accessions).\n", nhmm, (long) ns->nprimary, (long) ns->nsecondary);
else
printf("Indexed %d HMMs (%ld names).\n", nhmm, (long) ns->nprimary);
printf("SSI index written to file %s\n", ssifile);
free(ssifile);
esl_alphabet_Destroy(abc);
esl_newssi_Close(ns);
return;
}
/* Function: p7_hmmcache_SetNumericNames()
* Synopsis: Rename each profile in cache with a numeric name.
*
* Purpose: Rename every profile in profile cache <cache>
* with a numeric code, starting from "000000001".
*
* The code is nine digits long, left padded with
* 0's.
*
* Returns: <eslOK> on success.
*
* Throws: <eslEMEM> on allocation failure.
*/
int
p7_hmmcache_SetNumericNames(P7_HMMCACHE *cache)
{
int namelen = 9; /* 9 digit numeric code: 000000001, 000000002... */
P7_PROFILE *gm;
P7_OPROFILE *om;
int i;
int status;
for (i = 0; i < cache->n; i++)
{
gm = cache->gmlist[i];
if (gm->name) free(gm->name);
if (( status = esl_sprintf(&(gm->name), "%0*d", namelen, i+1)) != eslOK) return status;
om = cache->omlist[i];
if (om->name) free(om->name);
if (( status = esl_sprintf(&(om->name), "%0*d", namelen, i+1)) != eslOK) return status;
}
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *alifile = NULL; /* alignment file name */
int infmt = eslMSAFILE_UNKNOWN; /* format code for alifile */
int outfmt = eslMSAFILE_UNKNOWN; /* output format for fetched msa's */
ESLX_MSAFILE *afp = NULL; /* open alignment file */
FILE *ofp = NULL; /* output stream for alignments */
int status; /* easel return code */
/***********************************************
* 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 configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) < 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
if (esl_opt_IsOn(go, "--informat")) {
infmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslMSAFILE_UNKNOWN) esl_fatal("%s is not a valid input alignment file format for --informat", esl_opt_GetString(go, "--informat"));
}
outfmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--outformat"));
if (outfmt == eslMSAFILE_UNKNOWN) esl_fatal("%s is not a valid output alignment file format for --outformat", esl_opt_GetString(go, "--outformat"));
alifile = esl_opt_GetArg(go, 1);
/* Open the alignment file. */
if ( (status = eslx_msafile_Open(NULL, alifile, NULL, infmt, NULL, &afp)) != eslOK)
eslx_msafile_OpenFailure(afp, status);
/* Open the SSI index, if any */
if (! esl_opt_GetBoolean(go, "--index"))
{
if (afp->bf->mode_is == eslBUFFER_FILE ||
afp->bf->mode_is == eslBUFFER_ALLFILE ||
afp->bf->mode_is == eslBUFFER_MMAP)
{
char *ssifile = NULL;
esl_sprintf(&ssifile, "%s.ssi", afp->bf->filename);
status = esl_ssi_Open(ssifile, &(afp->ssi));
if (status == eslERANGE ) esl_fatal("SSI index %s has 64-bit offsets; this system doesn't support them", ssifile);
else if (status == eslEFORMAT) esl_fatal("SSI index %s has an unrecognized format. Try recreating, w/ esl-afetch --index", ssifile);
else if (status == eslENOTFOUND) afp->ssi = NULL;
else if (status != eslOK) esl_fatal("SSI index %s: open failed, error code %d\n", ssifile, status);
free(ssifile);
}
}
/* Open the output file, if any
*/
if (esl_opt_GetBoolean(go, "-O"))
{
if ((ofp = fopen(esl_opt_GetArg(go, 2), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetArg(go, 2));
}
else if (esl_opt_GetString(go, "-o") != NULL)
{
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/* Hand off control flow as appropriate */
if (esl_opt_GetBoolean(go, "--index"))
{
if (esl_opt_ArgNumber(go) != 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
create_ssi_index(go, afp);
}
else if (esl_opt_GetBoolean(go, "-f"))
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
multifetch(go, ofp, outfmt, esl_opt_GetArg(go, 2), afp);
}
else
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
onefetch(go, ofp, outfmt, esl_opt_GetArg(go, 2), afp);
if (ofp != stdout) printf("\n\nRetrieved alignment %s.\n", esl_opt_GetArg(go, 2));
}
eslx_msafile_Close(afp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 1, argc, argv, banner, usage);
ESL_ALPHABET *abc = NULL;
char *hmmfile = esl_opt_GetArg(go, 1);
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
P7_OPROFILE *om = NULL;
char *fname = NULL;
char *pname = NULL;
FILE *ffp = NULL;
FILE *pfp = NULL;
int nmodel = 0;
uint64_t totM = 0;
int status;
status = p7_hmmfile_Open(hmmfile, NULL, &hfp);
if (status == eslENOTFOUND) esl_fatal("Failed to open HMM file %s for reading.\n", hmmfile);
else if (status == eslEFORMAT) esl_fatal("File %s does not appear to be in a recognized HMM format.\n", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error %d in opening HMM file %s.\n", status, hmmfile);
esl_sprintf(&fname, "%s.h3f", hmmfile);
esl_sprintf(&pname, "%s.h3f", hmmfile);
if ((ffp = fopen(fname, "wb")) == NULL) esl_fatal("failed to open %s\n", fname);
if ((pfp = fopen(pname, "wb")) == NULL) esl_fatal("failed to open %s\n", pname);
free(fname);
free(pname);
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) == eslOK)
{
if (nmodel == 0) { /* first time initialization, now that alphabet known */
bg = p7_bg_Create(abc);
p7_bg_SetLength(bg, 400);
}
if (esl_opt_GetBoolean(go, "-v")) printf("%s\n", hmm->name);
nmodel++;
totM += hmm->M;
gm = p7_profile_Create(hmm->M, abc);
p7_ProfileConfig(hmm, bg, gm, 400, p7_LOCAL);
om = p7_oprofile_Create(gm->M, abc);
p7_oprofile_Convert(gm, om);
p7_oprofile_Write(ffp, pfp, om);
p7_profile_Destroy(gm);
p7_oprofile_Destroy(om);
p7_hmm_Destroy(hmm);
}
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 != eslEOF) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
fclose(ffp);
fclose(pfp);
p7_bg_Destroy(bg);
p7_hmmfile_Close(hfp);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return 0;
}
static void
utest_ReadWrite(P7_HMM *hmm, P7_OPROFILE *om)
{
char *msg = "oprofile read/write unit test failure";
ESL_ALPHABET *abc = NULL;
P7_OPROFILE *om2 = NULL;
char tmpfile[16] = "esltmpXXXXXX";
char *mfile = NULL;
char *ffile = NULL;
char *pfile = NULL;
char *ssifile = NULL;
FILE *fp = NULL;
FILE *mfp = NULL;
FILE *ffp = NULL;
FILE *pfp = NULL;
ESL_NEWSSI *nssi = NULL;
P7_HMMFILE *hfp = NULL;
uint16_t fh = 0;
float tolerance = 0.001;
char errbuf[eslERRBUFSIZE];
/* 1. A mini version of hmmpress: save the test HMM to a file along with its associated .h3{mfpi} files
*/
if ( esl_tmpfile_named(tmpfile, &fp) != eslOK) esl_fatal(msg);
if ( esl_sprintf(&mfile, "%s.h3m", tmpfile) != eslOK) esl_fatal(msg);
if ( esl_sprintf(&ffile, "%s.h3f", tmpfile) != eslOK) esl_fatal(msg);
if ( esl_sprintf(&pfile, "%s.h3p", tmpfile) != eslOK) esl_fatal(msg);
if ( esl_sprintf(&ssifile, "%s.h3i", tmpfile) != eslOK) esl_fatal(msg);
if ( esl_newssi_Open(ssifile, TRUE, &nssi) != eslOK) esl_fatal(msg);
if (( mfp = fopen(mfile, "wb")) == NULL) esl_fatal(msg);
if (( ffp = fopen(ffile, "wb")) == NULL) esl_fatal(msg);
if (( pfp = fopen(pfile, "wb")) == NULL) esl_fatal(msg);
/* the disk offsets are all 0 by construction, if there's only one
* HMM in the file - but don't want to forget them, if we change the
* unit test in the future to be multi HMM
*/
if ((om->offs[p7_MOFFSET] = ftello(mfp)) == -1) esl_fatal(msg);
if ((om->offs[p7_FOFFSET] = ftello(ffp)) == -1) esl_fatal(msg);
if ((om->offs[p7_POFFSET] = ftello(pfp)) == -1) esl_fatal(msg);
if ( p7_hmmfile_WriteASCII(fp, -1, hmm) != eslOK) esl_fatal(msg);
if ( p7_hmmfile_WriteBinary(mfp, -1, hmm) != eslOK) esl_fatal(msg);
if ( p7_oprofile_Write(ffp, pfp, om) != eslOK) esl_fatal(msg);
if ( esl_newssi_AddFile(nssi, tmpfile, 0, &fh) != eslOK) esl_fatal(msg);
if ( esl_newssi_AddKey (nssi, hmm->name, fh, om->offs[p7_MOFFSET], 0, 0) != eslOK) esl_fatal(msg);
if ( esl_newssi_Write(nssi) != eslOK) esl_fatal(msg);
fclose(fp);
fclose(mfp);
fclose(ffp);
fclose(pfp);
esl_newssi_Close(nssi);
/* 2. read the optimized profile back in */
if ( p7_hmmfile_Open(tmpfile, NULL, &hfp) != eslOK) esl_fatal(msg);
if ( p7_oprofile_ReadMSV(hfp, &abc, &om2) != eslOK) esl_fatal(msg);
if ( p7_oprofile_ReadRest(hfp, om2) != eslOK) esl_fatal(msg);
/* 3. it should be identical to the original */
if ( p7_oprofile_Compare(om, om2, tolerance, errbuf) != eslOK) esl_fatal("%s\n%s", msg, errbuf);
p7_oprofile_Destroy(om2);
p7_hmmfile_Close(hfp);
esl_alphabet_Destroy(abc);
remove(ssifile);
remove(ffile);
remove(pfile);
remove(mfile);
remove(tmpfile);
free(ssifile);
free(mfile);
free(ffile);
free(pfile);
}