static void
emit_alignment(ESL_GETOPTS *go, FILE *ofp, int outfmt, ESL_RANDOMNESS *r, P7_HMM *hmm)
{
ESL_MSA *msa = NULL;
ESL_SQ **sq = NULL;
P7_TRACE **tr = NULL;
int N = esl_opt_GetInteger(go, "-N");
int optflags = p7_ALL_CONSENSUS_COLS;
int i;
if ((tr = malloc(sizeof(P7_TRACE *) * N)) == NULL) esl_fatal("failed to allocate trace array");
if ((sq = malloc(sizeof(ESL_SQ *) * N)) == NULL) esl_fatal("failed to allocate seq array");
for (i = 0; i < N; i++)
{
if ((sq[i] = esl_sq_CreateDigital(hmm->abc)) == NULL) esl_fatal("failed to allocate seq");
if ((tr[i] = p7_trace_Create()) == NULL) esl_fatal("failed to allocate trace");
}
for (i = 0; i < N; i++)
{
if (p7_CoreEmit(r, hmm, sq[i], tr[i]) != eslOK) esl_fatal("Failed to emit sequence");
if (esl_sq_FormatName(sq[i], "%s-sample%d", hmm->name, i+1) != eslOK) esl_fatal("Failed to set sequence name\n");
}
p7_tracealign_Seqs(sq, tr, N, hmm->M, optflags, hmm, &msa);
eslx_msafile_Write(ofp, msa, outfmt);
for (i = 0; i < N; i++) p7_trace_Destroy(tr[i]); free(tr);
for (i = 0; i < N; i++) esl_sq_Destroy(sq[i]); free(sq);
esl_msa_Destroy(msa);
return;
}
/* Function: p7_tophits_Alignment()
* Synopsis: Create a multiple alignment of all the included domains.
*
* Purpose: Create a multiple alignment of all domains marked
* "includable" in the top hits list <th>, and return it in
* <*ret_msa>.
*
* Use of <optflags> is identical to <optflags> in <p7_tracealign_Seqs()>.
* Possible flags include <p7_DIGITIZE>, <p7_ALL_CONSENSUS_COLS>,
* and <p7_TRIM>; they may be OR'ed together. Otherwise, pass
* <p7_DEFAULT> to set no flags.
*
* Caller may optionally provide <inc_sqarr>, <inc_trarr>, and
* <inc_n> to include additional sequences in the alignment
* (the jackhmmer query, for example). Otherwise, pass <NULL, NULL, 0>.
*
* Returns: <eslOK> on success, and <*ret_msa> points to a new MSA that
* the caller is responsible for freeing.
*
* Returns <eslFAIL> if there are no reported domains that
* satisfy reporting thresholds, in which case <*ret_msa>
* is <NULL>.
*
* Throws: <eslEMEM> on allocation failure; <eslECORRUPT> on
* unexpected internal data corruption.
*
* Xref: J4/29: incept.
* J4/76: added inc_sqarr, inc_trarr, inc_n, optflags
*/
int
p7_tophits_Alignment(const P7_TOPHITS *th, const ESL_ALPHABET *abc,
ESL_SQ **inc_sqarr, P7_TRACE **inc_trarr, int inc_n,
int optflags, ESL_MSA **ret_msa)
{
ESL_SQ **sqarr = NULL;
P7_TRACE **trarr = NULL;
ESL_MSA *msa = NULL;
int ndom = 0;
int h, d, y;
int M;
int status;
/* How many domains will be included in the new alignment?
* We also set model size M here; every alignment has a copy.
*/
for (h = 0; h < th->N; h++)
if (th->hit[h]->flags & p7_IS_INCLUDED)
{
for (d = 0; d < th->hit[h]->ndom; d++)
if (th->hit[h]->dcl[d].is_included)
ndom++;
}
if (inc_n+ndom == 0) { status = eslFAIL; goto ERROR; }
if (inc_n) M = inc_trarr[0]->M;
else M = th->hit[0]->dcl[0].ad->M;
/* Allocation */
ESL_ALLOC(sqarr, sizeof(ESL_SQ *) * (ndom + inc_n));
ESL_ALLOC(trarr, sizeof(P7_TRACE *) * (ndom + inc_n));
/* Inclusion of preexisting seqs, traces: make copy of pointers */
for (y = 0; y < inc_n; y++) { sqarr[y] = inc_sqarr[y]; trarr[y] = inc_trarr[y]; }
for (; y < (ndom+inc_n); y++) { sqarr[y] = NULL; trarr[y] = NULL; }
/* Make faux sequences, traces from hit list */
y = inc_n;
for (h = 0; h < th->N; h++)
if (th->hit[h]->flags & p7_IS_INCLUDED)
{
for (d = 0; d < th->hit[h]->ndom; d++)
if (th->hit[h]->dcl[d].is_included)
{
if ((status = p7_alidisplay_Backconvert(th->hit[h]->dcl[d].ad, abc, &(sqarr[y]), &(trarr[y]))) != eslOK) goto ERROR;
y++;
}
}
/* Make the multiple alignment */
if ((status = p7_tracealign_Seqs(sqarr, trarr, inc_n+ndom, M, optflags, NULL, &msa)) != eslOK) goto ERROR;
/* Clean up */
for (y = inc_n; y < ndom+inc_n; y++) esl_sq_Destroy(sqarr[y]);
for (y = inc_n; y < ndom+inc_n; y++) p7_trace_Destroy(trarr[y]);
free(sqarr);
free(trarr);
*ret_msa = msa;
return eslOK;
ERROR:
if (sqarr != NULL) { for (y = inc_n; y < ndom+inc_n; y++) if (sqarr[y] != NULL) esl_sq_Destroy(sqarr[y]); free(sqarr); }
if (trarr != NULL) { for (y = inc_n; y < ndom+inc_n; y++) if (trarr[y] != NULL) p7_trace_Destroy(trarr[y]); free(trarr); }
if (msa != NULL) esl_msa_Destroy(msa);
*ret_msa = NULL;
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *hmmfile = NULL; /* HMM file name */
char *seqfile = NULL; /* sequence file name */
char *mapfile = NULL; /* optional mapped MSA file name */
int infmt = eslSQFILE_UNKNOWN;
int outfmt = eslMSAFILE_STOCKHOLM;
P7_HMMFILE *hfp = NULL; /* open HMM file */
ESL_SQFILE *sqfp = NULL; /* open sequence file */
char *outfile = NULL; /* output filename */
FILE *ofp = stdout; /* output stream */
ESL_SQ **sq = NULL; /* array of sequences */
void *p = NULL; /* tmp ptr for reallocation */
int nseq = 0; /* # of sequences in <seqfile> */
int mapseq = 0; /* # of sequences in mapped MSA */
int totseq = 0; /* # of seqs in all sources */
ESL_ALPHABET *abc = NULL; /* alphabet (set from the HMM file)*/
P7_HMM *hmm = NULL;
P7_TRACE **tr = NULL; /* array of tracebacks */
ESL_MSA *msa = NULL; /* resulting multiple alignment */
int msaopts = 0; /* flags to p7_tracealign_Seqs() */
int idx; /* counter over seqs, traces */
int status; /* easel/hmmer return code */
char errbuf[eslERRBUFSIZE];
/* Parse the 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) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
hmmfile = esl_opt_GetArg(go, 1);
seqfile = esl_opt_GetArg(go, 2);
if (strcmp(hmmfile, "-") == 0 && strcmp(seqfile, "-") == 0)
cmdline_failure(argv[0], "Either <hmmfile> or <seqfile> may be '-' (to read from stdin), but not both.\n");
msaopts |= p7_ALL_CONSENSUS_COLS; /* default as of 3.1 */
if (esl_opt_GetBoolean(go, "--trim")) msaopts |= p7_TRIM;
/* If caller declared an input format, decode it
*/
if (esl_opt_IsOn(go, "--informat")) {
infmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslSQFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized input sequence file format\n", esl_opt_GetString(go, "--informat"));
}
/* Determine output alignment file format */
outfmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--outformat"));
if (outfmt == eslMSAFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized output MSA file format\n", esl_opt_GetString(go, "--outformat"));
/* Open output stream */
if ( (outfile = esl_opt_GetString(go, "-o")) != NULL)
{
if ((ofp = fopen(outfile, "w")) == NULL)
cmdline_failure(argv[0], "failed to open -o output file %s for writing\n", outfile);
}
/* If caller forced an alphabet on us, create the one the caller wants
*/
if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
/* Read one HMM, and make sure there's only one.
*/
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);
if (status == eslEFORMAT) p7_Fail("Bad file format in HMM file %s:\n%s\n", hfp->fname, hfp->errbuf);
else if (status == eslEINCOMPAT) p7_Fail("HMM in %s is not in the expected %s alphabet\n", hfp->fname, esl_abc_DecodeType(abc->type));
else if (status == eslEOF) p7_Fail("Empty HMM file %s? No HMM data found.\n", hfp->fname);
else if (status != eslOK) p7_Fail("Unexpected error in reading HMMs from %s\n", hfp->fname);
status = p7_hmmfile_Read(hfp, &abc, NULL);
if (status != eslEOF) p7_Fail("HMM file %s does not contain just one HMM\n", hfp->fname);
p7_hmmfile_Close(hfp);
/* We're going to build up two arrays: sequences and traces.
* If --mapali option is chosen, the first set of sequences/traces is from the provided alignment
*/
if ( (mapfile = esl_opt_GetString(go, "--mapali")) != NULL)
{
map_alignment(mapfile, hmm, &sq, &tr, &mapseq);
}
totseq = mapseq;
/* Read digital sequences into an array (possibly concat'ed onto mapped seqs)
*/
status = esl_sqfile_OpenDigital(abc, seqfile, infmt, NULL, &sqfp);
if (status == eslENOTFOUND) p7_Fail("Failed to open sequence file %s for reading\n", seqfile);
else if (status == eslEFORMAT) p7_Fail("Sequence file %s is empty or misformatted\n", seqfile);
else if (status != eslOK) p7_Fail("Unexpected error %d opening sequence file %s\n", status, seqfile);
ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq + 1));
sq[totseq] = esl_sq_CreateDigital(abc);
nseq = 0;
while ((status = esl_sqio_Read(sqfp, sq[totseq+nseq])) == eslOK)
{
nseq++;
ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq+nseq+1));
sq[totseq+nseq] = esl_sq_CreateDigital(abc);
}
if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s):\n%s\n",
sqfp->filename, esl_sqfile_GetErrorBuf(sqfp));
else if (status != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s", status, sqfp->filename);
esl_sqfile_Close(sqfp);
totseq += nseq;
/* Remaining initializations, including trace array allocation
*/
ESL_RALLOC(tr, p, sizeof(P7_TRACE *) * totseq);
for (idx = mapseq; idx < totseq; idx++)
tr[idx] = p7_trace_CreateWithPP();
p7_tracealign_computeTraces(hmm, sq, mapseq, totseq - mapseq, tr);
p7_tracealign_Seqs(sq, tr, totseq, hmm->M, msaopts, hmm, &msa);
eslx_msafile_Write(ofp, msa, outfmt);
for (idx = 0; idx <= totseq; idx++) esl_sq_Destroy(sq[idx]); /* including sq[nseq] because we overallocated */
for (idx = 0; idx < totseq; idx++) p7_trace_Destroy(tr[idx]);
free(sq);
free(tr);
esl_msa_Destroy(msa);
p7_hmm_Destroy(hmm);
if (ofp != stdout) fclose(ofp);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return eslOK;
ERROR:
return status;
}
