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;
}
static void
onefetch_subseq(ESL_GETOPTS *go, FILE *ofp, ESL_SQFILE *sqfp, char *newname, char *key, uint32_t given_start, uint32_t given_end)
{
int start, end;
int do_revcomp;
ESL_SQ *sq = esl_sq_Create();
if (sqfp->data.ascii.ssi == NULL) esl_fatal("no ssi index");
/* reverse complement indicated by coords. */
/* -c 52: would be 52,0, so watch out for given_end = 0 case */
if (given_end != 0 && given_start > given_end)
{ start = given_end; end = given_start; do_revcomp = TRUE; }
else
{ start = given_start; end = given_end; do_revcomp = FALSE; }
if (esl_sqio_FetchSubseq(sqfp, key, start, end, sq) != eslOK) esl_fatal(esl_sqfile_GetErrorBuf(sqfp));
if (newname != NULL) esl_sq_SetName(sq, newname);
else esl_sq_FormatName(sq, "%s/%d-%d", key, given_start, (given_end == 0) ? sq->L : given_end);
/* Two ways we might have been asked to revcomp: by coord, or by -r option */
/* (If both happen, they'll cancel each other out) */
if (do_revcomp)
if (esl_sq_ReverseComplement(sq) != eslOK) esl_fatal("Failed to reverse complement %s; is it a protein?\n", sq->name);
if (esl_opt_GetBoolean(go, "-r"))
if (esl_sq_ReverseComplement(sq) != eslOK) esl_fatal("Failed to reverse complement %s; is it a protein?\n", sq->name);
esl_sqio_Write(ofp, sq, eslSQFILE_FASTA, FALSE);
esl_sq_Destroy(sq);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = p7_CreateDefaultApp(options, 1, argc, argv, banner, usage);
ESL_RANDOMNESS *rng = esl_randomness_CreateFast(esl_opt_GetInteger(go, "-s"));
char *hmmfile = esl_opt_GetArg(go, 1);
int L = esl_opt_GetInteger(go, "-L");
int N = esl_opt_GetInteger(go, "-N");
ESL_ALPHABET *abc = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
P7_TRACE *tr = p7_trace_Create();
ESL_SQ *sq = NULL;
char errbuf[eslERRBUFSIZE];
int i;
int status;
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);
p7_hmmfile_Close(hfp);
bg = p7_bg_Create(abc); p7_bg_SetLength(bg, L);
gm = p7_profile_Create(hmm->M, abc); p7_ProfileConfig(hmm, bg, gm, L, p7_LOCAL);
sq = esl_sq_CreateDigital(abc);
for (i = 0; i < N; i++)
{
p7_ProfileEmit(rng, hmm, gm, bg, sq, tr);
esl_sq_FormatName(sq, "%s-sample%d", hmm->name, i);
esl_sqio_Write(stdout, sq, eslSQFILE_FASTA, FALSE);
if (p7_trace_Validate(tr, abc, sq->dsq, errbuf) != eslOK) esl_fatal(errbuf);
esl_sq_Reuse(sq);
p7_trace_Reuse(tr);
}
esl_sq_Destroy(sq);
p7_trace_Destroy(tr);
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(hmm);
esl_alphabet_Destroy(abc);
esl_randomness_Destroy(rng);
esl_getopts_Destroy(go);
return 0;
}
static void
emit_sequences(ESL_GETOPTS *go, FILE *ofp, int outfmt, ESL_RANDOMNESS *r, P7_HMM *hmm)
{
ESL_SQ *sq = NULL;
P7_TRACE *tr = NULL;
P7_BG *bg = NULL;
P7_PROFILE *gm = NULL;
int do_profile = esl_opt_GetBoolean(go, "-p");
int N = esl_opt_GetInteger(go, "-N");
int L = esl_opt_GetInteger(go, "-L");
int mode = p7_LOCAL;
int nseq;
int status;
if (esl_opt_GetBoolean(go, "--local")) mode = p7_LOCAL;
else if (esl_opt_GetBoolean(go, "--unilocal")) mode = p7_UNILOCAL;
else if (esl_opt_GetBoolean(go, "--glocal")) mode = p7_GLOCAL;
else if (esl_opt_GetBoolean(go, "--uniglocal")) mode = p7_UNIGLOCAL;
if ((sq = esl_sq_CreateDigital(hmm->abc)) == NULL) esl_fatal("failed to allocate sequence");
if ((tr = p7_trace_Create()) == NULL) esl_fatal("failed to allocate trace");
if ((bg = p7_bg_Create(hmm->abc)) == NULL) esl_fatal("failed to create null model");
if ((gm = p7_profile_Create(hmm->M, hmm->abc)) == NULL) esl_fatal("failed to create profile");
if (p7_ProfileConfig(hmm, bg, gm, L, mode) != eslOK) esl_fatal("failed to configure profile");
if (p7_bg_SetLength(bg, L) != eslOK) esl_fatal("failed to reconfig null model length");
if (p7_hmm_Validate (hmm, NULL, 0.0001) != eslOK) esl_fatal("whoops, HMM is bad!");
if (p7_profile_Validate(gm, NULL, 0.0001) != eslOK) esl_fatal("whoops, profile is bad!");
for (nseq = 1; nseq <= N; nseq++)
{
if (do_profile) status = p7_ProfileEmit(r, hmm, gm, bg, sq, tr);
else status = p7_CoreEmit (r, hmm, sq, tr);
if (status) esl_fatal("Failed to emit sequence\n");
status = esl_sq_FormatName(sq, "%s-sample%d", hmm->name, nseq);
if (status) esl_fatal("Failed to set sequence name\n");
status = esl_sqio_Write(ofp, sq, outfmt, FALSE);
if (status != eslOK) esl_fatal("Failed to write sequence\n");
p7_trace_Reuse(tr);
esl_sq_Reuse(sq);
}
esl_sq_Destroy(sq);
p7_trace_Destroy(tr);
p7_bg_Destroy(bg);
p7_profile_Destroy(gm);
return;
}
/* seq_generation()
*
* Generating sequences.
*/
static int
seq_generation(ESL_GETOPTS *go, ESL_RANDOMNESS *r, FILE *ofp, int outfmt)
{
ESL_ALPHABET *abc = NULL;
ESL_SQ *sq = NULL;
double *fq = NULL;
int alphatype = eslUNKNOWN; // static checkers can't see that 1 of --rna, --dna, --amino must be true
int N = esl_opt_GetInteger(go, "-N");
int L = esl_opt_GetInteger(go, "-L");
int i;
int status;
if (L <= 0) esl_fatal("To generate sequences, set -L option (length of generated seqs) > 0 ");
if (esl_opt_GetBoolean(go, "--rna")) alphatype = eslRNA;
if (esl_opt_GetBoolean(go, "--dna")) alphatype = eslDNA;
if (esl_opt_GetBoolean(go, "--amino")) alphatype = eslAMINO;
abc = esl_alphabet_Create(alphatype);
sq = esl_sq_CreateDigital(abc);
esl_sq_GrowTo(sq, L);
/* Pick the iid frequency distribution to use */
ESL_ALLOC(fq, sizeof(double) * abc->K);
switch (alphatype) {
case eslRNA:
case eslDNA: esl_vec_DSet(fq, 4, 0.25); break;
case eslAMINO: esl_composition_SW34(fq); break;
default: esl_vec_DSet(fq, abc->K, 1.0 / (double) abc->K); break;
}
/* generate */
for (i = 0; i < N; i++)
{
esl_rsq_xIID(r, fq, abc->K, L, sq->dsq);
if (N > 1) esl_sq_FormatName(sq, "random%d", i);
else esl_sq_SetName(sq, "random");
sq->n = L;
esl_sqio_Write(ofp, sq, outfmt, FALSE);
}
free(fq);
esl_alphabet_Destroy(abc);
esl_sq_Destroy(sq);
return eslOK;
ERROR:
if (fq != NULL) free(fq);
esl_alphabet_Destroy(abc);
esl_sq_Destroy(sq);
return status;
}
static void
emit_consensus(ESL_GETOPTS *go, FILE *ofp, int outfmt, P7_HMM *hmm)
{
ESL_SQ *sq = NULL;
if ((sq = esl_sq_CreateDigital(hmm->abc)) == NULL) esl_fatal("failed to allocate sequence");
if (p7_emit_SimpleConsensus(hmm, sq) != eslOK) esl_fatal("failed to create simple consensus seq");
if (esl_sq_FormatName(sq, "%s-consensus", hmm->name) != eslOK) esl_fatal("failed to set sequence name");
if (esl_sqio_Write(ofp, sq, outfmt, FALSE) != eslOK) esl_fatal("failed to write sequence");
esl_sq_Destroy(sq);
return;
}
static void
emit_fancycons(ESL_GETOPTS *go, FILE *ofp, int outfmt, P7_HMM *hmm)
{
ESL_SQ *sq = NULL;
float minl = esl_opt_GetReal(go, "--minl");
float minu = esl_opt_GetReal(go, "--minu");
if ((sq = esl_sq_Create()) == NULL) esl_fatal("failed to allocate sequence");
if (p7_emit_FancyConsensus(hmm, minl, minu, sq) != eslOK) esl_fatal("failed to create consensus seq");
if (esl_sq_FormatName(sq, "%s-consensus", hmm->name) != eslOK) esl_fatal("failed to set sequence name");
if (esl_sqio_Write(ofp, sq, outfmt, FALSE) != eslOK) esl_fatal("failed to write sequence");
esl_sq_Destroy(sq);
return;
}
static int
synthesize_negatives(ESL_GETOPTS *go, struct cfg_s *cfg, int nneg)
{
ESL_SQ *sq = esl_sq_CreateDigital(cfg->abc);
int a;
int i;
int L1,L2,L3,d1n,d2n;
for (i = 0; i < nneg; i++)
{
/* Select a random test seq, to use its same segments */
a = esl_rnd_Roll(cfg->r, cfg->ntest);
L1 = cfg->test_lens[a].L1;
L2 = cfg->test_lens[a].L2;
L3 = cfg->test_lens[a].L3;
d1n = cfg->test_lens[a].d1n;
d2n = cfg->test_lens[a].d2n;
esl_sq_GrowTo(sq, cfg->test_lens[a].L);
esl_sq_FormatName(sq, "decoy%d", i+1);
esl_sq_FormatDesc(sq, "L=%d in segments: %d/%d/%d/%d/%d", cfg->test_lens[a].L, L1, d1n, L2, d2n, L3);
sq->n = cfg->test_lens[a].L;
fprintf(cfg->negsummfp, "%-15s %5d %5d %5d %5d %5d %5d",
sq->name, (int) sq->n,
L1, d1n, L2, d2n, L3);
sq->dsq[0] = sq->dsq[cfg->test_lens[a].L+1] = eslDSQ_SENTINEL;
set_random_segment(go, cfg, cfg->negsummfp, sq->dsq+1, L1);
set_random_segment(go, cfg, cfg->negsummfp, sq->dsq+1+L1, d1n);
set_random_segment(go, cfg, cfg->negsummfp, sq->dsq+1+L1+d1n, L2);
set_random_segment(go, cfg, cfg->negsummfp, sq->dsq+1+L1+d1n+L2, d2n);
set_random_segment(go, cfg, cfg->negsummfp, sq->dsq+1+L1+d1n+L2+d2n, L3);
fprintf(cfg->negsummfp, "\n");
esl_sqio_Write(cfg->out_seqfp, sq, eslSQFILE_FASTA, FALSE);
esl_sq_Reuse(sq);
}
esl_sq_Destroy(sq);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = p7_CreateDefaultApp(options, 2, argc, argv, banner, usage);
ESL_RANDOMNESS *rng = esl_randomness_CreateFast(esl_opt_GetInteger(go, "-s"));
ESL_ALPHABET *abc = NULL;
char *ghmmfile = esl_opt_GetArg(go, 1); /* HMMs parameterized for sequence generation */
char *ahmmfile = esl_opt_GetArg(go, 2); /* HMMs parameterized for alignment */
int N = esl_opt_GetInteger(go, "-N");
P7_HMMFILE *ghfp = NULL;
P7_HMMFILE *ahfp = NULL;
P7_HMM *ghmm = NULL;
P7_HMM *ahmm = NULL;
P7_PROFILE *ggm = NULL;
P7_PROFILE *agm = NULL;
P7_OPROFILE *aom = NULL;
P7_BG *bg = NULL;
ESL_SQ *sq = NULL;
P7_TRACE *reftr = p7_trace_Create();
P7_TRACE *testtr = p7_trace_Create();
P7_TRACE_METRICS *tmetrics = p7_trace_metrics_Create();
P7_REFMX *rmx = p7_refmx_Create(100,100);
// P7_FILTERMX *ox = NULL;
P7_HARDWARE *hw;
if ((hw = p7_hardware_Create ()) == NULL) p7_Fail("Couldn't get HW information data structure");
P7_SPARSEMASK *sm = p7_sparsemask_Create(100, 100, hw->simd);
P7_SPARSEMX *sxv = p7_sparsemx_Create(NULL);
int idx;
char errbuf[eslERRBUFSIZE];
int status;
p7_Init();
/* open HMM file containing models parameterized for generation (sampling) of seqs */
status = p7_hmmfile_OpenE(ghmmfile, NULL, &ghfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", ghmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", ghmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, ghmmfile, errbuf);
/* open HMM file containing models parameterized for alignment (may be the same as ghmmfile) */
status = p7_hmmfile_OpenE(ahmmfile, NULL, &ahfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", ahmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", ahmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, ahmmfile, errbuf);
while ( (status = p7_hmmfile_Read(ghfp, &abc, &ghmm)) == eslOK) /* <abc> gets set on first read */
{
/* read the counterpart HMM from <ahfp> */
status = p7_hmmfile_Read(ahfp, &abc, &ahmm);
if (status == eslEFORMAT) p7_Fail("Bad file format in HMM file %s:\n%s\n", ahfp->fname, ahfp->errbuf);
else if (status == eslEINCOMPAT) p7_Fail("HMM in %s is not in the expected %s alphabet\n", ahfp->fname, esl_abc_DecodeType(abc->type));
else if (status == eslEOF) p7_Fail("Empty HMM file %s? No HMM data found.\n", ahfp->fname);
else if (status != eslOK) p7_Fail("Unexpected error in reading HMMs from %s\n", ahfp->fname);
/* try to validate that they're the "same" */
if (ahmm->M != ghmm->M || strcmp(ahmm->name, ghmm->name) != 0) p7_Fail("<gen-hmmfile>, <ali-hmmfile> contain different set or order of models");
/* deferred one-time creation of structures that need to know the alphabet */
if (!bg) bg = p7_bg_Create(abc);
if (!sq) sq = esl_sq_CreateDigital(abc);
ggm = p7_profile_Create(ghmm->M, abc);
agm = p7_profile_Create(ahmm->M, abc);
aom = p7_oprofile_Create(ahmm->M, abc, hw->simd);
p7_profile_ConfigCustom(ggm, ghmm, bg, esl_opt_GetInteger(go, "--gL"), esl_opt_GetReal(go, "--gnj"), esl_opt_GetReal(go, "--gpglocal"));
p7_profile_ConfigCustom(agm, ahmm, bg, 100, esl_opt_GetReal(go, "--anj"), esl_opt_GetReal(go, "--apglocal"));
p7_oprofile_Convert(agm, aom);
for (idx = 1; idx <= N; idx++)
{
p7_ProfileEmit(rng, ghmm, ggm, bg, sq, reftr);
if (esl_opt_GetBoolean(go, "--dumpseqs")) {
esl_sq_FormatName(sq, "seq%d", idx);
esl_sqio_Write(stdout, sq, eslSQFILE_FASTA, FALSE);
}
p7_bg_SetLength(bg, sq->n);
p7_profile_SetLength(agm, sq->n);
p7_sparsemask_Reinit(sm, agm->M, sq->n);
p7_sparsemask_AddAll(sm);
if (esl_opt_GetBoolean(go, "--vit")) p7_ReferenceViterbi(sq->dsq, sq->n, agm, rmx, testtr, /*opt_vsc=*/NULL);
else p7_SparseViterbi (sq->dsq, sq->n, agm, sm, sxv, testtr, /*opt_vsc=*/NULL);
p7_trace_metrics(reftr, testtr, tmetrics);
p7_sparsemask_Reuse(sm);
p7_sparsemx_Reuse(sxv);
//p7_filtermx_Reuse(ox);
p7_refmx_Reuse(rmx);
esl_sq_Reuse(sq);
p7_trace_Reuse(reftr);
p7_trace_Reuse(testtr);
}
p7_oprofile_Destroy(aom);
p7_profile_Destroy(ggm);
p7_profile_Destroy(agm);
p7_hmm_Destroy(ghmm);
p7_hmm_Destroy(ahmm);
}
/* we leave the loop with <status> set by a p7_hmmfile_Read() on ghfp; if all is well, status=eslEOF */
if (status == eslEFORMAT) p7_Fail("Bad file format in HMM file %s:\n%s\n", ghfp->fname, ghfp->errbuf);
else if (status == eslEINCOMPAT) p7_Fail("HMM in %s is not in the expected %s alphabet\n", ghfp->fname, esl_abc_DecodeType(abc->type));
else if (status != eslEOF) p7_Fail("Unexpected error in reading HMMs from %s\n", ghfp->fname);
p7_trace_metrics_Dump(stdout, tmetrics);
p7_hmmfile_Close(ghfp);
p7_hmmfile_Close(ahfp);
// p7_filtermx_Destroy(ox);
p7_sparsemask_Destroy(sm);
p7_sparsemx_Destroy(sxv);
p7_refmx_Destroy(rmx);
p7_trace_metrics_Destroy(tmetrics);
p7_trace_Destroy(testtr);
p7_trace_Destroy(reftr);
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
esl_randomness_Destroy(rng);
esl_getopts_Destroy(go);
}
/* Function: p7_alidisplay_Backconvert()
* Synopsis: Convert an alidisplay to a faux trace and subsequence.
* Incept: SRE, Wed Dec 10 09:49:28 2008 [Janelia]
*
* Purpose: Convert alignment display object <ad> to a faux subsequence
* and faux subsequence trace, returning them in <ret_sq> and
* <ret_tr>.
*
* The subsequence <*ret_sq> is digital; ascii residues in
* <ad> are digitized using digital alphabet <abc>.
*
* The subsequence and trace are suitable for passing as
* array elements to <p7_MultipleAlignment>. This is the
* main purpose of backconversion. Results of a profile
* search are stored in a hit list as a processed
* <P7_ALIDISPLAY>, not as a <P7_TRACE> and <ESL_SQ>, to
* reduce space and to reduce communication overhead in
* parallelized search implementations. After reduction
* to a final hit list, a master may want to construct a
* multiple alignment of all the significant hits.
*
* Returns: <eslOK> on success.
*
* Throws: <eslEMEM> on allocation failures. <eslECORRUPT> on unexpected internal
* data corruption. On any exception, <*ret_sq> and <*ret_tr> are
* <NULL>.
*
* Xref: J4/29.
*/
int
p7_alidisplay_Backconvert(const P7_ALIDISPLAY *ad, const ESL_ALPHABET *abc, ESL_SQ **ret_sq, P7_TRACE **ret_tr)
{
ESL_SQ *sq = NULL; /* RETURN: faux subsequence */
P7_TRACE *tr = NULL; /* RETURN: faux trace */
int subL = 0; /* subsequence length in the <ad> */
int a, i, k; /* coords for <ad>, <sq->dsq>, model */
char st; /* state type: MDI */
int status;
/* Make a first pass over <ad> just to calculate subseq length */
for (a = 0; a < ad->N; a++)
if (! esl_abc_CIsGap(abc, ad->aseq[a])) subL++;
/* Allocations */
if ((sq = esl_sq_CreateDigital(abc)) == NULL) { status = eslEMEM; goto ERROR; }
if ((status = esl_sq_GrowTo(sq, subL)) != eslOK) goto ERROR;
if ((tr = (ad->ppline == NULL) ? p7_trace_Create() : p7_trace_CreateWithPP()) == NULL) { status = eslEMEM; goto ERROR; }
if ((status = p7_trace_GrowTo(tr, subL+6)) != eslOK) goto ERROR; /* +6 is for SNB/ECT */
/* Construction of dsq, trace */
sq->dsq[0] = eslDSQ_SENTINEL;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_S, 0, 0) : p7_trace_AppendWithPP(tr, p7T_S, 0, 0, 0.0))) != eslOK) goto ERROR;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_N, 0, 0) : p7_trace_AppendWithPP(tr, p7T_N, 0, 0, 0.0))) != eslOK) goto ERROR;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_B, 0, 0) : p7_trace_AppendWithPP(tr, p7T_B, 0, 0, 0.0))) != eslOK) goto ERROR;
k = ad->hmmfrom;
i = 1;
for (a = 0; a < ad->N; a++)
{
if (esl_abc_CIsResidue(abc, ad->model[a])) { st = (esl_abc_CIsResidue(abc, ad->aseq[a]) ? p7T_M : p7T_D); } else st = p7T_I;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, st, k, i) : p7_trace_AppendWithPP(tr, st, k, i, p7_alidisplay_DecodePostProb(ad->ppline[a])))) != eslOK) goto ERROR;
switch (st) {
case p7T_M: sq->dsq[i] = esl_abc_DigitizeSymbol(abc, ad->aseq[a]); k++; i++; break;
case p7T_I: sq->dsq[i] = esl_abc_DigitizeSymbol(abc, ad->aseq[a]); i++; break;
case p7T_D: k++; break;
}
}
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_E, 0, 0) : p7_trace_AppendWithPP(tr, p7T_E, 0, 0, 0.0))) != eslOK) goto ERROR;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_C, 0, 0) : p7_trace_AppendWithPP(tr, p7T_C, 0, 0, 0.0))) != eslOK) goto ERROR;
if ((status = ((ad->ppline == NULL) ? p7_trace_Append(tr, p7T_T, 0, 0) : p7_trace_AppendWithPP(tr, p7T_T, 0, 0, 0.0))) != eslOK) goto ERROR;
sq->dsq[i] = eslDSQ_SENTINEL;
/* some sanity checks */
if (tr->N != ad->N + 6) ESL_XEXCEPTION(eslECORRUPT, "backconverted trace ended up with unexpected size (%s/%s)", ad->sqname, ad->hmmname);
if (k != ad->hmmto + 1) ESL_XEXCEPTION(eslECORRUPT, "backconverted trace didn't end at expected place on model (%s/%s)", ad->sqname, ad->hmmname);
if (i != subL + 1) ESL_XEXCEPTION(eslECORRUPT, "backconverted subseq didn't end at expected length (%s/%s)", ad->sqname, ad->hmmname);
/* Set up <sq> annotation as a subseq of a source sequence */
if ((status = esl_sq_FormatName(sq, "%s/%ld-%ld", ad->sqname, ad->sqfrom, ad->sqto)) != eslOK) goto ERROR;
if ((status = esl_sq_FormatDesc(sq, "[subseq from] %s", ad->sqdesc[0] != '\0' ? ad->sqdesc : ad->sqname)) != eslOK) goto ERROR;
if ((status = esl_sq_SetSource (sq, ad->sqname)) != eslOK) goto ERROR;
if (ad->sqacc[0] != '\0') { if ((status = esl_sq_SetAccession (sq, ad->sqacc)) != eslOK) goto ERROR; }
sq->n = subL;
sq->start = ad->sqfrom;
sq->end = ad->sqto;
sq->C = 0;
sq->W = subL;
sq->L = ad->L;
tr->M = ad->M;
tr->L = ad->L;
*ret_sq = sq;
*ret_tr = tr;
return eslOK;
ERROR:
if (sq != NULL) esl_sq_Destroy(sq);
if (tr != NULL) p7_trace_Destroy(tr);
*ret_sq = NULL;
*ret_tr = NULL;
return status;
}
/* seq_shuffling()
* SRE, Tue Jan 22 08:35:51 2008 [Market Street Cafe, Leesburg]
*
* Shuffling of input sequences.
*
* Fixed-length (L>0) vs. full-length (L=0) modes handled differently.
* In fixed-length mode:
* <shuff->seq> only needs to be allocated once, for L
* <targ> is an allocated copy of a random subseq of length L
* sequences < L residues long can't be shuffled
* In full-length mode:
* <shuff->seq> is grown to length <sq->n> for each input seq
* <targ> just points to <sq->seq>
*/
static int
seq_shuffling(ESL_GETOPTS *go, ESL_RANDOMNESS *r, FILE *ofp, int outfmt)
{
char *seqfile = esl_opt_GetArg(go, 1);
int infmt = eslSQFILE_UNKNOWN;
ESL_SQFILE *sqfp = NULL;
ESL_SQ *sq = esl_sq_Create();
ESL_SQ *shuff = esl_sq_Create();
char *targ = NULL;
int N = esl_opt_GetInteger(go, "-N");
int L = esl_opt_GetInteger(go, "-L"); /* L>0 means select random fixed-len subseqs */
int kmers = 0;
int i;
int status;
if (esl_opt_GetString(go, "--informat") != NULL) {
infmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslSQFILE_UNKNOWN) esl_fatal("%s is not a valid input sequence file format for --informat");
}
if (esl_opt_IsOn(go, "-k")) kmers = esl_opt_GetInteger(go, "-k");
status = esl_sqfile_Open(seqfile, infmt, NULL, &sqfp);
if (status == eslENOTFOUND) esl_fatal("No such file %s", seqfile);
else if (status == eslEFORMAT) esl_fatal("Format of seqfile %s unrecognized.", seqfile);
else if (status == eslEINVAL) esl_fatal("Can't autodetect stdin or .gz.");
else if (status != eslOK) esl_fatal("Open failed, code %d.", status);
if (L>0) {
esl_sq_GrowTo(shuff, L);
shuff->n = L;
ESL_ALLOC(targ, sizeof(char) * (L+1));
}
while ((status = esl_sqio_Read(sqfp, sq)) == eslOK)
{
if (L == 0) { /* shuffling entire sequence */
esl_sq_GrowTo(shuff, sq->n); /* make sure shuff can hold sq */
shuff->n = sq->n;
targ = sq->seq;
} else {
if (sq->n < L) continue; /* reject seqs < L long */
}
for (i = 0; i < N; i++)
{
if (L > 0) { /* fixed-len mode: copy a random subseq */
int pos = esl_rnd_Roll(r, sq->n - L + 1);
strncpy(targ, sq->seq + pos, L);
targ[L] = '\0';
}
/* Do the requested kind of shuffling */
if (esl_opt_GetBoolean(go, "-m")) esl_rsq_CShuffle (r, targ, shuff->seq); /* monoresidue shuffling */
else if (esl_opt_GetBoolean(go, "-d")) esl_rsq_CShuffleDP (r, targ, shuff->seq); /* diresidue shuffling */
else if (esl_opt_IsOn (go, "-k")) esl_rsq_CShuffleKmers(r, targ, kmers, shuff->seq); /* diresidue shuffling */
else if (esl_opt_GetBoolean(go, "-0")) esl_rsq_CMarkov0 (r, targ, shuff->seq); /* 0th order Markov */
else if (esl_opt_GetBoolean(go, "-1")) esl_rsq_CMarkov1 (r, targ, shuff->seq); /* 1st order Markov */
else if (esl_opt_GetBoolean(go, "-r")) esl_rsq_CReverse ( targ, shuff->seq); /* reverse */
else if (esl_opt_IsOn (go, "-w")) { /* regionally shuffle */
int W= esl_opt_GetInteger(go, "-w"); esl_rsq_CShuffleWindows(r, targ, W, shuff->seq);
}
/* Set the name of the shuffled sequence */
if (N > 1) esl_sq_FormatName(shuff, "%s-shuffled-%d", sq->name, i);
else esl_sq_FormatName(shuff, "%s-shuffled", sq->name);
/* Output the resulting sequence */
esl_sqio_Write(ofp, shuff, outfmt, FALSE);
/* don't need to reuse the shuffled sequence: we will use exactly the same memory */
}
esl_sq_Reuse(sq);
}
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);
if (L>0) free(targ);
esl_sq_Destroy(shuff);
esl_sq_Destroy(sq);
esl_sqfile_Close(sqfp);
return eslOK;
ERROR:
if (targ != NULL) free(targ);
esl_sq_Destroy(shuff);
esl_sq_Destroy(sq);
esl_sqfile_Close(sqfp);
return status;
}
/* Each test sequence will contain one or two domains, depending on whether --single is set.
*/
static int
synthesize_positives(ESL_GETOPTS *go, struct cfg_s *cfg, char *testname, ESL_STACK *teststack, int *ret_ntest)
{
ESL_SQ *domain1, *domain2;
ESL_SQ *sq;
void *p;
int64_t L; /* total length of synthetic test seq */
int d1n, d2n; /* lengths of two domains */
int L1,L2,L3; /* lengths of three random regions */
int i,j;
int ntest = 0;
int ndomains = ( (esl_opt_GetBoolean(go, "--single") == TRUE) ? 1 : 2);
int status;
while (esl_stack_ObjectCount(teststack) >= ndomains)
{
ESL_RALLOC(cfg->test_lens, p, (cfg->ntest+1) * sizeof(struct testseq_s));
/* Pop our one or two test domains off the stack */
esl_stack_PPop(teststack, &p);
domain1 = p;
d1n = domain1->n;
if (ndomains == 2)
{
esl_stack_PPop(teststack, &p);
domain2 = p;
d2n = domain2->n;
}
else
{
domain2 = NULL;
d2n = 0;
}
/* Select a random total sequence length */
if (d1n+d2n > cfg->db_maxL) esl_fatal("can't construct test seq; no db seq >= %d residues\n", d1n+d2n);
do {
if (esl_ssi_FindNumber(cfg->dbfp->data.ascii.ssi, esl_rnd_Roll(cfg->r, cfg->db_nseq), NULL, NULL, NULL, &L, NULL) != eslOK)
esl_fatal("failed to look up a random seq");
} while (L < d1n+d2n);
/* Now figure out the embedding */
if (ndomains == 2)
{
/* Select random lengths of three flanking domains;
* Imagine picking two "insert after" points i,j in sequence 1..L', for
* L' = L-d1n-d2n (the total length of nonhomologous test seq)
*/
do {
i = esl_rnd_Roll(cfg->r, L - d1n - d2n + 1 ); /* i = 0..L' */
j = esl_rnd_Roll(cfg->r, L - d1n - d2n + 1 ); /* j = 0..L' */
} while (i > j);
/* now 1 .. i = random region 1 (if i==0, there's none);
* i+1 .. i+d1n = domain 1
* i+d1n+1 .. j+d1n = random region 2 (if i==j, there's none);
* j+d1n+1 .. j+d1n+d2n = domain 2
* j+d1n+d2n+1 .. L = random region 3 (if j == L-d1n-d2n, there's none);
*/
L1 = i;
L2 = j-i;
L3 = L - d1n - d2n - j;
}
else
{ /* embedding one domain */
i = esl_rnd_Roll(cfg->r, L - d1n + 1 ); /* i = 0..L' */
/* now 1 .. i = random region 1 (if i==0, there's none);
* i+1 .. i+d1n = domain 1
* i+d1n+1 .. L = random region 2 (if i==j, there's none);
*/
L1 = i;
L2 = L - d1n - L1;
L3 = 0;
}
sq = esl_sq_CreateDigital(cfg->abc);
esl_sq_GrowTo(sq, L);
sq->n = L;
if (ndomains == 2)
{
esl_sq_FormatName(sq, "%s/%d/%d-%d/%d-%d", testname, cfg->ntest, i+1, i+d1n, j+d1n+1, j+d1n+d2n);
esl_sq_FormatDesc(sq, "domains: %s %s", domain1->name, domain2->name);
}
else
{
esl_sq_FormatName(sq, "%s/%d/%d-%d", testname, cfg->ntest, i+1, i+d1n);
esl_sq_FormatDesc(sq, "domain: %s", domain1->name);
}
fprintf(cfg->possummfp, "%-35s %5d %5d %5d %5d %5d %5d", sq->name, (int) sq->n, L1, d1n, L2, d2n, L3);
sq->dsq[0] = sq->dsq[L+1] = eslDSQ_SENTINEL;
set_random_segment(go, cfg, cfg->possummfp, sq->dsq+1, L1);
memcpy(sq->dsq+i+1, domain1->dsq+1, sizeof(ESL_DSQ) * d1n);
fprintf(cfg->possummfp, " %-24s %5d %5d", domain1->name, 1, d1n);
set_random_segment(go, cfg, cfg->possummfp, sq->dsq+i+d1n+1, L2);
if (ndomains == 2)
{
memcpy(sq->dsq+j+d1n+1, domain2->dsq+1, sizeof(ESL_DSQ) * d2n);
fprintf(cfg->possummfp, " %-24s %5d %5d", domain2->name, 1, d2n);
set_random_segment(go, cfg, cfg->possummfp, sq->dsq+j+d1n+d2n+1, L3);
}
fprintf(cfg->possummfp, "\n");
cfg->test_lens[cfg->ntest].L = L;
cfg->test_lens[cfg->ntest].L1 = L1;
cfg->test_lens[cfg->ntest].d1n = d1n;
cfg->test_lens[cfg->ntest].L2 = L2;
cfg->test_lens[cfg->ntest].d2n = d2n;
cfg->test_lens[cfg->ntest].L3 = L3;
cfg->ntest++;
ntest++;
esl_sqio_Write(cfg->out_seqfp, sq, eslSQFILE_FASTA, FALSE);
esl_sq_Destroy(domain1);
if (ndomains == 2) esl_sq_Destroy(domain2);
esl_sq_Destroy(sq);
}
*ret_ntest = ntest;
return eslOK;
ERROR:
esl_fatal("Failure in synthesize_positives");
return status;
}