/* 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;
}
/* msa_shuffling()
*
* 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;
if ( (status = esl_msafile_Open(NULL, msafile, NULL, infmt, NULL, &afp)) != eslOK)
esl_msafile_OpenFailure(afp, status);
while ((status = esl_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) esl_msafile_ReadFailure(afp, 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_msafile_Write(ofp, shuf, afp->format);
}
esl_msa_Destroy(shuf);
esl_msa_Destroy(msa);
}
esl_msafile_Close(afp);
return eslOK;
}
/* utest_fragments()
* This exercises the building code that deals with fragments,
* creating traces with B->X->{MDI}k and {MDI}k->X->E
* transitions, and making sure we can make MSAs correctly
* from them using p7_tracealign_MSA(). This code was initially
* buggy when first written; bugs first detected by Elena,
* Nov 2009
*/
static void
utest_fragments(void)
{
char *failmsg = "failure in build.c::utest_fragments() unit test";
char msafile[16] = "p7tmpXXXXXX"; /* tmpfile name template */
FILE *ofp = NULL;
ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO);
ESL_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
ESL_MSA *dmsa = NULL;
ESL_MSA *postmsa = NULL;
P7_HMM *hmm = NULL;
P7_TRACE **trarr = NULL;
int i;
/* Write an MSA that tests fragment/missing data transitions.
* When built with Handmodelmaker (using the RF line):
* seq1 forces B->X->Mk and Mk->X->E missing data transitions;
* seq2 forces B->X->Ik and Ik->X->E missing data transitions;
* seq3 forces B->X->Dk and Dk->X->E missing data transitions.
*
* The first two cases can arise from fragment definition in
* model construction, or in an input file.
*
* The X->Dk and Dk->X cases should never happen, but we don't
* prohibit them. They can only arise in an input file, because
* esl_msa_MarkFragments() converts everything before/after
* first/last residue to ~, and won't leave a gap character in
* between.
*
* There's nothing being tested by seq4 and seq5; they're just there.
*/
if (esl_tmpfile_named(msafile, &ofp) != eslOK) esl_fatal(failmsg);
fprintf(ofp, "# STOCKHOLM 1.0\n");
fprintf(ofp, "#=GC RF xxxxx.xxxxxxxxxxxx.xxx\n");
fprintf(ofp, "seq1 ~~~~~~GHIKLMNPQRST~~~~\n");
fprintf(ofp, "seq2 ~~~~~aGHIKLMNPQRSTa~~~\n");
fprintf(ofp, "seq3 ~~~~~~~HIKLMNPQRS~~~~~\n");
fprintf(ofp, "seq4 ACDEF.GHIKLMNPQRST.VWY\n");
fprintf(ofp, "seq5 ACDEF.GHIKLMNPQRST.VWY\n");
fprintf(ofp, "//\n");
fclose(ofp);
/* Read the original as text for comparison to postmsa. Make a digital copy for construction */
if (esl_msafile_Open(NULL, msafile, NULL, eslMSAFILE_UNKNOWN, NULL, &afp)!= eslOK) esl_fatal(failmsg);
if (esl_msafile_Read(afp, &msa) != eslOK) esl_fatal(failmsg);
if ((dmsa = esl_msa_Clone(msa)) == NULL) esl_fatal(failmsg);
if (esl_msa_Digitize(abc, dmsa, NULL) != eslOK) esl_fatal(failmsg);
if (p7_Handmodelmaker(dmsa, NULL, &hmm, &trarr) != eslOK) esl_fatal(failmsg);
for (i = 0; i < dmsa->nseq; i++)
if (p7_trace_Validate(trarr[i], abc, dmsa->ax[i], NULL) != eslOK) esl_fatal(failmsg);
/* The example is contrived such that the traces should give exactly the
* same (text) alignment as the input alignment; no tracedoctoring.
* Not a trivial test; for example, sequence 2 has a B->X->I transition that
* can be problematic to handle.
*/
if (p7_tracealign_MSA(dmsa, trarr, hmm->M, p7_DEFAULT, &postmsa) != eslOK) esl_fatal(failmsg);
for (i = 0; i < msa->nseq; i++)
if (strcmp(msa->aseq[i], postmsa->aseq[i]) != 0) esl_fatal(failmsg);
p7_trace_DestroyArray(trarr, msa->nseq);
p7_hmm_Destroy(hmm);
esl_msa_Destroy(msa);
esl_msa_Destroy(dmsa);
esl_msa_Destroy(postmsa);
esl_msafile_Close(afp);
esl_alphabet_Destroy(abc);
remove(msafile);
return;
}
