static void
read_test_msas_text(char *a2mfile, char *stkfile)
{
char msg[] = "A2M msa text-mode read unit test failed";
ESL_MSAFILE *afp1 = NULL;
ESL_MSAFILE *afp2 = NULL;
ESL_MSA *msa1, *msa2, *msa3, *msa4;
FILE *a2mfp, *stkfp;
char a2mfile2[32] = "esltmpa2m2XXXXXX";
char stkfile2[32] = "esltmpstk2XXXXXX";
/* vvvv-- everything's the same as the digital utest except these NULLs */
if ( esl_msafile_Open(NULL, a2mfile, NULL, eslMSAFILE_A2M, NULL, &afp1) != eslOK) esl_fatal(msg);
if ( esl_msafile_Open(NULL, stkfile, NULL, eslMSAFILE_STOCKHOLM, NULL, &afp2) != eslOK) esl_fatal(msg);
if ( esl_msafile_a2m_Read (afp1, &msa1) != eslOK) esl_fatal(msg);
if ( esl_msafile_stockholm_Read(afp2, &msa2) != eslOK) esl_fatal(msg);
if ( esl_msa_Compare(msa1, msa2) != eslOK) esl_fatal(msg);
if ( esl_msafile_a2m_Read (afp1, &msa3) != eslEOF) esl_fatal(msg);
if ( esl_msafile_stockholm_Read(afp2, &msa3) != eslEOF) esl_fatal(msg);
esl_msafile_Close(afp2);
esl_msafile_Close(afp1);
if ( esl_tmpfile_named(a2mfile2, &a2mfp) != eslOK) esl_fatal(msg);
if ( esl_tmpfile_named(stkfile2, &stkfp) != eslOK) esl_fatal(msg);
if ( esl_msafile_a2m_Write (a2mfp, msa2) != eslOK) esl_fatal(msg);
if ( esl_msafile_stockholm_Write(stkfp, msa1, eslMSAFILE_PFAM) != eslOK) esl_fatal(msg);
fclose(a2mfp);
fclose(stkfp);
if ( esl_msafile_Open(NULL, a2mfile2, NULL, eslMSAFILE_A2M, NULL, &afp1) != eslOK) esl_fatal(msg);
if ( esl_msafile_Open(NULL, stkfile2, NULL, eslMSAFILE_STOCKHOLM, NULL, &afp2) != eslOK) esl_fatal(msg);
if ( esl_msafile_a2m_Read (afp1, &msa3) != eslOK) esl_fatal(msg);
if ( esl_msafile_stockholm_Read(afp2, &msa4) != eslOK) esl_fatal(msg);
if ( esl_msa_Compare(msa3, msa4) != eslOK) esl_fatal(msg);
remove(a2mfile2);
remove(stkfile2);
esl_msafile_Close(afp2);
esl_msafile_Close(afp1);
esl_msa_Destroy(msa1);
esl_msa_Destroy(msa2);
esl_msa_Destroy(msa3);
esl_msa_Destroy(msa4);
}
int
main(int argc, char **argv)
{
char msg[] = "PSI-BLAST MSA i/o module test driver failed";
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 0, argc, argv, banner, usage);
char pbfile[32] = "esltmppbXXXXXX";
char stkfile[32] = "esltmpstkXXXXXX";
FILE *pbfp, *stkfp;
int testnumber;
int ngoodtests = 2;
char tmpfile[32];
FILE *ofp;
int expected_alphatype;
int expected_nseq;
int expected_alen;
if ( esl_tmpfile_named(pbfile, &pbfp) != eslOK) esl_fatal(msg);
if ( esl_tmpfile_named(stkfile, &stkfp) != eslOK) esl_fatal(msg);
write_test_msas(pbfp, stkfp);
fclose(pbfp);
fclose(stkfp);
read_test_msas_digital(pbfile, stkfile);
read_test_msas_text (pbfile, stkfile);
/* Various "good" files that should be parsed correctly */
for (testnumber = 1; testnumber <= ngoodtests; testnumber++)
{
strcpy(tmpfile, "esltmpXXXXXX");
if (esl_tmpfile_named(tmpfile, &ofp) != eslOK) esl_fatal(msg);
switch (testnumber) {
case 1: utest_write_good1 (ofp, &expected_alphatype, &expected_nseq, &expected_alen); break;
case 2: utest_write_good2 (ofp, &expected_alphatype, &expected_nseq, &expected_alen); break;
}
fclose(ofp);
utest_goodfile(tmpfile, testnumber, expected_alphatype, expected_nseq, expected_alen);
remove(tmpfile);
}
remove(pbfile);
remove(stkfile);
esl_getopts_Destroy(go);
return 0;
}
static void
generate_testfile(ESL_RANDOMNESS *rng, char *tmpfile, int *is_data, int nlines)
{
char *msg = "esl_recorder:: test file generator failed";
FILE *fp = NULL;
int in_block = esl_rnd_Roll(rng, 2); /* TRUE | FALSE */
int nblock = 1 + esl_rnd_Roll(rng, 10); /* 1..10 */
int i;
if (esl_tmpfile_named(tmpfile, &fp) != eslOK) esl_fatal(msg);
for (i = 0; i < nlines; i++)
{
is_data[i] = in_block ? TRUE : FALSE;
fprintf(fp, "%c%d\n", (in_block ? '#' : ' '), i);
if (--nblock == 0) {
in_block = ! in_block;
nblock = 1 + esl_rnd_Roll(rng, 10); /* 1..10 */
}
}
fclose(fp);
}
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);
}
/* 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;
}
