def setUpClass(cls):
cls.alphabet = Alphabet.amino()
with SequenceFile(
pkg_resources.resource_filename(
"tests", "data/seqs/938293.PRJEB85.HG003687.faa")) as f:
f.set_digital(cls.alphabet)
cls.references = list(f)
def test_pksi(self):
alphabet = Alphabet.amino()
path = pkg_resources.resource_filename(__name__, "data/seqs/PKSI.faa")
with SequenceFile(path) as seqs_file:
seqs_file.set_digital(alphabet)
seqs = list(seqs_file)
hits = next(pyhmmer.phmmer(seqs[-1:], seqs, cpus=1))
hits.sort()
with self.table("A0A089QRB9.tbl") as table:
lines = iter(filter(lambda line: not line.startswith("#"), table))
it = ((hit, domain) for hit in hits for domain in hit.domains)
for line, (hit, domain) in itertools.zip_longest(lines, it):
fields = list(filter(None, line.strip().split(" ")))
self.assertIsNot(line, None)
self.assertIsNot(hit, None)
self.assertEqual(hit.name.decode(), fields[0])
self.assertAlmostEqual(hit.score, float(fields[7]), delta=0.1)
self.assertAlmostEqual(hit.bias, float(fields[8]), delta=0.1)
self.assertAlmostEqual(hit.evalue, float(fields[6]), delta=0.1)
self.assertAlmostEqual(domain.i_evalue,
float(fields[12]),
delta=0.1)
self.assertAlmostEqual(domain.score,
float(fields[13]),
delta=0.1)
self.assertEqual(domain.alignment.hmm_from, int(fields[15]))
self.assertEqual(domain.alignment.hmm_to, int(fields[16]))
self.assertEqual(domain.alignment.target_from, int(fields[17]))
self.assertEqual(domain.alignment.target_to, int(fields[18]))
self.assertEqual(domain.env_from, int(fields[19]))
self.assertEqual(domain.env_to, int(fields[20]))
def test_consensus(self):
# if not set, HMM is None
abc = Alphabet.amino()
hmm = HMM(100, abc)
self.assertIs(hmm.consensus, None)
# if the HMM is fully configured, the consensus should be as
# long as the number of nodes
self.assertEqual(len(self.hmm.consensus), self.hmm.M)
def test_no_queries(self):
alphabet = Alphabet.amino()
path = pkg_resources.resource_filename(__name__, "data/seqs/PKSI.faa")
with SequenceFile(path) as seqs_file:
seqs_file.set_digital(alphabet)
seqs = list(seqs_file)
hits = pyhmmer.phmmer([], seqs, cpus=1)
self.assertIs(None, next(hits, None))
def test_background_error(self):
# check that errors occuring in worker threads are recovered and raised
# in the main threads (a common error is mismatching the HMM and the
# sequence alphabets).
seqs = [TextSequence().digitize(Alphabet.dna())]
with self.hmm_file("PF02826") as hmm_file:
hmm = next(hmm_file)
self.assertRaises(ValueError, self.get_hits, hmm, seqs)
def test_build_protein(self):
abc = Alphabet.amino()
builder = Builder(alphabet=abc)
seq = self.proteins[1].digitize(abc)
hmm, profile, opt = builder.build(seq, Background(abc))
self.assertEqual(hmm.name, seq.name)
self.assertEqual(hmm.alphabet, abc)
self.assertEqual(profile.alphabet, abc)
self.assertEqual(opt.alphabet, abc)
def test_build_dna(self):
abc = Alphabet.dna()
builder = Builder(alphabet=abc)
hmm, profile, opt = builder.build(self.dna.digitize(abc),
Background(abc))
self.assertEqual(hmm.name, self.dna.name)
self.assertEqual(hmm.alphabet, abc)
self.assertEqual(profile.alphabet, abc)
self.assertEqual(opt.alphabet, abc)
self.assertEqual(hmm.M, self.ecori.M)
def test_alphabet_mismatch(self):
pipeline = Pipeline(alphabet=Alphabet.dna())
# mismatch between pipeline alphabet and query alphabet
dsq = TextSequence(sequence="IRGIY").digitize(self.alphabet)
self.assertRaises(AlphabetMismatch, pipeline.scan_seq, dsq, self.hmms)
# mismatch between pipeline alphabet and database alphabet
dsq = TextSequence(sequence="ATGC").digitize(pipeline.alphabet)
self.assertRaises(AlphabetMismatch, pipeline.scan_seq, dsq, self.hmms)
def test_accession(self):
abc = Alphabet.amino()
hmm = HMM(100, abc)
self.assertIs(hmm.accession, None)
hmm.accession = b"TST001"
self.assertEqual(hmm.accession, b"TST001")
hmm.accession = b""
self.assertEqual(hmm.accession, b"")
hmm.accession = None
self.assertEqual(hmm.accession, None)
def test_name(self):
abc = Alphabet.amino()
hmm = HMM(100, abc)
self.assertIs(hmm.name, None)
hmm.name = b"Test"
self.assertEqual(hmm.name, b"Test")
hmm.name = b""
self.assertEqual(hmm.name, b"")
hmm.name = None
self.assertEqual(hmm.name, None)
def test_build_msa_dna(self):
abc = Alphabet.dna()
builder = Builder(alphabet=abc)
with MSAFile(os.path.join(self.testdata, "3box.sto")) as msa_file:
msa_file.set_digital(abc)
msa = next(msa_file)
msa.name = b"3box"
with HMMFile(os.path.join(self.testdata, "3box.hmm")) as hmm_file:
hmm_exp = next(hmm_file)
hmm, profile, opt = builder.build_msa(msa, Background(abc))
self.assertEqual(hmm.name, b"3box")
self.assertEqual(hmm.M, hmm_exp.M)
def test_eq(self):
abc = Alphabet.amino()
self.assertEqual(self.hmm, self.hmm)
self.assertEqual(self.hmm, self.hmm.copy())
self.assertNotEqual(self.hmm, HMM(100, abc))
self.assertNotEqual(self.hmm, 1)
def test_copy(self):
abc = Alphabet.amino()
builder = Builder(alphabet=abc)
copy = builder.copy()
self.assertEqual(builder.alphabet, copy.alphabet)
self.assertEqual(builder.seed, copy.seed)