def test_bytes_protein(track_abundance):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 6, True, track_abundance=track_abundance)
mh.add_protein('AGYYG')
mh.add_protein(u'AGYYG')
mh.add_protein(b'AGYYG')
assert len(mh.get_mins()) == 4
def test_protein_hp(track_abundance, hp):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 6, True, dayhoff=False, hp=hp, track_abundance=track_abundance)
mh.add_protein('AGYYG')
if hp:
assert len(mh.get_mins()) == 1
else:
assert len(mh.get_mins()) == 4
def test_protein_dayhoff(track_abundance, dayhoff):
# verify that we can hash protein/aa sequences
mh = MinHash(10,
6,
True,
dayhoff=dayhoff,
hp=False,
track_abundance=track_abundance)
mh.add_protein('AGYYG')
assert len(mh.get_mins()) == 4
def test_bytes_protein_dayhoff(track_abundance, dayhoff):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 6, True, dayhoff=dayhoff, hp=False,
track_abundance=track_abundance)
expected_moltype = 'protein'
if dayhoff:
expected_moltype = 'dayhoff'
assert mh.moltype == expected_moltype
mh.add_protein('AGYYG')
mh.add_protein('AGYYG')
mh.add_protein(b'AGYYG')
assert len(mh.get_mins()) == 4
def test_bytes_protein_hp(track_abundance, hp):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 6, True, dayhoff=False, hp=hp, track_abundance=track_abundance)
expected_moltype = 'protein'
if hp:
expected_moltype = 'hp'
assert mh.moltype == expected_moltype
mh.add_protein('AGYYG')
mh.add_protein(u'AGYYG')
mh.add_protein(b'AGYYG')
if hp:
assert len(mh.get_mins()) == 1
else:
assert len(mh.get_mins()) == 4
def test_basic_dna_bad_2(track_abundance):
# test behavior on bad DNA
mh = MinHash(1, 6, track_abundance=track_abundance)
with pytest.raises(ValueError):
mh.add_protein('YYYY')
def test_protein_short(track_abundance):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 9, True, track_abundance=track_abundance)
mh.add_protein('AG')
assert len(mh.get_mins()) == 0, mh.get_mins()
def test_basic_dna_bad_2(track_abundance):
# test behavior on bad DNA
mh = MinHash(1, 6, track_abundance=track_abundance)
with pytest.raises(ValueError):
mh.add_protein('YYYY')
def test_protein_short(track_abundance):
# verify that we can hash protein/aa sequences
mh = MinHash(10, 9, True, track_abundance=track_abundance)
mh.add_protein('AG')
assert len(mh.get_mins()) == 0, mh.get_mins()
