def test_mh_similarity_downsample_errors(track_abundance):
# test downsample=False (default) argument to MinHash.similarity
# max_hash = 50
a = MinHash(0, 20, max_hash=50, track_abundance=track_abundance)
# max_hash = 100
b = MinHash(0, 20, max_hash=100, track_abundance=track_abundance)
a_values = {1: 5, 3: 3, 5: 2, 8: 2}
b_values = {1: 3, 3: 2, 5: 1, 6: 1, 8: 1, 10: 1}
if track_abundance:
a.set_abundances(a_values)
b.set_abundances(b_values)
else:
a.add_many(a_values.keys())
b.add_many(b_values.keys())
# error, incompatible max hash
with pytest.raises(ValueError) as e:
a.similarity(b, ignore_abundance=True) # downsample=False
assert 'mismatch in scaled; comparison fail' in str(e.value)
with pytest.raises(ValueError) as e:
a.similarity(b, ignore_abundance=False) # downsample=False
assert 'mismatch in scaled; comparison fail' in str(e.value)
with pytest.raises(ValueError) as e:
b.similarity(a, ignore_abundance=True) # downsample=False
assert 'mismatch in scaled; comparison fail' in str(e.value)
with pytest.raises(ValueError) as e:
b.similarity(a, ignore_abundance=False) # downsample=false
assert 'mismatch in scaled; comparison fail' in str(e.value)
def test_mh_similarity_downsample_true(track_abundance):
# verify sim(a, b) == sim(b, a), with and without ignore_abundance
# max_hash = 50
a = MinHash(0, 20, max_hash=50, track_abundance=track_abundance)
# max_hash = 100
b = MinHash(0, 20, max_hash=100, track_abundance=track_abundance)
a_values = {1: 5, 3: 3, 5: 2, 8: 2}
b_values = {1: 3, 3: 2, 5: 1, 6: 1, 8: 1, 10: 1}
if track_abundance:
a.set_abundances(a_values)
b.set_abundances(b_values)
else:
a.add_many(a_values.keys())
b.add_many(b_values.keys())
# downsample=True => no error; values should match either way
x = a.similarity(b, ignore_abundance=True, downsample=True)
y = b.similarity(a, ignore_abundance=True, downsample=True)
assert x == y
# downsample=True => no error; values should match either way
x = a.similarity(b, ignore_abundance=False, downsample=True)
y = b.similarity(a, ignore_abundance=False, downsample=True)
assert x == y
def test_mh_jaccard_similarity():
# check actual Jaccard value for a non-trivial case
a = MinHash(0, 20, max_hash=50, track_abundance=False)
b = MinHash(0, 20, max_hash=50, track_abundance=False)
a.add_many([1, 3, 5, 8])
b.add_many([1, 3, 5, 6, 8, 10])
assert a.similarity(b) == 4. / 6.
def test_mh_similarity_downsample_jaccard_value():
# check jaccard value after downsampling
# max_hash = 50
a = MinHash(0, 20, max_hash=50, track_abundance=False)
# max_hash = 100
b = MinHash(0, 20, max_hash=100, track_abundance=False)
a.add_many([1, 3, 5, 8, 70])
b.add_many([1, 3, 5, 6, 8, 10, 70])
# the hash=70 will be truncated by downsampling
assert a.similarity(b, downsample=True) == 4. / 6.
def test_add_many(track_abundance):
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
b = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
a.add_many(list(range(0, 100, 2)))
a.add_many(list(range(0, 100, 2)))
assert len(a) == 50
assert all(c % 2 == 0 for c in a.get_mins())
for h in range(0, 100, 2):
b.add_hash(h)
b.add_hash(h)
assert len(b) == 50
assert a == b
def test_add_many(track_abundance):
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
b = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
a.add_many(list(range(0, 100, 2)))
a.add_many(list(range(0, 100, 2)))
assert len(a) == 50
assert all(c % 2 == 0 for c in a.get_mins())
for h in range(0, 100, 2):
b.add_hash(h)
b.add_hash(h)
assert len(b) == 50
assert a == b
def test_remove_many(track_abundance):
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
a.add_many(list(range(0, 100, 2)))
orig_sig = signature.SourmashSignature(a)
orig_md5 = orig_sig.md5sum()
a.remove_many(list(range(0, 100, 3)))
new_sig = signature.SourmashSignature(a)
new_md5 = new_sig.md5sum()
assert orig_md5 == "f1cc295157374f5c07cfca5f867188a1"
assert new_md5 == "dd93fa319ef57f4a019c59ee1a8c73e2"
assert orig_md5 != new_md5
assert len(a) == 33
assert all(c % 6 != 0 for c in a.get_mins())
def test_remove_many(track_abundance):
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=5000)
a.add_many(list(range(0, 100, 2)))
orig_sig = signature.SourmashSignature(a)
orig_md5 = orig_sig.md5sum()
a.remove_many(list(range(0, 100, 3)))
new_sig = signature.SourmashSignature(a)
new_md5 = new_sig.md5sum()
assert orig_md5 == "f1cc295157374f5c07cfca5f867188a1"
assert new_md5 == "dd93fa319ef57f4a019c59ee1a8c73e2"
assert orig_md5 != new_md5
assert len(a) == 33
assert all(c % 6 != 0 for c in a.get_mins())