def test_mh_count_common(track_abundance):
a = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 40, 2):
a.add_hash(i)
b = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 80, 4):
b.add_hash(i)
assert a.count_common(b) == 10
assert b.count_common(a) == 10
def test_mh_count_common(track_abundance):
a = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 40, 2):
a.add_hash(i)
b = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 80, 4):
b.add_hash(i)
assert a.count_common(b) == 10
assert b.count_common(a) == 10
def test_abundance_count_common():
a = MinHash(20, 5, False, track_abundance=True)
b = MinHash(20, 5, False, track_abundance=False)
a.add_sequence('AAAAA')
a.add_sequence('AAAAA')
assert a.get_mins() == [2110480117637990133]
assert a.get_mins(with_abundance=True) == {2110480117637990133: 2}
b.add_sequence('AAAAA')
b.add_sequence('GGGGG')
assert a.count_common(b) == 1
assert a.count_common(b) == b.count_common(a)
assert b.get_mins(with_abundance=True) == [2110480117637990133,
10798773792509008305]
def test_mh_asymmetric(track_abundance):
a = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 40, 2):
a.add_hash(i)
# different size: 10
b = MinHash(10, 10, track_abundance=track_abundance)
for i in range(0, 80, 4):
b.add_hash(i)
assert a.count_common(b) == 10
assert b.count_common(a) == 10
with pytest.raises(TypeError):
a.compare(b)
a = a.downsample_n(10)
assert a.compare(b) == 0.5
assert b.compare(a) == 0.5
def test_mh_asymmetric(track_abundance):
a = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 40, 2):
a.add_hash(i)
# different size: 10
b = MinHash(10, 10, track_abundance=track_abundance)
for i in range(0, 80, 4):
b.add_hash(i)
assert a.count_common(b) == 10
assert b.count_common(a) == 10
with pytest.raises(TypeError):
a.compare(b)
a = a.downsample_n(10)
assert a.compare(b) == 0.5
assert b.compare(a) == 0.5
def test_mh_jaccard_asymmetric_num(track_abundance):
a = MinHash(20, 10, track_abundance=track_abundance)
for i in range(0, 40, 2):
a.add_hash(i)
# different size: 10
b = MinHash(10, 10, track_abundance=track_abundance)
for i in range(0, 80, 4):
b.add_hash(i)
assert a.count_common(b) == 10
assert b.count_common(a) == 10
# with 'jaccard', this will raise an error b/c different num
with pytest.raises(TypeError):
a.jaccard(b)
a = a.downsample_n(10)
# CTB note: this used to be 'compare', is now 'jaccard'
assert a.jaccard(b) == 0.5
assert b.jaccard(a) == 0.5
def test_abundance_simple_2():
a = MinHash(20, 5, False, track_abundance=True)
b = MinHash(20, 5, False, track_abundance=True)
a.add_sequence('AAAAA')
assert a.get_mins() == [2110480117637990133]
assert a.get_mins(with_abundance=True) == {2110480117637990133: 1}
a.add_sequence('AAAAA')
assert a.get_mins() == [2110480117637990133]
assert a.get_mins(with_abundance=True) == {2110480117637990133: 2}
b.add_sequence('AAAAA')
assert a.count_common(b) == 1
def test_abundance_simple_2():
a = MinHash(20, 5, False, track_abundance=True)
b = MinHash(20, 5, False, track_abundance=True)
a.add_sequence('AAAAA')
assert a.get_mins() == [2110480117637990133]
assert a.get_mins(with_abundance=True) == {2110480117637990133: 1}
a.add_sequence('AAAAA')
assert a.get_mins() == [2110480117637990133]
assert a.get_mins(with_abundance=True) == {2110480117637990133: 2}
b.add_sequence('AAAAA')
assert a.count_common(b) == 1
def test_mh_count_common_notmh(track_abundance):
a = MinHash(20, 5, track_abundance=track_abundance)
b = set()
with pytest.raises(TypeError):
a.count_common(b)
def test_mh_count_common_diff_ksize(track_abundance):
a = MinHash(20, 5, track_abundance=track_abundance)
b = MinHash(20, 6, track_abundance=track_abundance)
with pytest.raises(ValueError):
a.count_common(b)
def test_mh_count_common_diff_seed(track_abundance):
a = MinHash(20, 5, False, track_abundance=track_abundance, seed=1)
b = MinHash(20, 5, True, track_abundance=track_abundance, seed=2)
with pytest.raises(ValueError):
a.count_common(b)
def test_mh_count_common_diff_maxhash(track_abundance):
a = MinHash(0, 5, False, track_abundance=track_abundance, max_hash=1)
b = MinHash(0, 5, True, track_abundance=track_abundance, max_hash=2)
with pytest.raises(ValueError):
a.count_common(b)
def test_mh_count_common_diff_ksize(track_abundance):
a = MinHash(20, 5, track_abundance=track_abundance)
b = MinHash(20, 6, track_abundance=track_abundance)
with pytest.raises(ValueError):
a.count_common(b)
def test_mh_count_common_diff_seed(track_abundance):
a = MinHash(20, 5, False, track_abundance=track_abundance, seed=1)
b = MinHash(20, 5, True, track_abundance=track_abundance, seed=2)
with pytest.raises(ValueError):
a.count_common(b)
def test_mh_count_common_diff_maxhash(track_abundance):
a = MinHash(0, 5, False, track_abundance=track_abundance, max_hash=1)
b = MinHash(0, 5, True, track_abundance=track_abundance, max_hash=2)
with pytest.raises(ValueError):
a.count_common(b)
