def test_one_state(self):
# create random data that sums to one row-wise
a_fuzzy = np.random.standard_normal((100, 1))
a_fuzzy = np.exp(a_fuzzy) / np.sum(np.exp(a_fuzzy), 1)[:, None]
# check with both overlap handlings
_fuzzy_to_discrete(a_fuzzy=a_fuzzy)
def test_normal_output(self):
a_fuzzy = np.array([
[0.3, 0.7, 0],
[0.2, 0.5, 0.3],
[0.1, 0.8, 0.1],
[0.4, 0.4, 0.2],
[0.5, 0.3, 0.2],
[0.6, 0.3, 0.1],
[0.3, 0.3, 0.4],
[0.2, 0.2, 0.6],
])
# note: removing the overlap should have no effect in this case since there is none.
# there should also be no critical clusters in this case
a_actual_1, c_1 = _fuzzy_to_discrete(a_fuzzy,
n_most_likely=2,
remove_overlap=True)
a_actual_2, c_2 = _fuzzy_to_discrete(a_fuzzy,
n_most_likely=2,
remove_overlap=False)
a_expected = np.array([
[False, True, False],
[False, False, False],
[False, True, False],
[False, False, False],
[True, False, False],
[True, False, False],
[False, False, True],
[False, False, True],
])
np.testing.assert_array_equal(a_actual_1, a_expected)
np.testing.assert_array_equal(a_actual_2, a_expected)
assert len(c_1) == 0
assert len(c_2) == 0
def _create_states(
self,
probs: Union[np.ndarray, Lineage],
n_cells: int,
check_row_sums: bool = False,
return_not_enough_cells: bool = False,
) -> pd.Series:
if n_cells <= 0:
raise ValueError(
f"Expected `n_cells` to be positive, found `{n_cells}`.")
a_discrete, not_enough_cells = _fuzzy_to_discrete(
a_fuzzy=probs,
n_most_likely=n_cells,
remove_overlap=False,
raise_threshold=0.2,
check_row_sums=check_row_sums,
)
states = _series_from_one_hot_matrix(
membership=a_discrete,
index=self.adata.obs_names,
names=probs.names if isinstance(probs, Lineage) else None,
)
return (states,
not_enough_cells) if return_not_enough_cells else states
def test_passing_lineage_object(self):
a_fuzzy = np.array([
[0.3, 0.7, 0],
[0.2, 0.5, 0.3],
[0.1, 0.8, 0.1],
[0.4, 0.4, 0.2],
[0.5, 0.3, 0.2],
[0.6, 0.3, 0.1],
[0.3, 0.3, 0.4],
[0.2, 0.2, 0.6],
])
a_fuzzy_lin = Lineage(a_fuzzy, names=["0", "1", "2"])
b_np, c_np = _fuzzy_to_discrete(a_fuzzy=a_fuzzy, n_most_likely=2)
b_l, c_l = _fuzzy_to_discrete(a_fuzzy=a_fuzzy_lin, n_most_likely=2)
np.testing.assert_array_equal(b_np, b_l)
assert len(c_np) == 0
assert len(c_l) == 0
def test_critical_samples(self):
a_fuzzy = np.array(
[
[0.3, 0.7, 0],
[0.3, 0.6, 0.1],
[0.0, 0.7, 0.3],
[0.1, 0.9, 0],
[0.4, 0.4, 0.2],
[0.5, 0.3, 0.2],
[0.6, 0.3, 0.1],
[0.3, 0.3, 0.4],
[0.2, 0.2, 0.6],
]
)
_, c_1 = _fuzzy_to_discrete(a_fuzzy, n_most_likely=3, remove_overlap=False)
_, c_2 = _fuzzy_to_discrete(a_fuzzy, n_most_likely=3, remove_overlap=True)
assert c_1 == np.array(2)
np.testing.assert_array_equal(c_2, np.array([1, 2]))
def test_normal_run(self):
# create random data that sums to one row-wise
a_fuzzy = np.random.standard_normal((100, 3))
a_fuzzy = np.exp(a_fuzzy) / np.sum(np.exp(a_fuzzy), 1)[:, None]
# check with both overlap handlings
_fuzzy_to_discrete(a_fuzzy=a_fuzzy)
_fuzzy_to_discrete(a_fuzzy=a_fuzzy, n_most_likely=30, remove_overlap=True)
_fuzzy_to_discrete(a_fuzzy=a_fuzzy, n_most_likely=30, remove_overlap=False)
def test_raise_threshold(self):
a_fuzzy = np.repeat(np.array([0.9, 0.1])[None, :], 10, 0)
with pytest.raises(ValueError):
_fuzzy_to_discrete(a_fuzzy, n_most_likely=3, remove_overlap=True)
with pytest.raises(ValueError):
_fuzzy_to_discrete(a_fuzzy, n_most_likely=3, remove_overlap=False)
def test_too_many_cells(self):
a_fuzzy = np.random.standard_normal((100, 3))
a_fuzzy = np.exp(a_fuzzy) / np.sum(np.exp(a_fuzzy), 1)[:, None]
with pytest.raises(ValueError):
_fuzzy_to_discrete(a_fuzzy=a_fuzzy, n_most_likely=50)
def test_normalization(self):
a_fuzzy = np.random.standard_normal((100, 3))
with pytest.raises(ValueError):
_fuzzy_to_discrete(a_fuzzy=a_fuzzy)
