def test_normalise_errors_invalid_method():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
with pytest.raises(ValueError):
normalise.normalise(df, plate_id="Metadata_plate", method="invalid")
def test_check_control_within_function():
# dataframe with missing controls in one plate
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 10) + (["drug"] * 8 + ["DMSO"] * 2) * 4
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
missing_control_df = pd.DataFrame(list(zip(x, y, z, plate, compound)),
columns=colnames)
with pytest.raises(RuntimeError):
normalise.normalise(missing_control_df, plate_id="Metadata_plate")
def test_parallel_normalise():
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
standard_output = normalise.normalise(df, plate_id="Metadata_plate")
parallel_output = normalise.normalise(df,
plate_id="Metadata_plate",
parallel=True)
assert standard_output.equals(parallel_output)
def normalise(self, **kwargs):
"""normalise data via morar.normalise.normalise"""
df = normalise.normalise(self,
metadata_string=self.metadata_string,
prefix=self.prefix **kwargs)
return DataFrame(df, metadata_string=self.metadata_string,
prefix=self.metadata_prefix)
def test_normalise_errors_invalid_method():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
with pytest.raises(ValueError):
normalise.normalise(df, plate_id="Metadata_plate", method="invalid")
def test_parallel_normalise():
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
standard_output = normalise.normalise(df, plate_id="Metadata_plate")
parallel_output = normalise.normalise(df, plate_id="Metadata_plate", parallel=True)
assert standard_output.equals(parallel_output)
def normalise(self, **kwargs):
"""normalise data via morar.normalise.normalise"""
df = normalise.normalise(self,
metadata_string=self.metadata_string,
prefix=self.prefix**kwargs)
return DataFrame(df,
metadata_string=self.metadata_string,
prefix=self.metadata_prefix)
def test_normalise_subtract_returns_correct_values():
# simple dataframe to check actual values
x = [4, 4, 4, 2, 2]
compound = ["drug"] * 3 + ["DMSO"] * 2
plate = ["plate_1"] * 5
colnames = ["f1", "Metadata_compound", "Metadata_plate"]
simple_df = pd.DataFrame(list(zip(x, compound, plate)), columns=colnames)
out = normalise.normalise(simple_df, plate_id="Metadata_plate", method="subtract")
assert isinstance(out, pd.DataFrame)
assert out["f1"].tolist() == [2, 2, 2, 0, 0]
def test_check_control_within_function():
# dataframe with missing controls in one plate
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 10) + (["drug"] * 8 + ["DMSO"] * 2) * 4
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
missing_control_df = pd.DataFrame(
list(zip(x, y, z, plate, compound)), columns=colnames
)
with pytest.raises(RuntimeError):
normalise.normalise(missing_control_df, plate_id="Metadata_plate")
def test_normalise_returns_dataframe_divide():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
out = normalise.normalise(df, plate_id="Metadata_plate", method="divide")
assert isinstance(out, pd.DataFrame)
def test_normalise_divide_returns_correct_values():
# simple dataframe to check actual values
x = [4, 4, 4, 2, 2]
compound = ["drug"] * 3 + ["DMSO"] * 2
plate = ["plate_1"] * 5
colnames = ["f1", "Metadata_compound", "Metadata_plate"]
simple_df = pd.DataFrame(list(zip(x, compound, plate)), columns=colnames)
out = normalise.normalise(simple_df,
plate_id="Metadata_plate",
method="divide")
assert isinstance(out, pd.DataFrame)
assert out["f1"].tolist() == [2, 2, 2, 1, 1]
def test_normalise_returns_correct_size():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
out = normalise.normalise(df, plate_id="Metadata_plate")
assert out.shape[0] == df.shape[0]
def test_normalise_non_default_cols():
# dataframe with weird columns names
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = ["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 + [
"plate4"
] * 10 + ["plate5"] * 10
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "meta_plate", "meta_cmpd"]
non_default_df = pd.DataFrame(list(zip(x, y, z, plate, compound)),
columns=colnames)
out = normalise.normalise(non_default_df,
compound="meta_cmpd",
plate_id="meta_plate",
metadata_string="meta")
assert isinstance(out, pd.DataFrame)
def test_normalise_returns_dataframe_divide():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
out = normalise.normalise(df, plate_id="Metadata_plate", method="divide")
assert isinstance(out, pd.DataFrame)
def test_normalise_returns_correct_size():
# create test DataFrame
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "Metadata_plate", "Metadata_compound"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
out = normalise.normalise(df, plate_id="Metadata_plate")
assert out.shape[0] == df.shape[0]
def test_normalise_extra_metadata_cols():
# dataframe with weird columns names
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (["plate1"] * 10 + ["plate2"] * 10 + ["plate3"] * 10 +
["plate4"] * 10 + ["plate5"] * 10)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
extra_metadata = ["A", "B"] * 25
colnames = ["A", "B", "C", "meta_plate", "meta_cmpd", "metadata_extra"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound, extra_metadata)))
df.columns = colnames
out = normalise.normalise(df,
metadata_string="meta",
compound="meta_cmpd",
plate_id="meta_plate")
assert df.shape == out.shape
assert df.columns.tolist() == out.columns.tolist()
def test_normalise_non_default_cols():
# dataframe with weird columns names
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
colnames = ["A", "B", "C", "meta_plate", "meta_cmpd"]
non_default_df = pd.DataFrame(list(zip(x, y, z, plate, compound)), columns=colnames)
out = normalise.normalise(
non_default_df,
compound="meta_cmpd",
plate_id="meta_plate",
metadata_string="meta",
)
assert isinstance(out, pd.DataFrame)
def test_normalise_extra_metadata_cols():
# dataframe with weird columns names
x = np.random.randn(50).tolist()
y = np.random.randn(50).tolist()
z = np.random.randn(50).tolist()
plate = (
["plate1"] * 10
+ ["plate2"] * 10
+ ["plate3"] * 10
+ ["plate4"] * 10
+ ["plate5"] * 10
)
compound = (["drug"] * 8 + ["DMSO"] * 2) * 5
extra_metadata = ["A", "B"] * 25
colnames = ["A", "B", "C", "meta_plate", "meta_cmpd", "metadata_extra"]
df = pd.DataFrame(list(zip(x, y, z, plate, compound, extra_metadata)))
df.columns = colnames
out = normalise.normalise(
df, metadata_string="meta", compound="meta_cmpd", plate_id="meta_plate"
)
assert df.shape == out.shape
assert df.columns.tolist() == out.columns.tolist()