def test_scale_merged_intensities_phenix(hewl_merged, ref_hewl,
mean_intensity_method):
"""
Compare phenix.french_wilson to scale_merged_intensities(). Current
test criteria are that >95% of I, SigI, F, and SigF are within 2%.
"""
mtz = hewl_merged.dropna()
scaled = scale_merged_intensities(
mtz, "IMEAN", "SIGIMEAN", mean_intensity_method=mean_intensity_method)
# Assert no reflections were dropped
assert len(scaled) == len(ref_hewl)
# Intensities should be identical
assert np.array_equal(scaled["IMEAN"].to_numpy(), ref_hewl["I"].to_numpy())
assert np.array_equal(scaled["SIGIMEAN"].to_numpy(),
ref_hewl["SIGI"].to_numpy())
rsF = scaled["FW-F"].to_numpy()
rsSigF = scaled["FW-SIGF"].to_numpy()
refF = ref_hewl["F"].to_numpy()
refSigF = ref_hewl["SIGF"].to_numpy()
rsI = scaled["FW-I"].to_numpy()
rsSigI = scaled["FW-SIGI"].to_numpy()
refI = ref_hewl["I"].to_numpy()
refSigI = ref_hewl["SIGI"].to_numpy()
assert (np.isclose(rsI, refI, rtol=0.02).sum() / len(scaled)) >= 0.95
assert (np.isclose(rsSigI, refSigI, rtol=0.02).sum() / len(scaled)) >= 0.95
assert (np.isclose(rsF, refF, rtol=0.02).sum() / len(scaled)) >= 0.95
assert (np.isclose(rsSigF, refSigF, rtol=0.02).sum() / len(scaled)) >= 0.95
def test_scale_merged_intensities_validdata(hewl_merged, inplace, output_columns,
mean_intensity_method):
"""
Confirm scale_merged_intensities() returns all positive values
"""
scaled = scale_merged_intensities(hewl_merged, "IMEAN", "SIGIMEAN",
output_columns=output_columns,
inplace=inplace,
mean_intensity_method=mean_intensity_method)
# Confirm inplace returns same object if true
if inplace:
assert id(scaled) == id(hewl_merged)
else:
assert id(scaled) != id(hewl_merged)
defaults = ("FW-I", "FW-SIGI", "FW-F", "FW-SIGF")
if output_columns:
o1, o2, o3, o4 = output_columns
else:
o1, o2, o3, o4 = defaults
# Confirm output columns are of desired types
assert isinstance(scaled[o1].dtype, rs.IntensityDtype)
assert isinstance(scaled[o2].dtype, rs.StandardDeviationDtype)
assert isinstance(scaled[o3].dtype, rs.StructureFactorAmplitudeDtype)
assert isinstance(scaled[o4].dtype, rs.StandardDeviationDtype)
# Confirm output columns are strictly positive
assert (scaled[o1].to_numpy() >= 0).all()
assert (scaled[o2].to_numpy() >= 0).all()
assert (scaled[o3].to_numpy() >= 0).all()
assert (scaled[o4].to_numpy() >= 0).all()
def test_scale_merged_intensities_dropna(data_hewl_all, dropna):
"""
Test scale_merged_intensities() using data containing NaNs in different
columns.
"""
keys = ["IMEAN", "SIGIMEAN"]
nan_in_intensity = data_hewl_all[keys].isna().to_numpy().any()
if nan_in_intensity and not dropna:
with pytest.raises(ValueError):
scaled = scale_merged_intensities(data_hewl_all, "IMEAN",
"SIGIMEAN", dropna=dropna)
else:
scaled = scale_merged_intensities(data_hewl_all, "IMEAN", "SIGIMEAN",
dropna=dropna)
assert id(scaled) != id(data_hewl_all)
assert (scaled["FW-I"].to_numpy() >= 0.).all()