def test_save_to_json(tmp_path, ecoli_obj):
fname = str(tmp_path / "ecoli_obj.json")
save_to_json(ecoli_obj, fname)
assert isfile(fname)
new_icd = load_json_model(fname)
# Data
assert np.allclose(new_icd.M, ecoli_obj.M)
assert np.allclose(new_icd.A, ecoli_obj.A)
assert np.allclose(new_icd.X, ecoli_obj.X)
# Names
assert new_icd.imodulon_names == ecoli_obj.imodulon_names
assert new_icd.sample_names == ecoli_obj.sample_names
assert new_icd.gene_names == ecoli_obj.gene_names
# Tables
assert new_icd.gene_table.equals(ecoli_obj.gene_table.replace("", np.nan))
assert new_icd.sample_table.equals(
ecoli_obj.sample_table.replace("", np.nan))
assert new_icd.imodulon_table.equals(
ecoli_obj.imodulon_table.replace("", np.nan))
# Thresholds
assert np.allclose(
[new_icd.thresholds[x] for x in new_icd.imodulon_names],
[ecoli_obj.thresholds[x] for x in ecoli_obj.imodulon_names],
)
assert new_icd.cutoff_optimized == ecoli_obj.cutoff_optimized
assert new_icd.dagostino_cutoff == ecoli_obj.dagostino_cutoff
def load_bsub_data():
"""
Load *Bacillus subtilis* :class:`~pymodulon.core.IcaData` object from
:cite:`Rychel2020a`
Returns
-------
bsub_data: ~pymodulon.core.IcaData
*B. subtilis* :class:`~pymodulon.core.IcaData` object
"""
return load_json_model(path.join(_data_dir, "objects", "bsub_data.json.gz"))
def load_staph_data():
"""
Load *Staphylococcus aureus* :class:`~pymodulon.core.IcaData` object from
:cite:`Poudel2020`
Returns
-------
staph_data: ~pymodulon.core.IcaData
*S. aureus* :class:`~pymodulon.core.IcaData` object
"""
return load_json_model(path.join(_data_dir, "objects", "staph_data.json.gz"))
def load_ecoli_data():
"""
Load *Escherichia coli* :class:`~pymodulon.core.IcaData` object from
:cite:`Sastry2019`
Returns
-------
ecoli_data: ~pymodulon.core.IcaData
*E. coli* :class:`~pymodulon.core.IcaData` object
"""
return load_json_model(path.join(_data_dir, "objects", "ecoli_data.json.gz"))
def test_init_thresholds(ecoli_obj, mini_obj_opt, caplog, tmp_path):
# Test all threshold initialization options
M = mini_obj_opt.M
A = mini_obj_opt.A
# Reduce TRN size
trn = ecoli_obj.trn.iloc[:10].copy()
# Define custom thresholds
thresh_list = list(np.arange(1, 2, 0.1))
thresh_dict = dict(zip(M.columns, thresh_list))
thresholds_opts = [None, thresh_dict]
threshold_method_opts = ["kmeans", "dagostino"]
trn_opts = [None, trn]
optimize_cutoff_opts = [True, False]
dagostino_cutoff_opts = [None, 600]
for opt1 in thresholds_opts:
for opt2 in threshold_method_opts:
for opt3 in trn_opts:
for opt4 in optimize_cutoff_opts:
for opt5 in dagostino_cutoff_opts:
caplog.clear()
obj1 = IcaData(
M,
A,
thresholds=opt1,
threshold_method=opt2,
trn=opt3,
optimize_cutoff=opt4,
dagostino_cutoff=opt5,
)
fname = str(tmp_path / "test_data.json")
save_to_json(obj1, fname)
obj2 = load_json_model(fname)
for obj in [obj1, obj2]:
if opt1 is not None:
assert obj.thresholds == thresh_dict
assert obj.dagostino_cutoff is None
assert not obj.cutoff_optimized
if opt4: # optimize_cutoff == True
assert caplog.messages == [
"Using manually input thresholds. "
"D'agostino optimization will not be "
"performed."
]
else:
assert caplog.text == ""
# opt1 == None
elif opt2 == "kmeans" or opt3 is None:
assert obj.dagostino_cutoff is None
assert not obj.cutoff_optimized
if opt4: # optimize_cutoff == True
assert caplog.messages == [
"Using Kmeans threshold "
"method. D'agostino optimization will not "
"be performed."
]
else:
assert caplog.text == ""
# opt2 == 'dagostino'
# opt3 == trn
elif opt4:
assert obj.cutoff_optimized
assert obj.dagostino_cutoff is not None
assert caplog.messages == [
"Optimizing iModulon thresholds, may take "
"2-3 minutes..."
]
# opt4 == False
elif opt5 is None:
assert not obj.cutoff_optimized
assert obj.dagostino_cutoff == 550
assert caplog.messages == [
"Using the default "
"dagostino_cutoff of 550. This may not be "
"optimal for your dataset. Use "
"ica_data.reoptimize_thresholds() to find the "
"optimal threshold."
]
elif opt5 == 600:
assert not obj.cutoff_optimized
assert obj.dagostino_cutoff == 600
else:
raise ValueError("Missing test case!")