def test_handle_regressions_default_jitter(mocker, fake_phenotype_data,
output_dir,
fake_phenotype_data_desc_conf):
def fake_build_regression(*args):
return {"pvalue_regression_male": 0, "pvalue_regression_female": 0}
mocked = mocker.patch(
"dae.pheno_browser.prepare_data."
"PreparePhenoBrowserBase.build_regression",
side_effect=fake_build_regression,
)
reg = GPFConfigParser.load_config(fake_phenotype_data_desc_conf,
pheno_conf_schema)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir,
reg)
regressand = fake_phenotype_data.get_measure("i1.m1")
for i in prep.handle_regressions(regressand):
pass
mocked.assert_called()
measure, reg_measure, jitter = mocked.call_args_list[0][0]
assert jitter == 0.12
measure, reg_measure, jitter = mocked.call_args_list[1][0]
assert jitter == 0.13
def test_augment_measure_with_identical_measures(fake_phenotype_data,
output_dir):
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.age")
regressor = fake_phenotype_data.get_measure("i1.age")
df = prep._augment_measure_values_df(regressor, "test regression",
regressand)
assert df is None
def test_build_regression_identical_measures(mocker, fake_phenotype_data,
output_dir):
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.age")
regressor = fake_phenotype_data.get_measure("i1.age")
jitter = 0.32403423849
assert prep.build_regression(regressand, regressor, jitter) == {}
def test_augment_measure_with_nonexistent_regressor(fake_phenotype_data,
output_dir):
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i2.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
regressor.instrument_name = None
df = prep._augment_measure_values_df(regressor, "test regression",
regressand)
assert df is None
def build_pheno_browser(dbfile,
pheno_name,
output_dir,
pheno_regressions=None):
phenodb = pheno_db.PhenotypeStudy(pheno_name, dbfile=dbfile)
prep = PreparePhenoBrowserBase(pheno_name, phenodb, output_dir,
pheno_regressions)
prep.run()
def test_handle_regressions_regressand_is_regressor(
fake_phenotype_data, output_dir, fake_phenotype_data_desc_conf):
reg = GPFConfigParser.load_config(fake_phenotype_data_desc_conf,
pheno_conf_schema)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir,
reg)
regressand = fake_phenotype_data.get_measure("i1.age")
with pytest.raises(StopIteration):
next(prep.handle_regressions(regressand))
def test_handle_regressions_non_continuous_or_ordinal_measure(
fake_phenotype_data, output_dir, fake_phenotype_data_desc_conf):
reg = GPFConfigParser.load_config(fake_phenotype_data_desc_conf,
pheno_conf_schema)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir,
reg)
regressand_categorical = fake_phenotype_data.get_measure("i1.m5")
regressand_raw = fake_phenotype_data.get_measure("i1.m6")
with pytest.raises(StopIteration):
next(prep.handle_regressions(regressand_categorical))
with pytest.raises(StopIteration):
next(prep.handle_regressions(regressand_raw))
def test_build_regression_aug_df_is_none(mocker, fake_phenotype_data,
output_dir):
def fake_augment_df(*args):
return None
mocker.patch(
"dae.pheno_browser.prepare_data."
"PreparePhenoBrowserBase._augment_measure_values_df",
side_effect=fake_augment_df,
)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
jitter = 0.32403423849
assert prep.build_regression(regressand, regressor, jitter) == {}
def test_has_regression_measure(fake_phenotype_data, output_dir,
regressions_conf):
reg = GPFConfigParser.load_config(regressions_conf, regression_conf_schema)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir,
reg)
expected_reg_measures = [
("regressor1", "i1"),
("regressor2", "i1"),
("common_regressor", ""),
("common_regressor", "i1"),
("common_regressor", "i2"),
("regressor1", "i2"),
]
for e in expected_reg_measures:
assert prep._has_regression_measure(*e)
def test_augment_measure(fake_phenotype_data, output_dir):
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
df = prep._augment_measure_values_df(regressor, "test regression",
regressand)
roles = list(df["role"].unique())
assert len(roles) == 3
for role in [Role.parent, Role.sib, Role.prb]:
assert role in roles
assert list(df) == [
"person_id",
"family_id",
"role",
"sex",
"status",
"test regression",
"i1.m1",
]
assert len(df) > 0
def test_augment_measure_regressor_no_instrument_name(fake_phenotype_data,
output_dir):
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
exp_df = prep._augment_measure_values_df(regressor, "test regression",
regressand)
regressor.instrument_name = None
df = prep._augment_measure_values_df(regressor, "test regression",
regressand)
assert list(df) == [
"person_id",
"family_id",
"role",
"sex",
"status",
"test regression",
"i1.m1",
]
assert len(df) > 0
assert list(df["test regression"]) == list(exp_df["test regression"])
def test_build_regression_min_unique_vals(mocker, fake_phenotype_data,
output_dir):
fake_df = pd.DataFrame({
"i1.m1": [1, 1, 1, 1, 1, 1],
"age": [1, 2, 3, 4, 5, 6],
"role": [
Role.prb,
Role.prb,
Role.prb,
Role.prb,
Role.prb,
Role.prb,
],
"sex": [
Sex.male,
Sex.female,
Sex.male,
Sex.female,
Sex.male,
Sex.female,
],
})
def fake_augment_df(*args):
return fake_df
mocker.patch(
"dae.pheno_browser.prepare_data."
"PreparePhenoBrowserBase._augment_measure_values_df",
side_effect=fake_augment_df,
)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
jitter = 0.32403423849
assert prep.build_regression(regressand, regressor, jitter) == {}
def test_handle_regressions(mocker, fake_phenotype_data, output_dir,
fake_phenotype_data_desc_conf):
def fake_build_regression(dependent_measure, independent_measure, jitter):
return {
"regressand": dependent_measure,
"regressor": independent_measure,
"jitter": jitter,
"pvalue_regression_male": 0,
"pvalue_regression_female": 0,
}
mocked = mocker.patch(
"dae.pheno_browser.prepare_data."
"PreparePhenoBrowserBase.build_regression",
side_effect=fake_build_regression,
)
reg = GPFConfigParser.load_config(fake_phenotype_data_desc_conf,
pheno_conf_schema)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir,
reg)
regressand = fake_phenotype_data.get_measure("i1.m1")
res = [r for r in prep.handle_regressions(regressand) if r is not None]
assert len(res) == 2
assert sorted([r["regression_id"] for r in res]) == sorted(["age", "nviq"])
mocked.assert_called()
measure, reg_measure, jitter = mocked.call_args_list[0][0]
assert measure.measure_id == "i1.m1"
assert reg_measure.measure_id == "i1.age"
assert jitter == 0.12
measure, reg_measure, jitter = mocked.call_args_list[1][0]
assert measure.measure_id == "i1.m1"
assert reg_measure.measure_id == "i1.iq"
assert jitter == 0.13
def test_build_regression(mocker, fake_phenotype_data, output_dir):
fake_df = pd.DataFrame({
# Only two unique values, in order to test
# the MIN_UNIQUE_VALUES check
"i1.m1": [1, 2, 1, 2, 1, 2],
"age": [1, 2, 3, 4, 5, 6],
"role": [
Role.prb,
Role.prb,
Role.prb,
Role.prb,
Role.prb,
Role.prb,
],
"sex": [
Sex.male,
Sex.female,
Sex.male,
Sex.female,
Sex.male,
Sex.female,
],
})
def fake_augment_df(*args):
return fake_df
def fake_linregres(*args):
class Result:
pvalues = {"age": 0.123456}
res_male = Result()
res_female = Result()
res_female.pvalues["age"] = 0.654321
return (res_male, res_female)
def fake_savefig(*args):
return ("figsmall", "fig")
mocked_linregres = mocker.patch(
"dae.pheno_browser.prepare_data.draw_linregres",
side_effect=fake_linregres,
)
mocker.patch(
"dae.pheno_browser.prepare_data.PreparePhenoBrowserBase.save_fig",
side_effect=fake_savefig,
)
mocker.patch(
"dae.pheno_browser.prepare_data."
"PreparePhenoBrowserBase._augment_measure_values_df",
side_effect=fake_augment_df,
)
prep = PreparePhenoBrowserBase("fake", fake_phenotype_data, output_dir)
regressand = fake_phenotype_data.get_measure("i1.m1")
regressor = fake_phenotype_data.get_measure("i1.age")
jitter = 0.32403423849
res = prep.build_regression(regressand, regressor, jitter)
assert res is not None
assert type(res) is dict
mocked_linregres.assert_called_once()
df, col1, col2, jitter = mocked_linregres.call_args[0]
assert col1 == "age"
assert col2 == "i1.m1"
assert jitter == 0.32403423849