def test_preprocess_expansion(write_yaml, shared_datadir, add_cli_arguments):
config_path = write_yaml(
{
"file_prefix": str(shared_datadir / "dummy"),
"split_size": {"training": 0.6, "testing": 0.2, "validation": 0.2},
}
)
add_cli_arguments(config_path)
expansion_main()
with open(shared_datadir / "dummy_template_library.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 10
with open(shared_datadir / "dummy_training.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 6
with open(shared_datadir / "dummy_testing.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 2
with open(shared_datadir / "dummy_validation.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 2
data = pd.read_hdf(shared_datadir / "dummy_unique_templates.hdf5", "table")
config = Config(config_path)
assert len(data) == 2
assert "retro_template" in data.columns
assert "library_occurence" in data.columns
for column in config["metadata_headers"]:
assert column in data.columns
def test_preprocess_expansion_bad_product(
write_yaml, shared_datadir, add_cli_arguments
):
config_path = write_yaml(
{
"file_prefix": str(shared_datadir / "dummy_sani"),
"split_size": {"training": 0.6, "testing": 0.2, "validation": 0.2},
}
)
add_cli_arguments(config_path)
with pytest.raises(MoleculeException):
expansion_main()
def test_preprocess_expansion_skip_bad_product(
write_yaml, shared_datadir, add_cli_arguments
):
config_path = write_yaml(
{
"file_prefix": str(shared_datadir / "dummy_sani"),
"split_size": {"training": 0.6, "testing": 0.2, "validation": 0.2},
"remove_unsanitizable_products": True,
}
)
add_cli_arguments(config_path)
expansion_main()
with open(shared_datadir / "dummy_sani_template_library.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 10
def test_preprocess_recommender(write_yaml, shared_datadir, add_cli_arguments):
config_path = write_yaml({
"file_prefix": str(shared_datadir / "dummy"),
"split_size": {
"training": 0.6,
"testing": 0.2,
"validation": 0.2
},
})
add_cli_arguments(config_path)
expansion_main()
with open(shared_datadir / "dummy_template_library.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 10
os.remove(shared_datadir / "dummy_training.csv")
os.remove(shared_datadir / "dummy_testing.csv")
os.remove(shared_datadir / "dummy_validation.csv")
os.remove(shared_datadir / "dummy_unique_templates.hdf5")
recommender_main()
with open(shared_datadir / "dummy_training.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 6
with open(shared_datadir / "dummy_testing.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 2
with open(shared_datadir / "dummy_validation.csv", "r") as fileobj:
lines = fileobj.read().splitlines()
assert len(lines) == 2
data = pd.read_hdf(shared_datadir / "dummy_unique_templates.hdf5", "table")
assert len(data) == 2