def test_create_torsiondrive_dataset():
"""
Make sure we can correclt make a dataset using the scan enumerator.
"""
factory = TorsiondriveDatasetFactory()
scan_filter = workflow_components.ScanEnumerator()
scan_filter.add_torsion_scan(smarts="[*:1]~[*:2]-[#8:3]-[#1:4]",
scan_rage=(-90, 90),
scan_increment=10)
factory.add_workflow_components(scan_filter)
conformer_generator = workflow_components.StandardConformerGenerator(
max_conformers=1)
factory.add_workflow_components(conformer_generator)
mols = Molecule.from_file(get_data("tautomers_small.smi"),
"smi",
allow_undefined_stereo=True)
dataset = factory.create_dataset(dataset_name="test name",
molecules=mols,
description="Force field test",
tagline="A test dataset",
processors=1)
assert dataset.n_molecules > 0
assert dataset.n_records > 0
for entry in dataset.dataset.values():
assert entry.keywords.dihedral_ranges == [(-90, 90)]
assert entry.keywords.grid_spacing == [10]
def test_torsiondrive_scan_keywords(fractal_compute_server):
"""
Test running torsiondrives with unique keyword settings which overwrite the global grid spacing and scan range.
"""
client = FractalClient(fractal_compute_server)
molecules = Molecule.from_smiles("CO")
factory = TorsiondriveDatasetFactory()
scan_enum = workflow_components.ScanEnumerator()
scan_enum.add_torsion_scan(smarts="[*:1]~[#6:2]-[#8:3]~[*:4]")
factory.add_workflow_components(scan_enum)
factory.clear_qcspecs()
factory.add_qc_spec(method="openff_unconstrained-1.1.0",
basis="smirnoff",
program="openmm",
spec_description="scan range test",
spec_name="openff-1.1.0")
dataset = factory.create_dataset(
dataset_name="Torsiondrive scan keywords",
molecules=molecules,
description="Testing scan keywords which overwrite the global settings",
tagline="Testing scan keywords which overwrite the global settings")
# now set the keywords
keys = list(dataset.dataset.keys())
entry = dataset.dataset[keys[0]]
entry.keywords = {"grid_spacing": [5], "dihedral_ranges": [(-10, 10)]}
# now submit
dataset.submit(client=client)
fractal_compute_server.await_services(max_iter=50)
# make sure of the results are complete
ds = client.get_collection("TorsionDriveDataset", dataset.dataset_name)
# get the entry
record = ds.get_record(ds.df.index[0], "openff-1.1.0")
assert record.keywords.grid_spacing == [5]
assert record.keywords.grid_spacing != dataset.grid_spacing
assert record.keywords.dihedral_ranges == [(-10, 10)]
assert record.keywords.dihedral_ranges != dataset.dihedral_ranges
def test_torsiondrive_submissions(fractal_compute_server, specification):
"""
Test submitting a torsiondrive dataset and computing it.
"""
client = FractalClient(fractal_compute_server)
qc_spec, driver = specification
program = qc_spec["program"]
if not has_program(program):
pytest.skip(f"Program '{program}' not found.")
molecules = Molecule.from_smiles("CO")
factory = TorsiondriveDatasetFactory(driver=driver)
factory.add_qc_spec(**qc_spec,
spec_name="default",
spec_description="test",
overwrite=True)
dataset = factory.create_dataset(
dataset_name=f"Test torsiondrives info {program}, {driver}",
molecules=molecules,
description="Test torsiondrive dataset",
tagline="Testing torsiondrive datasets",
)
# force a metadata validation error
dataset.metadata.long_description = None
with pytest.raises(DatasetInputError):
dataset.submit(client=client)
# re-add the description so we can submit the data
dataset.metadata.long_description = "Test basics dataset"
# now submit again
dataset.submit(client=client)
fractal_compute_server.await_services(max_iter=50)
# make sure of the results are complete
ds = client.get_collection("TorsionDriveDataset", dataset.dataset_name)
# check the metadata
meta = Metadata(**ds.data.metadata)
assert meta == dataset.metadata
# check the provenance
assert dataset.provenance == ds.data.provenance
# check the qc spec
for qc_spec in dataset.qc_specifications.values():
spec = ds.data.specs[qc_spec.spec_name]
assert spec.description == qc_spec.spec_description
assert spec.qc_spec.driver == dataset.driver
assert spec.qc_spec.method == qc_spec.method
assert spec.qc_spec.basis == qc_spec.basis
assert spec.qc_spec.program == qc_spec.program
# check the keywords
keywords = client.query_keywords(spec.qc_spec.keywords)[0]
assert keywords.values["maxiter"] == qc_spec.maxiter
assert keywords.values["scf_properties"] == qc_spec.scf_properties
# query the dataset
ds.query(qc_spec.spec_name)
for index in ds.df.index:
record = ds.df.loc[index].default
# this will take some time so make sure it is running with no error
assert record.status.value == "COMPLETE", print(record.dict())
assert record.error is None
assert len(record.final_energy_dict) == 24