def test_create_dataset_atom_map():
"""Test creating a dataset with molecules with atom maps."""
factory = OptimizationDatasetFactory()
mol = Molecule.from_smiles("CCCC([O-])=O")
mol.properties['atom_map'] = {1: 1, 2: 2, 3: 3, 4: 4}
_ = factory.create_dataset(dataset_name="test name",
molecules=mol,
description="Force field test",
tagline="A test dataset")
def test_optimization_submissions_with_pcm(fractal_compute_server):
"""Test submitting an Optimization dataset to a snowflake server with PCM."""
client = FractalClient(fractal_compute_server)
program = "psi4"
if not has_program(program):
pytest.skip(f"Program '{program}' not found.")
# use a single small molecule due to the extra time PCM takes
molecules = Molecule.from_smiles("C")
factory = OptimizationDatasetFactory(driver="gradient")
factory.add_qc_spec(method="hf",
basis="sto-3g",
program=program,
spec_name="default",
spec_description="test",
implicit_solvent=PCMSettings(units="au",
medium_Solvent="water"),
overwrite=True)
dataset = factory.create_dataset(
dataset_name="Test optimizations info with pcm water",
molecules=molecules,
description="Test optimization dataset",
tagline="Testing optimization 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_results()
# make sure of the results are complete
ds = client.get_collection("OptimizationDataset", 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
assert record.status.value == "COMPLETE"
assert record.error is None
assert len(record.trajectory) > 1
result = record.get_trajectory()[0]
assert "CURRENT DIPOLE X" in result.extras["qcvars"].keys()
assert "SCF QUADRUPOLE XX" in result.extras["qcvars"].keys()
# make sure the PCM result was captured
assert result.extras["qcvars"]["PCM POLARIZATION ENERGY"] < 0
def test_optimization_submissions(fractal_compute_server, specification):
"""Test submitting an Optimization dataset to a snowflake server."""
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_file(get_data("butane_conformers.pdb"), "pdb")
factory = OptimizationDatasetFactory(driver=driver)
factory.add_qc_spec(**qc_spec,
spec_name="default",
spec_description="test",
overwrite=True)
dataset = factory.create_dataset(
dataset_name=f"Test optimizations info {program}, {driver}",
molecules=molecules[:2],
description="Test optimization dataset",
tagline="Testing optimization 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_results()
# make sure of the results are complete
ds = client.get_collection("OptimizationDataset", 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
assert record.status.value == "COMPLETE"
assert record.error is None
assert len(record.trajectory) > 1
# if we used psi4 make sure the properties were captured
if program == "psi4":
result = record.get_trajectory()[0]
assert "CURRENT DIPOLE X" in result.extras["qcvars"].keys()
assert "SCF QUADRUPOLE XX" in result.extras["qcvars"].keys()
def test_adding_specifications(fractal_compute_server):
"""
Test adding specifications to datasets.
Here we are testing multiple scenarios:
1) Adding an identical specification to a dataset
2) Adding a spec with the same name as another but with different options
3) overwrite a spec which was added but never used.
"""
client = FractalClient(fractal_compute_server)
mol = Molecule.from_smiles("CO")
# make a dataset
factory = OptimizationDatasetFactory()
opt_dataset = factory.create_dataset(
dataset_name="Specification error check",
molecules=mol,
description="test adding new compute specs to datasets",
tagline="test adding new compute specs")
opt_dataset.clear_qcspecs()
# add a new mm spec
opt_dataset.add_qc_spec(method="openff-1.0.0",
basis="smirnoff",
program="openmm",
spec_description="default openff spec",
spec_name="openff-1.0.0")
# submit the optimizations and let the compute run
opt_dataset.submit(client=client)
fractal_compute_server.await_results()
fractal_compute_server.await_services()
# grab the collection
ds = client.get_collection(opt_dataset.type, opt_dataset.dataset_name)
# now try and add the specification again this should return True
assert opt_dataset._add_dataset_specification(
spec=opt_dataset.qc_specifications["openff-1.0.0"],
procedure_spec=opt_dataset.optimization_procedure.get_optimzation_spec(
),
dataset=ds) is True
# now change part of the spec but keep the name the same
opt_dataset.clear_qcspecs()
opt_dataset.add_qc_spec(method="openff-1.2.1",
basis="smirnoff",
spec_name="openff-1.0.0",
program="openmm",
spec_description="openff-1.2.1 with wrong name.")
# now try and add this specification with the same name but different settings
with pytest.raises(QCSpecificationError):
opt_dataset._add_dataset_specification(
spec=opt_dataset.qc_specifications["openff-1.0.0"],
procedure_spec=opt_dataset.optimization_procedure.
get_optimzation_spec(),
dataset=ds)
# now add a new specification but no compute and make sure it is overwritten
opt_dataset.clear_qcspecs()
opt_dataset.add_qc_spec(method="ani1x",
basis=None,
program="torchani",
spec_name="ani",
spec_description="a ani spec")
assert opt_dataset._add_dataset_specification(
spec=opt_dataset.qc_specifications["ani"],
procedure_spec=opt_dataset.optimization_procedure.get_optimzation_spec(
),
dataset=ds) is True
# now change the spec slightly and add again
opt_dataset.clear_qcspecs()
opt_dataset.add_qc_spec(method="ani1ccx",
basis=None,
program="torchani",
spec_name="ani",
spec_description="a ani spec")
assert opt_dataset._add_dataset_specification(
spec=opt_dataset.qc_specifications["ani"],
procedure_spec=opt_dataset.optimization_procedure.get_optimzation_spec(
),
dataset=ds) is True