def test_get_bonds(self):
mol1 = Molecule.from_file(os.path.join(test_dir, "t1.xyz"))
msc = MoleculeStructureComparator()
# noinspection PyProtectedMember
bonds = msc._get_bonds(mol1)
bonds_ref = [(0, 1), (0, 2), (0, 3), (0, 23), (3, 4), (3, 5), (5, 6),
(5, 7), (7, 8), (7, 9), (7, 21), (9, 10), (9, 11),
(9, 12), (12, 13), (12, 14), (12, 15), (15, 16), (15, 17),
(15, 18), (18, 19), (18, 20), (18, 21), (21, 22),
(21, 23), (23, 24), (23, 25)]
self.assertEqual(bonds, bonds_ref)
mol2 = Molecule.from_file(os.path.join(test_dir, "MgBH42.xyz"))
bonds = msc._get_bonds(mol2)
self.assertEqual(bonds, [(1, 3), (2, 3), (3, 4), (3, 5), (6, 8),
(7, 8), (8, 9), (8, 10)])
msc = MoleculeStructureComparator(ignore_ionic_bond=False)
bonds = msc._get_bonds(mol2)
self.assertEqual(bonds,
[(0, 1), (0, 2), (0, 3), (0, 5), (0, 6), (0, 7),
(0, 8), (0, 9), (1, 3), (2, 3), (3, 4), (3, 5),
(6, 8), (7, 8), (8, 9), (8, 10)])
mol1 = Molecule.from_file(
os.path.join(test_dir, "molecule_with_halogen_bonds_1.xyz"))
msc = MoleculeStructureComparator()
# noinspection PyProtectedMember
bonds = msc._get_bonds(mol1)
self.assertEqual(bonds, [(0, 12), (0, 13), (0, 14), (0, 15), (1, 12),
(1, 16), (1, 17), (1, 18), (2, 4), (2, 11),
(2, 19), (3, 5), (3, 10), (3, 20), (4, 6),
(4, 10), (5, 11), (5, 12), (6, 7), (6, 8),
(6, 9)])
def test_structural_change_in_geom_opt(self):
qcout_path = os.path.join(test_dir, "mol_1_3_bond.qcout")
qcout = QcOutput(qcout_path)
mol1 = qcout.data[0]["molecules"][0]
mol2 = qcout.data[0]["molecules"][-1]
priority_bonds = [[0, 1], [0, 2], [1, 3], [1, 4], [1, 7], [2, 5], [2, 6], [2, 8], [4, 6], [4, 10], [6, 9]]
msc = MoleculeStructureComparator(priority_bonds=priority_bonds)
self.assertTrue(msc.are_equal(mol1, mol2))
def test_structural_change_in_geom_opt(self):
qcout_path = os.path.join(test_dir, "mol_1_3_bond.qcout")
qcout = QcOutput(qcout_path)
mol1 = qcout.data[0]["molecules"][0]
mol2 = qcout.data[0]["molecules"][-1]
priority_bonds = [[0, 1], [0, 2], [1, 3], [1, 4], [1, 7], [2, 5], [2, 6], [2, 8], [4, 6], [4, 10], [6, 9]]
msc = MoleculeStructureComparator(priority_bonds=priority_bonds)
self.assertTrue(msc.are_equal(mol1, mol2))
def test_get_bonds(self):
mol = read_mol(os.path.join(test_dir, "t1.xyz"))
msc = MoleculeStructureComparator()
# noinspection PyProtectedMember
bonds = msc._get_bonds(mol)
bonds_ref = [(0, 1), (0, 2), (0, 3), (0, 23), (3, 4), (3, 5), (5, 6),
(5, 7), (7, 8), (7, 9), (7, 21), (9, 10), (9, 11),
(9, 12), (12, 13), (12, 14), (12, 15), (15, 16), (15, 17),
(15, 18), (18, 19), (18, 20), (18, 21), (21, 22),
(21, 23), (23, 24), (23, 25)]
self.assertEqual(bonds, bonds_ref)
def test_to_and_from_dict(self):
msc1 = MoleculeStructureComparator()
d1 = msc1.to_dict
d2 = MoleculeStructureComparator.from_dict(d1).to_dict
self.assertEqual(d1, d2)
thio1 = read_mol(os.path.join(test_dir, "thiophene1.xyz"))
# noinspection PyProtectedMember
msc2 = MoleculeStructureComparator(
bond_length_cap=0.2,
priority_bonds=msc1._get_bonds(thio1),
priority_cap=0.5)
d1 = msc2.to_dict
d2 = MoleculeStructureComparator.from_dict(d1).to_dict
self.assertEqual(d1, d2)
def test_to_and_from_dict(self):
msc1 = MoleculeStructureComparator()
d1 = msc1.as_dict()
d2 = MoleculeStructureComparator.from_dict(d1).as_dict()
self.assertEqual(d1, d2)
thio1 = Molecule.from_file(os.path.join(test_dir, "thiophene1.xyz"))
# noinspection PyProtectedMember
msc2 = MoleculeStructureComparator(
bond_length_cap=0.2, priority_bonds=msc1._get_bonds(thio1), priority_cap=0.5
)
d1 = msc2.as_dict()
d2 = MoleculeStructureComparator.from_dict(d1).as_dict()
self.assertEqual(d1, d2)
def test_get_13_bonds(self):
priority_bonds = [[0, 1], [0, 2], [1, 3], [1, 4], [1, 7], [2, 5],
[2, 6], [2, 8], [4, 6], [4, 10], [6, 9]]
bonds_13 = MoleculeStructureComparator.get_13_bonds(priority_bonds)
ans = ((0, 3), (0, 4), (0, 5), (0, 6), (0, 7), (0, 8), (1, 2), (1, 6),
(1, 10), (2, 4), (2, 9), (3, 4), (3, 7), (4, 7), (4, 9), (5, 6),
(5, 8), (6, 8), (6, 10))
self.assertEqual(bonds_13, tuple(ans))
def test_get_bonds(self):
mol1 = Molecule.from_file(os.path.join(test_dir, "t1.xyz"))
msc = MoleculeStructureComparator()
# noinspection PyProtectedMember
bonds = msc._get_bonds(mol1)
bonds_ref = [(0, 1), (0, 2), (0, 3), (0, 23), (3, 4), (3, 5), (5, 6),
(5, 7), (7, 8), (7, 9), (7, 21), (9, 10), (9, 11),
(9, 12), (12, 13), (12, 14), (12, 15), (15, 16), (15, 17),
(15, 18), (18, 19), (18, 20), (18, 21), (21, 22),
(21, 23), (23, 24), (23, 25)]
self.assertEqual(bonds, bonds_ref)
mol2 = Molecule.from_file(os.path.join(test_dir, "MgBH42.xyz"))
bonds = msc._get_bonds(mol2)
self.assertEqual(bonds, [(1, 3), (2, 3), (3, 4), (3, 5), (6, 8), (7, 8),
(8, 9), (8, 10)])
msc = MoleculeStructureComparator(ignore_ionic_bond=False)
bonds = msc._get_bonds(mol2)
self.assertEqual(bonds, [(0, 1), (0, 2), (0, 3), (0, 5), (0, 6), (0, 7),
(0, 8), (0, 9), (1, 3), (2, 3), (3, 4), (3, 5),
(6, 8), (7, 8), (8, 9), (8, 10)])
mol1 = Molecule.from_file(os.path.join(test_dir, "molecule_with_halogen_bonds_1.xyz"))
msc = MoleculeStructureComparator()
# noinspection PyProtectedMember
bonds = msc._get_bonds(mol1)
self.assertEqual(bonds, [(0, 12), (0, 13), (0, 14), (0, 15), (1, 12), (1, 16),
(1, 17), (1, 18), (2, 4), (2, 11), (2, 19), (3, 5),
(3, 10), (3, 20), (4, 6), (4, 10), (5, 11), (5, 12),
(6, 7), (6, 8), (6, 9)])
def _check_structure_change(mol1, mol2, fw_spec):
"""
Check whether structure is changed:
Return:
True: structure changed, False: unchanged
"""
if 'egsnl' not in fw_spec:
raise ValueError("Can't find initial SNL")
if isinstance(fw_spec['egsnl'], EGStructureNL):
egsnl_dict = fw_spec['egsnl'].as_dict()
else:
egsnl_dict = fw_spec['egsnl']
if 'known_bonds' not in egsnl_dict:
raise ValueError("Can't find known bonds information")
bonds = egsnl_dict['known_bonds']
msc = MoleculeStructureComparator(priority_bonds=bonds)
return not msc.are_equal(mol1, mol2)
def _check_structure_change(mol1, mol2, qcout_path):
"""
Check whether structure is changed:
Return:
True: structure changed, False: unchanged
"""
dirname = os.path.dirname(qcout_path)
fw_spec_path = os.path.join(dirname, "FW.json")
with zopen(zpath(fw_spec_path), 'rt') as f:
fw = json.load(f)
if 'egsnl' not in fw['spec']:
raise ValueError("Can't find initial SNL")
if 'known_bonds' not in fw['spec']['egsnl']:
raise ValueError("Can't find known bonds information")
bonds = fw['spec']['egsnl']['known_bonds']
msc = MoleculeStructureComparator(priority_bonds=bonds)
return not msc.are_equal(mol1, mol2)
def test_are_equal(self):
msc1 = MoleculeStructureComparator()
mol1 = Molecule.from_file(os.path.join(test_dir, "t1.xyz"))
mol2 = Molecule.from_file(os.path.join(test_dir, "t2.xyz"))
mol3 = Molecule.from_file(os.path.join(test_dir, "t3.xyz"))
self.assertFalse(msc1.are_equal(mol1, mol2))
self.assertTrue(msc1.are_equal(mol2, mol3))
thio1 = Molecule.from_file(os.path.join(test_dir, "thiophene1.xyz"))
thio2 = Molecule.from_file(os.path.join(test_dir, "thiophene2.xyz"))
# noinspection PyProtectedMember
msc2 = MoleculeStructureComparator(
priority_bonds=msc1._get_bonds(thio1))
self.assertTrue(msc2.are_equal(thio1, thio2))
def test_are_equal(self):
msc1 = MoleculeStructureComparator()
mol1 = Molecule.from_file(os.path.join(test_dir, "t1.xyz"))
mol2 = Molecule.from_file(os.path.join(test_dir, "t2.xyz"))
mol3 = Molecule.from_file(os.path.join(test_dir, "t3.xyz"))
self.assertFalse(msc1.are_equal(mol1, mol2))
self.assertTrue(msc1.are_equal(mol2, mol3))
thio1 = Molecule.from_file(os.path.join(test_dir, "thiophene1.xyz"))
thio2 = Molecule.from_file(os.path.join(test_dir, "thiophene2.xyz"))
# noinspection PyProtectedMember
msc2 = MoleculeStructureComparator(
priority_bonds=msc1._get_bonds(thio1))
self.assertTrue(msc2.are_equal(thio1, thio2))
hal1 = Molecule.from_file(os.path.join(test_dir, "molecule_with_halogen_bonds_1.xyz"))
hal2 = Molecule.from_file(os.path.join(test_dir, "molecule_with_halogen_bonds_2.xyz"))
msc3 = MoleculeStructureComparator(priority_bonds=msc1._get_bonds(hal1))
self.assertTrue(msc3.are_equal(hal1, hal2))
def test_are_equal(self):
msc1 = MoleculeStructureComparator()
mol1 = read_mol(os.path.join(test_dir, "t1.xyz"))
mol2 = read_mol(os.path.join(test_dir, "t2.xyz"))
mol3 = read_mol(os.path.join(test_dir, "t3.xyz"))
self.assertFalse(msc1.are_equal(mol1, mol2))
self.assertTrue(msc1.are_equal(mol2, mol3))
thio1 = read_mol(os.path.join(test_dir, "thiophene1.xyz"))
thio2 = read_mol(os.path.join(test_dir, "thiophene2.xyz"))
# noinspection PyProtectedMember
msc2 = MoleculeStructureComparator(
priority_bonds=msc1._get_bonds(thio1))
self.assertTrue(msc2.are_equal(thio1, thio2))
def get_redo_workflow(self,
qchem_input_params,
sp_params,
max_iterations=3):
"""
Identifies molecules which need to be re-run (for now, based only on
presence of negative frequencies) and then performs a frequency
flattening workflow on those molecules.
This is a hack. In the future, a frequency flattening workflow should be
used from the beginning.
:param qchem_input_params: dict
:param sp_params: For OptFreqSPFW, single-point calculations can be
treated differently from Opt and Freq. In this case, another dict
for sp must be used.
:param max_iterations: Maximum number of iterations for frequency
flattening. Default is 3.
:return: Workflow
"""
if self.db is None:
raise RuntimeError("Cannot access database to determine what"
"molecules need to be re-calculated.")
fws = []
collection = self.db.db["molecules"]
for mol in collection.find({}):
frequencies = mol["output"]["frequencies"]
if any([True if x < 0 else False for x in frequencies]):
min_molecule_perturb_scale = 0.1
max_molecule_perturb_scale = 0.3
scale_grid = 10
perturb_scale_grid = (max_molecule_perturb_scale -
min_molecule_perturb_scale) / scale_grid
msc = MoleculeStructureComparator()
old_molecule = None
for calc in mol["calcs_reversed"]:
if calc["task"]["type"] in ["freq", "frequency"
] and old_molecule is None:
negative_freq_vecs = calc.get(
"frequency_mode_vectors")[0]
old_coords = calc.get("initial_geometry")
old_molecule = Molecule.from_dict(
calc.get("initial_molecule"))
structure_successfully_perturbed = False
for molecule_perturb_scale in np.arange(
max_molecule_perturb_scale, min_molecule_perturb_scale,
-perturb_scale_grid):
new_coords = perturb_coordinates(
old_coords=old_coords,
negative_freq_vecs=negative_freq_vecs,
molecule_perturb_scale=molecule_perturb_scale,
reversed_direction=False)
new_molecule = Molecule(
species=old_molecule.species,
coords=new_coords,
charge=old_molecule.charge,
spin_multiplicity=old_molecule.spin_multiplicity)
if msc.are_equal(old_molecule, new_molecule):
structure_successfully_perturbed = True
break
if not structure_successfully_perturbed:
raise Exception(
"Unable to perturb coordinates to remove negative frequency without changing the bonding structure"
)
mol_id = mol["mol_id"]
dir_name = mol["dir_name"].split("/")[-1]
if dir_name not in listdir(self.base_dir):
os.mkdir(join(self.base_dir, dir_name))
fws.append(
OptFreqSPFW(molecule=new_molecule,
name="Flattening: {}/{}".format(
mol_id, dir_name),
qchem_cmd="qchem -slurm",
input_file=join(self.base_dir, dir_name,
mol_id + ".in"),
output_file=join(self.base_dir, dir_name,
mol_id + ".out"),
qclog_file=join(self.base_dir, dir_name,
mol_id + ".qclog"),
max_cores=32,
max_iterations=max_iterations,
qchem_input_params=qchem_input_params,
sp_params=sp_params,
db_file=self.db_file))
if len(fws) == 0:
return None
else:
return Workflow(fws)
def test_get_13_bonds(self):
priority_bonds = [[0, 1], [0, 2], [1, 3], [1, 4], [1, 7], [2, 5], [2, 6], [2, 8], [4, 6], [4, 10], [6, 9]]
bonds_13 = MoleculeStructureComparator.get_13_bonds(priority_bonds)
ans = ((0, 3), (0, 4), (0, 5), (0, 6), (0, 7), (0, 8), (1, 2), (1, 6), (1, 10), (2, 4), (2, 9), (3, 4),
(3, 7), (4, 7), (4, 9), (5, 6), (5, 8), (6, 8), (6, 10))
self.assertEqual(bonds_13, tuple(ans))
def opt_with_frequency_flattener(cls,
qchem_command,
multimode="openmp",
input_file="mol.qin",
output_file="mol.qout",
qclog_file="mol.qclog",
max_iterations=10,
max_molecule_perturb_scale=0.3,
reversed_direction=False,
ignore_connectivity=False,
**QCJob_kwargs):
"""
Optimize a structure and calculate vibrational frequencies to check if the
structure is in a true minima. If a frequency is negative, iteratively
perturbe the geometry, optimize, and recalculate frequencies until all are
positive, aka a true minima has been found.
Args:
qchem_command (str): Command to run QChem.
multimode (str): Parallelization scheme, either openmp or mpi.
input_file (str): Name of the QChem input file.
output_file (str): Name of the QChem output file.
max_iterations (int): Number of perturbation -> optimization -> frequency
iterations to perform. Defaults to 10.
max_molecule_perturb_scale (float): The maximum scaled perturbation that
can be applied to the molecule. Defaults to 0.3.
reversed_direction (bool): Whether to reverse the direction of the
vibrational frequency vectors. Defaults to False.
ignore_connectivity (bool): Whether to ignore differences in connectivity
introduced by structural perturbation. Defaults to False.
**QCJob_kwargs: Passthrough kwargs to QCJob. See
:class:`custodian.qchem.jobs.QCJob`.
"""
min_molecule_perturb_scale = 0.1
scale_grid = 10
perturb_scale_grid = (max_molecule_perturb_scale -
min_molecule_perturb_scale) / scale_grid
msc = MoleculeStructureComparator()
if not os.path.exists(input_file):
raise AssertionError('Input file must be present!')
orig_opt_input = QCInput.from_file(input_file)
orig_opt_rem = copy.deepcopy(orig_opt_input.rem)
orig_freq_rem = copy.deepcopy(orig_opt_input.rem)
orig_freq_rem["job_type"] = "freq"
first = True
for ii in range(max_iterations):
yield (QCJob(qchem_command=qchem_command,
multimode=multimode,
input_file=input_file,
output_file=output_file,
qclog_file=qclog_file,
suffix=".opt_" + str(ii),
backup=first,
**QCJob_kwargs))
first = False
opt_outdata = QCOutput(output_file + ".opt_" + str(ii)).data
if opt_outdata["structure_change"] == "unconnected_fragments":
print(
"Unstable molecule broke into unconnected fragments! Exiting..."
)
break
else:
freq_QCInput = QCInput(molecule=opt_outdata.get(
"molecule_from_optimized_geometry"),
rem=orig_freq_rem,
opt=orig_opt_input.opt,
pcm=orig_opt_input.pcm,
solvent=orig_opt_input.solvent)
freq_QCInput.write_file(input_file)
yield (QCJob(qchem_command=qchem_command,
multimode=multimode,
input_file=input_file,
output_file=output_file,
qclog_file=qclog_file,
suffix=".freq_" + str(ii),
backup=first,
**QCJob_kwargs))
outdata = QCOutput(output_file + ".freq_" + str(ii)).data
errors = outdata.get("errors")
if len(errors) != 0:
raise AssertionError(
'No errors should be encountered while flattening frequencies!'
)
if outdata.get('frequencies')[0] > 0.0:
print("All frequencies positive!")
break
else:
negative_freq_vecs = outdata.get(
"frequency_mode_vectors")[0]
old_coords = outdata.get("initial_geometry")
old_molecule = outdata.get("initial_molecule")
structure_successfully_perturbed = False
for molecule_perturb_scale in np.arange(
max_molecule_perturb_scale,
min_molecule_perturb_scale, -perturb_scale_grid):
new_coords = perturb_coordinates(
old_coords=old_coords,
negative_freq_vecs=negative_freq_vecs,
molecule_perturb_scale=molecule_perturb_scale,
reversed_direction=reversed_direction)
new_molecule = Molecule(
species=outdata.get('species'),
coords=new_coords,
charge=outdata.get('charge'),
spin_multiplicity=outdata.get('multiplicity'))
if msc.are_equal(old_molecule,
new_molecule) or ignore_connectivity:
structure_successfully_perturbed = True
break
if not structure_successfully_perturbed:
raise Exception(
"Unable to perturb coordinates to remove negative frequency without changing the bonding structure"
)
new_opt_QCInput = QCInput(molecule=new_molecule,
rem=orig_opt_rem,
opt=orig_opt_input.opt,
pcm=orig_opt_input.pcm,
solvent=orig_opt_input.solvent)
new_opt_QCInput.write_file(input_file)