def test_label_molecules(verbose=False):
"""Test labeling/getting stats on labeling molecules"""
molecules = read_molecules(
get_data_filename('molecules/AlkEthOH-tripos.mol2.gz'),
verbose=verbose)
ffxml = get_data_filename('forcefield/Frosst_AlkEthOH.ffxml')
get_molecule_parameterIDs(molecules, ffxml)
def test_smirnoff_energies_vs_parmatfrosst(verbose=False):
"""Test evaluation of energies from parm@frosst ffxml files versus energies of equivalent systems."""
from openeye import oechem
prefix = 'AlkEthOH_'
molecules = ['r118', 'r12', 'c1161', 'r0', 'c100', 'c38', 'c1266']
# Loop over molecules, load OEMols and prep for comparison/do comparison
for molnm in molecules:
f_prefix = os.path.join('molecules', prefix + molnm)
mol2file = get_data_filename(f_prefix + '.mol2')
prmtop = get_data_filename(f_prefix + '.top')
crd = get_data_filename(f_prefix + '.crd')
# Load special parm@frosst with parm99/parm@frosst bugs re-added for testing
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH_parmAtFrosst.ffxml'))
# Load OEMol
mol = oechem.OEGraphMol()
ifs = oechem.oemolistream(mol2file)
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
oechem.OETriposAtomNames(mol)
# Do comparison
results = compare_molecule_energies(prmtop,
crd,
forcefield,
mol,
verbose=verbose)
def test_improper(verbose=False):
"""Test implement of impropers on benzene."""
from openeye import oechem
# Load benzene
ifs = oechem.oemolistream(get_data_filename('molecules/benzene.mol2'))
mol = oechem.OEMol()
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
ifs.close()
# Load forcefield
ffxml = get_data_filename('forcefield/benzene_minimal.ffxml')
ff = ForceField(ffxml)
# Load AMBER files and compare
crd = get_data_filename('molecules/benzene.crd')
top = get_data_filename('molecules/benzene.top')
g0, g1, e0, e1 = compare_molecule_energies(top,
crd,
ff,
mol,
skip_assert=True)
# Check that torsional energies the same to 1 in 10^6
rel_error = np.abs((g0['torsion'] - g1['torsion']) / g0['torsion'])
if rel_error > 2e-5: #Note that this will not be tiny because we use six-fold impropers and they use a single improper
raise Exception(
"Improper torsion energy for benzene differs too much (relative error %.4g) between AMBER and SMIRNOFF."
% rel_error)
def test_merge_system():
"""Test merging of a system created from AMBER and another created from SMIRNOFF."""
#Create System from AMBER
prefix = os.path.join('systems', 'amber', 'cyclohexane_ethanol_0.4_0.6')
prmtop = get_data_filename(prefix + '.prmtop')
incrd = get_data_filename(prefix + '.inpcrd')
from openforcefield.typing.engines.smirnoff import create_system_from_amber
topology0, system0, positions0 = create_system_from_amber(prmtop, incrd)
from openeye import oechem
# Load simple OEMol
ifs = oechem.oemolistream(
get_data_filename('molecules/AlkEthOH_c100.mol2'))
mol = oechem.OEMol()
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
oechem.OETriposAtomNames(mol)
# Load forcefield file
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
topology1, system1, positions1 = create_system_from_molecule(
forcefield, mol)
merge_system(topology0,
topology1,
system0,
system1,
positions0,
positions1,
verbose=True)
def test_gromacs_roundtrip():
"""Save a System (a mixture) to GROMACS, read back in, verify yields same energy and force terms."""
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
filename = get_data_filename(
os.path.join('systems', 'packmol_boxes',
'cyclohexane_ethanol_0.4_0.6.pdb'))
from simtk.openmm.app import PDBFile
pdbfile = PDBFile(filename)
mol2files = [
get_data_filename(os.path.join('systems', 'monomers', 'ethanol.mol2')),
get_data_filename(
os.path.join('systems', 'monomers', 'cyclohexane.mol2'))
]
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
# setup system
system = forcefield.createSystem(pdbfile.topology, mols)
# Create ParmEd structure, save to AMBER
a, topfile = tempfile.mkstemp(suffix='.top')
a, grofile = tempfile.mkstemp(suffix='.gro')
save_system_to_gromacs(pdbfile.topology, system, pdbfile.positions,
topfile, grofile)
# Read back in and cross-check energies
top = parmed.load_file(topfile)
gro = parmed.load_file(grofile)
gromacssys = top.createSystem(nonbondedMethod=app.NoCutoff,
constraints=None,
implicitSolvent=None)
groups0, groups1, energy0, energy1 = compare_system_energies(
pdbfile.topology,
pdbfile.topology,
gromacssys,
system,
pdbfile.positions,
verbose=False)
# Remove temp files
os.remove(topfile)
os.remove(grofile)
def test_create_system_molecules_parmatfrosst_gbsa(verbose=False):
"""Test creation of a System object from small molecules to test parm@frosst forcefield with GBSA support.
"""
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH_GBSA.ffxml'))
for f in check_AlkEthOH(forcefield,
"to test Parm@Frosst parameters",
verbose=verbose):
yield f
def test_create_system_boxes_smirnoff99Frosst(verbose=False):
"""Test creation of a System object from some boxes of mixed solvents to test parm@frosst forcefield.
"""
forcefield = ForceField(
get_data_filename('forcefield/smirnoff99Frosst.ffxml'))
for f in check_boxes(forcefield,
description="to test Parm@frosst parameters",
verbose=verbose):
yield f
def create_yank_input_files():
"""Creates XML, PDB and YAML script files to run YANK calculations.
Starting from mol2 files, the script uses packmol to solvate the molecules
in TIP3P waters and produces PDB files of the vacuum and solvated boxes
together with OpenMM XML System files parametrized with smirnoff99Frosst.
A YANK script is generated for each system following the unified FreeSolv
protocol.
"""
database_filepath = os.path.join('..', 'database.json')
smirnoff_openmm_dir = os.path.join('..', 'openmmfiles')
pdb_dir = os.path.join('..', 'pdbfiles')
water_mol2_filepath = get_data_filename(
os.path.join('systems', 'monomers', 'tip3p_water.mol2'))
# Create directories that don't exist.
for directory_path in [smirnoff_openmm_dir, pdb_dir]:
if not os.path.exists(directory_path):
os.makedirs(directory_path)
# Read in database.
with open(database_filepath, 'r') as f:
database = json.load(f)
# Find all molecules.
mol2_files_pattern = os.path.join('..', 'mol2files_sybyl', '*.mol2')
tripos_filenames = list(glob.glob(mol2_files_pattern))
# Create YANK input files.
for i, tripos_filename in enumerate(tripos_filenames):
molecule_id, _ = os.path.splitext(
os.path.basename(tripos_filename)) # Remove path and extension.
print('\rSetting up {} ({}/{}) '.format(molecule_id, i + 1,
len(tripos_filenames)))
# Output file paths.
vacuum_filename = molecule_id + '_vacuum'
solvated_filename = molecule_id + '_solvated'
vacuum_pdb_filepath = os.path.join(pdb_dir, vacuum_filename + '.pdb')
solvated_pdb_filepath = os.path.join(pdb_dir,
solvated_filename + '.pdb')
vacuum_smirnoff_openmm_filepath = os.path.join(
smirnoff_openmm_dir, vacuum_filename + '.xml')
solvated_smirnoff_openmm_filepath = os.path.join(
smirnoff_openmm_dir, solvated_filename + '.xml')
# Export solvated and vacuum molecule in PDB format.
create_pdb_files(database[molecule_id]['smiles'], tripos_filename,
water_mol2_filepath, vacuum_pdb_filepath,
solvated_pdb_filepath)
# Create OpenMM XML file parametrized with SMIRNOFF.
create_xml_system_files(tripos_filename, water_mol2_filepath,
vacuum_pdb_filepath, solvated_pdb_filepath,
vacuum_smirnoff_openmm_filepath,
solvated_smirnoff_openmm_filepath)
def test_parameter_completeness_check(self):
"""Test that proper exceptions are raised if a force field fails to assign parameters to valence terms in a molecule."""
from openeye import oechem
mol = oechem.OEMol()
oechem.OEParseSmiles(mol, 'CCC')
oechem.OEAddExplicitHydrogens(mol)
oechem.OETriposAtomNames(mol)
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
topology = generateTopologyFromOEMol(mol)
# Test nonbonded error checking by wiping out required LJ parameter
params = ff.getParameter(paramID='n0001')
params['smirks'] = '[#136:1]'
ff.setParameter(paramID='n0001', params=params)
ff.setParameter(paramID='n0002', params=params)
with self.assertRaises(Exception):
system = ff.createSystem(topology, [mol])
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
# Test bond error checking by wiping out a required bond parameter
params = ff.getParameter(paramID='b0001')
params['smirks'] = '[#136:1]~[*:2]'
ff.setParameter(paramID='b0001', params=params)
with self.assertRaises(Exception):
system = ff.createSystem(topology, [mol])
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
# Test angle error checking by wiping out a required angle parameter
params = ff.getParameter(paramID='a0001')
params['smirks'] = '[#136:1]~[*:2]~[*:3]'
ff.setParameter(paramID='a0001', params=params)
with self.assertRaises(Exception):
system = ff.createSystem(topology, [mol])
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
# Test torsion error checking by wiping out a required torsion parameter
params = ff.getParameter(paramID='t0001')
params['smirks'] = '[#136:1]~[*:2]~[*:3]~[*:4]'
ff.setParameter(paramID='t0001', params=params)
ff.setParameter(paramID='t0004', params=params)
with self.assertRaises(Exception):
system = ff.createSystem(topology, [mol])
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
def initialize_system(dt=0.001,
temperature=100,
forcefield_file='forcefield/smirnoff99Frosst.offxml',
smiles="C1CCCCC1"):
mol = OEMol()
# OEParseSmiles(mol, 'CCOCCSCC')
# OEParseSmiles(mol, 'c1ccccc1')
OEParseSmiles(mol, smiles)
# OEParseSmiles(mol, 'C([C@@H]1[C@H]([C@@H]([C@H](C(O1)O)O)O)O)O')
OEAddExplicitHydrogens(mol)
masses = get_masses(mol)
num_atoms = mol.NumAtoms()
topology = generateTopologyFromOEMol(mol)
ff = ForceField(get_data_filename(forcefield_file))
nrgs, total_params, offsets = system_builder.construct_energies(
ff, mol, False)
# dt = 0.0025
# friction = 10.0
# temperature = 300
# gradient descent
dt = dt
friction = 10.0
temperature = temperature
a, b, c = get_abc_coefficents(masses, dt, friction, temperature)
buf_size = estimate_buffer_size(1e-10, a)
print("BUFFER SIZE", buf_size)
omegaOpts = oeomega.OEOmegaOptions()
omegaOpts.SetMaxConfs(1)
omega = oeomega.OEOmega(omegaOpts)
omega.SetStrictStereo(False)
if not omega(mol):
assert 0
x0 = mol_coords_to_numpy_array(mol) / 10
intg = custom_ops.Integrator_double(dt, buf_size, num_atoms, total_params,
a, b, c)
context = custom_ops.Context_double(nrgs, intg)
x0 = minimizer.minimize_newton_cg(nrgs, x0, total_params)
return nrgs, offsets, intg, context, x0, total_params
def check_boxes(forcefield, description="", chargeMethod=None, verbose=False):
"""Test creation of System from boxes of mixed solvents.
"""
# Read monomers
mols = list()
monomers = [
'water', 'cyclohexane', 'ethanol', 'propane', 'methane', 'butanol'
]
from openeye import oechem
mol = oechem.OEGraphMol()
for monomer in monomers:
filename = get_data_filename(
os.path.join('systems', 'monomers', monomer + '.sdf'))
ifs = oechem.oemolistream(filename)
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
if verbose: print('%d reference molecules loaded' % len(mols))
# Read systems.
boxes = [
'ethanol_water.pdb', 'cyclohexane_water.pdb',
'cyclohexane_ethanol_0.4_0.6.pdb',
'propane_methane_butanol_0.2_0.3_0.5.pdb'
]
from simtk.openmm.app import PDBFile
for box in boxes:
filename = get_data_filename(
os.path.join('systems', 'packmol_boxes', box))
pdbfile = PDBFile(filename)
f = partial(check_system_creation_from_topology,
forcefield,
pdbfile.topology,
mols,
pdbfile.positions,
chargeMethod=chargeMethod,
verbose=verbose)
f.description = 'Test creation of System object from %s %s' % (
box, description)
yield f
def test_change_parameters(verbose=False):
"""Test modification of forcefield parameters."""
from openeye import oechem
# Load simple OEMol
ifs = oechem.oemolistream(
get_data_filename('molecules/AlkEthOH_c100.mol2'))
mol = oechem.OEMol()
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
oechem.OETriposAtomNames(mol)
# Load forcefield file
ffxml = get_data_filename('forcefield/Frosst_AlkEthOH.ffxml')
ff = ForceField(ffxml)
topology = generateTopologyFromOEMol(mol)
# Create initial system
system = ff.createSystem(topology, [mol], verbose=verbose)
# Get initial energy before parameter modification
positions = positions_from_oemol(mol)
old_energy = get_energy(system, positions)
# Get params for an angle
params = ff.getParameter(smirks='[a,A:1]-[#6X4:2]-[a,A:3]')
# Modify params
params['k'] = '0.0'
ff.setParameter(params, smirks='[a,A:1]-[#6X4:2]-[a,A:3]')
# Write params
ff.writeFile(tempfile.TemporaryFile(suffix='.ffxml'))
# Make sure params changed energy! (Test whether they get rebuilt on system creation)
system = ff.createSystem(topology, [mol], verbose=verbose)
energy = get_energy(system, positions)
if verbose:
print("New energy/old energy:", energy, old_energy)
if np.abs(energy - old_energy) < 0.1:
raise Exception("Error: Parameter modification did not change energy.")
def check_AlkEthOH(forcefield,
description="",
chargeMethod=None,
verbose=False):
"""Test creation of System from AlkEthOH small molecules.
"""
from openeye import oechem
ifs = oechem.oemolistream(
get_data_filename('molecules/AlkEthOH-tripos.mol2.gz'))
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
args = {'verbose': verbose, 'chargeMethod': chargeMethod}
f = partial(check_system_creation_from_molecule, forcefield, mol,
**args)
f.description = 'Testing creation of system object from small molecules (%s) %s' % (
mol.GetTitle(), description)
yield f
def test_molecule_labeling(verbose=False):
"""Test using labelMolecules to see which parameters applied to an oemol."""
from openeye import oechem
mol = oechem.OEMol()
oechem.OEParseSmiles(mol, 'CCC')
oechem.OEAddExplicitHydrogens(mol)
ff = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
labels = ff.labelMolecules([mol], verbose=verbose)
# Check that force terms aren't empty
print(labels[0].keys())
if not 'HarmonicBondGenerator' in labels[0].keys():
raise Exception("No force term assigned for harmonic bonds.")
if not 'HarmonicAngleGenerator' in labels[0].keys():
raise Exception("No force term assigned for harmonic angles.")
if not 'PeriodicTorsionGenerator' in labels[0].keys():
raise Exception("No force term assigned for periodic torsions.")
if not 'NonbondedGenerator' in labels[0].keys():
raise Exception("No nonbonded force term assigned.")
def initialize(input_smiles, gp):
train_reference_args = []
train_args = []
train_offset_idxs = []
train_charge_idxs = []
for smi_idx, smiles in enumerate(input_smiles):
print("processing", smiles, smi_idx, "/", len(input_smiles))
mol = OEMol()
OEParseSmiles(mol, smiles)
OEAddExplicitHydrogens(mol)
masses = get_masses(mol)
num_atoms = mol.NumAtoms()
omegaOpts = oeomega.OEOmegaOptions()
omegaOpts.SetMaxConfs(1)
omega = oeomega.OEOmega(omegaOpts)
omega.SetStrictStereo(False)
if not omega(mol):
assert 0
topology = generateTopologyFromOEMol(mol)
reference_forcefield_file = 'forcefield/smirnoff99Frosst_perturbed.offxml'
ff = ForceField(get_data_filename(reference_forcefield_file))
params = system_builder.construct_energies(ff, mol, True)
train_reference_args.append((params[0], params[1], masses, mol))
params = system_builder.construct_energies(ff, mol, False)
# global_params = args[0]
# nrg_params = args[1]
# total_params = args[2]
# masses = args[3]
# mol = args[4]
# charge_idxs = args[5]
train_args.append((gp, params[0], params[1], masses, mol, params[3]))
train_offset_idxs.append(params[2])
train_charge_idxs.append(params[3])
label_confs = [generate_conformations(a) for a in train_reference_args]
return train_args, train_offset_idxs, train_charge_idxs, label_confs
def test_tip3p_constraints():
"""Test that TIP3P distance costraints are correctly applied."""
# TIP3P constrained distances.
tip3p_oh_distance = 0.9572 # angstrom
tip3p_hoh_angle = 104.52 # angle
tip3p_hh_distance = tip3p_oh_distance * np.sin(
np.radians(tip3p_hoh_angle / 2)) * 2
expected_distances = [
tip3p_oh_distance, tip3p_oh_distance, tip3p_hh_distance
]
# Load tip3p molecule as OEMol.
tip3p_mol2_filepath = get_data_filename(
os.path.join('systems', 'monomers', 'tip3p_water.mol2'))
tip3p_oemol = read_molecules(tip3p_mol2_filepath, verbose=False)[0]
# Extract topology and positions.
tip3p_topology = generateTopologyFromOEMol(tip3p_oemol)
tip3p_positions = positions_from_oemol(tip3p_oemol)
# Create tip3p water system.
ff = ForceField('forcefield/tip3p.ffxml')
system = ff.createSystem(tip3p_topology, [tip3p_oemol])
# Run dynamics.
integrator = openmm.VerletIntegrator(2.0 * unit.femtoseconds)
context = openmm.Context(system, integrator)
context.setPositions(tip3p_positions)
integrator.step(50)
# Constrained distances are correct.
state = context.getState(getPositions=True)
new_positions = state.getPositions(asNumpy=True) / unit.angstroms
distances = []
for atom_1, atom_2 in [(0, 1), (0, 2),
(1, 2)]: # pair of atoms O-H1, O-H2, H1-H2
distances.append(
np.linalg.norm(new_positions[atom_1] - new_positions[atom_2]))
err_msg = 'expected distances [O-H1, O-H2, H1-H2]: {} A, new distances: {} A'
assert np.allclose(expected_distances,
distances), err_msg.format(expected_distances,
distances)
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
name = sys.argv[1]+'_'+sys.argv[3]
filename = 'packmol_boxes/'+name+'.pdb'
pdb = PDBFile(filename)
forcefield = ForceField(get_data_filename('forcefield/smirnoff99Frosst.ffxml'))
params = forcefield.getParameter(smirks='[#1:1]-[#8]')
params['rmin_half']='0.01'
params['epsilon']='0.01'
forcefield.setParameter(params, smirks='[#1:1]-[#8]')
smirkseries = sys.argv[4]
eps = sys.argv[5]
rmin = sys.argv[6]
epsval = sys.argv[7]
rminval = sys.argv[8]
param = forcefield.getParameter(smirks=smirkseries)
param[eps] = epsval
param[rmin] = rminval
def test_read_ffxml():
"""Test reading of ffxml files.
"""
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
def create_yank_input_files():
"""Creates XML, PDB and YAML script files to run YANK calculations on a subset of the FreeSolv set.
Starting from mol2 files, the script uses packmol to solvate the molecules
in TIP3P waters and produces PDB files of the vacuum and solvated boxes
together with OpenMM XML System files parametrized with smirnoff99Frosst.
A YANK script is generated for each system following the unified FreeSolv
protocol.
"""
database_filepath = os.path.join('..', 'database.json')
smirnoff_openmm_dir = os.path.join('..', 'openmmfiles')
pdb_dir = os.path.join('..', 'pdbfiles')
water_mol2_filepath = get_data_filename(
os.path.join('systems', 'monomers', 'tip3p_water.mol2'))
# Create directories that don't exist.
for directory_path in [smirnoff_openmm_dir, pdb_dir]:
if not os.path.exists(directory_path):
os.makedirs(directory_path)
# Read in database.
with open(database_filepath, 'r') as f:
database = json.load(f)
# Find all molecules.
mol2_files_pattern = os.path.join('..', 'mol2files_sybyl', '*.mol2')
tripos_filenames = list(glob.glob(mol2_files_pattern))
# Remove everything with names which are NOT in the list of those we need to repeat.
file = open(
'../../modification_validation/filtered_data/FreeSolv/database_filtered.txt',
'r')
database_text = file.readlines()
file.close()
retain_mols = []
for line in database_text:
#If it's not a comment, parse it and add to list
if line[0] != '#':
retain_mols.append(line.split(';')[0])
fnm = []
for filename in tripos_filenames:
molid = os.path.basename(filename).replace('.mol2', '')
if molid in retain_mols:
fnm.append(filename)
tripos_filenames = fnm
# Create YANK input files.
for i, tripos_filename in enumerate(tripos_filenames):
molecule_id, _ = os.path.splitext(
os.path.basename(tripos_filename)) # Remove path and extension.
print('\rSetting up {} ({}/{}) '.format(molecule_id, i + 1,
len(tripos_filenames)))
# Output file paths.
vacuum_filename = molecule_id + '_vacuum'
solvated_filename = molecule_id + '_solvated'
vacuum_pdb_filepath = os.path.join(pdb_dir, vacuum_filename + '.pdb')
solvated_pdb_filepath = os.path.join(pdb_dir,
solvated_filename + '.pdb')
vacuum_smirnoff_openmm_filepath = os.path.join(
smirnoff_openmm_dir, vacuum_filename + '.xml')
solvated_smirnoff_openmm_filepath = os.path.join(
smirnoff_openmm_dir, solvated_filename + '.xml')
# Export solvated and vacuum molecule in PDB format.
create_pdb_files(database[molecule_id]['smiles'], tripos_filename,
water_mol2_filepath, vacuum_pdb_filepath,
solvated_pdb_filepath)
# Create OpenMM XML file parametrized with SMIRNOFF.
create_xml_system_files(tripos_filename, water_mol2_filepath,
vacuum_pdb_filepath, solvated_pdb_filepath,
vacuum_smirnoff_openmm_filepath,
solvated_smirnoff_openmm_filepath)
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
name = sys.argv[1]+'_'+sys.argv[3]
filename = 'packmol_boxes/'+name+'.pdb'
pdb = PDBFile(filename)
FF_filename = get_data_filename('forcefield/smirnoff99Frosst.ffxml')
#FF_filename = 'smirnoff99Frosst_with_AllConstraints.ffxml'
forcefield = ForceField(FF_filename)
params = forcefield.getParameter(smirks='[#1:1]-[#8]')
params['rmin_half']='0.01'
params['epsilon']='0.01'
forcefield.setParameter(params, smirks='[#1:1]-[#8]')
smirkseries1 = sys.argv[4]
eps = sys.argv[5]
rmin = sys.argv[6]
epsval1 = sys.argv[7]
rminval1 = sys.argv[8]
smirkseries2 = sys.argv[9]
def new_param_energy(coords, params, topology, vecs, P=1.01, T=293.15,NPT=False,V=None,P_conv=1.e5,V_conv=1.e-6,Ener_conv=1.e-3,N_part=250.):
"""
Return potential energies associated with specified parameter perturbations.
Parameters
----------
coords: coordinates from the simulation(s) ran on the given molecule
params: arbitrary length dictionary of changes in parameter across arbitrary number of states. Highest level key is the molecule AlkEthOH_ID,
second level of keys are the new state, the values of each of these subkeys are a arbitrary length list of length 3 lists where the
length 3 lists contain information on a parameter to change in the form: [SMIRKS, parameter type, parameter value]. I.e. :
params = {'AlkEthOH_c1143':{'State 1':[['[6X4:1]-[#1:2]','k','620'],['[6X4:1]-[#6X4:2]','length','1.53'],...],'State 2':[...],...}}
P: Pressure of the system. By default set to 1.01 bar.
T: Temperature of the system. By default set to 300 K.
Returns
-------
E_kn: a kxN matrix of the dimensional energies associated with the forcfield parameters used as input
u_kn: a kxN matrix of the dimensionless energies associated with the forcfield parameters used as input
"""
#-------------------
# CONSTANTS
#-------------------
kB = 0.0083145 #Boltzmann constant (Gas constant) in kJ/(mol*K)
beta = 1/(kB*T)
#-------------------
# PARAMETERS
#-------------------
params = params
mol2files = []
for i in params:
mol2files.append('monomers/'+i.rsplit(' ',1)[0]+'.mol2')
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
K = len(params['cyclohexane'].keys())
# Load forcefield file
#ffxml = 'smirnoff99Frosst_with_AllConstraints.ffxml'#
#print('The forcefield being used is smirnoff99Frosst_with_AllConstraints.ffxml')
ffxml = get_data_filename('forcefield/smirnoff99Frosst.ffxml')
print('The forcefield being used is smirnoff99Frosst.ffxml')
ff = ForceField(ffxml)
# Generate a topology
top = topology#generateTopologyFromOEMol(mol)
#-----------------
# MAIN
#-----------------
# Calculate energies
E_kn = np.zeros([K,len(coords)],np.float64)
u_kn = np.zeros([K,len(coords)],np.float64)
for i,j in enumerate(params):
AlkEthOH_id = j
for k,l in enumerate(params[AlkEthOH_id]):
print("Anotha one")
for m,n in enumerate(params[AlkEthOH_id][l]):
newparams = ff.getParameter(smirks=n[0])
newparams[n[1]]=n[2]
ff.setParameter(newparams,smirks=n[0])
system = ff.createSystem(top,mols,nonbondedMethod=PME,nonbondedCutoff=1.125*nanometers,ewaldErrorTolerance=1.e-5)
barostat = MonteCarloBarostat(P*bar, T*kelvin, 10)
system.addForce(barostat)
for o,p in enumerate(coords):
e = get_energy(system,p,vecs[o])
if not NPT:
E_kn[k,o] = e._value
u_kn[k,o] = e._value*beta
else:
E_kn[k,o] = e._value + P*P_conv*V[o]*V_conv*Ener_conv*N_part
u_kn[k,o] = (e._value + P*P_conv*V[o]*V_conv*Ener_conv*N_part)*beta
return E_kn,u_kn
# Cross-check energies of molecules from AlkEthOH set using SMIRNOFF xml file
# versus energies from AMBER .prmtop and .crd files (parm@frosst params)
datapath = './AlkEthOH_inputfiles/AlkEthOH_rings_filt1'
#molname = 'AlkEthOH_r0' #That fails, but it's complicated. Try cyclobutane
molname = 'AlkEthOH_r51'
mol_filename = os.path.join( datapath, molname+'.mol2')
# Check if we have this data file; if not we have to extract the archive
if not os.path.isfile( mol_filename):
print "Extracting archived molecule files."
tarfile = 'AlkEthOH_inputfiles.tar.gz'
os.system('tar -xf AlkEthOH_inputfiles.tar.gz')
# Load OEMol
mol = oechem.OEGraphMol()
ifs = oechem.oemolistream(mol_filename)
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol )
oechem.OETriposAtomNames(mol)
# Load forcefield
forcefield = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH_parmAtFrosst.offxml'))
# Compare energies
prmtop = os.path.join( datapath, molname+'.top')
crd = os.path.join( datapath, molname+'.crd')
results = compare_molecule_energies( prmtop, crd, forcefield, mol)
def new_param_energy(coords, params, topology, vecs, P=1.01, T=300.):
"""
Return potential energies associated with specified parameter perturbations.
Parameters
----------
coords: coordinates from the simulation(s) ran on the given molecule
params: arbitrary length dictionary of changes in parameter across arbitrary number of states. Highest level key is the molecule AlkEthOH_ID,
second level of keys are the new state, the values of each of these subkeys are a arbitrary length list of length 3 lists where the
length 3 lists contain information on a parameter to change in the form: [SMIRKS, parameter type, parameter value]. I.e. :
params = {'AlkEthOH_c1143':{'State 1':[['[6X4:1]-[#1:2]','k','620'],['[6X4:1]-[#6X4:2]','length','1.53'],...],'State 2':[...],...}}
P: Pressure of the system. By default set to 1.01 bar.
T: Temperature of the system. By default set to 300 K.
Returns
-------
E_kn: a kxN matrix of the dimensional energies associated with the forcfield parameters used as input
u_kn: a kxN matrix of the dimensionless energies associated with the forcfield parameters used as input
"""
#-------------------
# CONSTANTS
#-------------------
kB = 0.0083145 #Boltzmann constant (Gas constant) in kcal/(mol*K)
beta = 1 / (kB * T)
#-------------------
# PARAMETERS
#-------------------
params = params
# Determine number of states we wish to estimate potential energies for
mol2files = []
for i in params:
mol2files.append('monomers/' + i.rsplit(' ', 1)[0] + '.mol2')
# mol2files.append('monomers/'+i.rsplit(' ',1)[1]+'.mol2')
#print mol2files
#mol = 'Mol2_files/'+mols[0]+'.mol2'
#K = len(params[mols[0]].keys())
#if np.shape(params) != np.shape(N_k): raise "K_k and N_k must have same dimensions"
# Determine max number of samples to be drawn from any state
#mol2files = ['monomers/'+sys.argv[1]+'.mol2','monomers/'+sys.argv[4]+'.mol2']
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
K = len(params['cyclohexane'].keys())
# Load forcefield file
ffxml = get_data_filename('forcefield/smirnoff99Frosst.ffxml')
ff = ForceField(ffxml)
# Generate a topology
#from smarty.forcefield import generateTopologyFromOEMol
top = topology #generateTopologyFromOEMol(mol)
#-----------------
# MAIN
#-----------------
# Calculate energies
E_kn = np.zeros([K, len(coords)], np.float64)
u_kn = np.zeros([K, len(coords)], np.float64)
for i, j in enumerate(params):
AlkEthOH_id = j
for k, l in enumerate(params[AlkEthOH_id]):
for m, n in enumerate(params[AlkEthOH_id][l]):
newparams = ff.getParameter(smirks=n[0])
newparams[n[1]] = n[2]
ff.setParameter(newparams, smirks=n[0])
system = ff.createSystem(top,
mols,
nonbondedMethod=PME,
nonbondedCutoff=12. * angstroms)
barostat = MonteCarloBarostat(P * bar, T * kelvin, 25)
system.addForce(barostat)
#print system
for o, p in enumerate(coords):
e = get_energy(system, p, vecs[o])
print o, e
E_kn[k, o] = e._value
u_kn[k, o] = e._value * beta
return E_kn, u_kn
def new_param_energy_vac(coords, params, T=293.15):
"""
Return potential energies associated with specified parameter perturbations.
Parameters
----------
coords: coordinates from the simulation(s) ran on the given molecule
params: arbitrary length dictionary of changes in parameter across arbitrary number of states. Highest level key is the molecule AlkEthOH_ID,
second level of keys are the new state, the values of each of these subkeys are a arbitrary length list of length 3 lists where the
length 3 lists contain information on a parameter to change in the form: [SMIRKS, parameter type, parameter value]. I.e. :
params = {'AlkEthOH_c1143':{'State 1':[['[6X4:1]-[#1:2]','k','620'],['[6X4:1]-[#6X4:2]','length','1.53'],...],'State 2':[...],...}}
T: Temperature of the system. By default set to 300 K.
Returns
-------
E_kn: a kxN matrix of the dimensional energies associated with the forcfield parameters used as input
u_kn: a kxN matrix of the dimensionless energies associated with the forcfield parameters used as input
"""
#-------------------
# CONSTANTS
#-------------------
kB = 0.0083145 #Boltzmann constant (Gas constant) in kJ/(mol*K)
beta = 1/(kB*T)
#-------------------
# PARAMETERS
#-------------------
params = params
# Determine number of states we wish to estimate potential energies for
mols = []
for i in params:
mols.append(i)
mol = 'monomers/'+mols[0]+'.mol2'
K = len(params[mols[0]].keys())
#-------------
# SYSTEM SETUP
#-------------
verbose = False # suppress echos from OEtoolkit functions
ifs = oechem.oemolistream(mol)
mol = oechem.OEMol()
# This uses parm@frosst atom types, so make sure to use the forcefield-flavor reader
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol )
# Perceive tripos types
oechem.OETriposAtomNames(mol)
# Load forcefield file
#ffxml = 'smirnoff99Frosst_with_AllConstraints.ffxml'#
#print('The forcefield being used is smirnoff99Frosst_with_AllConstraints.ffxml')
ffxml = get_data_filename('forcefield/smirnoff99Frosst.ffxml')
print('The forcefield being used is smirnoff99Frosst.ffxml')
ff = ForceField(ffxml)
# Generate a topology
topology = generateTopologyFromOEMol(mol)
#-----------------
# MAIN
#-----------------
# Calculate energies
E_kn = np.zeros([K,len(coords)],np.float64)
u_kn = np.zeros([K,len(coords)],np.float64)
for i,j in enumerate(params):
AlkEthOH_id = j
for k,l in enumerate(params[AlkEthOH_id]):
print("Anotha one")
for m,n in enumerate(params[AlkEthOH_id][l]):
newparams = ff.getParameter(smirks=n[0])
newparams[n[1]]=n[2]
ff.setParameter(newparams,smirks=n[0])
system = ff.createSystem(topology, [mol])
#print(newparams)
for o,p in enumerate(coords):
e = get_energy_vac(system,p)
E_kn[k,o] = e._value
u_kn[k,o] = e._value*beta
return E_kn,u_kn
def test_protein_structure(self):
from simtk.openmm import app
pdbfile = utils.get_data_filename('proteins/T4-protein.pdb')
proteinpdb = app.PDBFile(pdbfile)
protein_structure = utils.generateProteinStructure(proteinpdb)
self.assertIsInstance(protein_structure, parmed.structure.Structure)
# Cross-check energies of molecules from AlkEthOH set using SMIRNOFF xml file
# versus energies from AMBER .prmtop and .crd files (parm@frosst params)
datapath = './AlkEthOH_inputfiles/AlkEthOH_rings_filt1'
#molname = 'AlkEthOH_r0' #That fails, but it's complicated. Try cyclobutane
molname = 'AlkEthOH_r51'
mol_filename = os.path.join(datapath, molname + '.mol2')
# Check if we have this data file; if not we have to extract the archive
if not os.path.isfile(mol_filename):
print "Extracting archived molecule files."
tarfile = 'AlkEthOH_inputfiles.tar.gz'
os.system('tar -xf AlkEthOH_inputfiles.tar.gz')
# Load OEMol
mol = oechem.OEGraphMol()
ifs = oechem.oemolistream(mol_filename)
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
oechem.OETriposAtomNames(mol)
# Load forcefield
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH_parmAtFrosst.ffxml'))
# Compare energies
prmtop = os.path.join(datapath, molname + '.top')
crd = os.path.join(datapath, molname + '.crd')
results = compare_molecule_energies(prmtop, crd, forcefield, mol)
def test_tip3p_solvated_molecule_energy():
"""Check the energy of a TIP3P solvated molecule is the same with SMIRNOFF and OpenMM.
This test makes also use of defining the force field with multiple FFXML files,
and test the energy of mixture systems (i.e. molecule plus water).
"""
# TODO remove this function when ParmEd#868 is fixed
def add_water_bonds(system):
"""Hack to make tip3p waters work with ParmEd Structure.createSystem.
Bond parameters need to be specified even when constrained.
"""
k_tip3p = 462750.4 * unit.kilojoule_per_mole / unit.nanometers**2
length_tip3p = 0.09572 * unit.nanometers
for force in system.getForces():
if isinstance(force, openmm.HarmonicBondForce):
force.addBond(particle1=0,
particle2=1,
length=length_tip3p,
k=k_tip3p)
force.addBond(particle1=0,
particle2=2,
length=length_tip3p,
k=k_tip3p)
monomers_dir = get_data_filename(os.path.join('systems', 'monomers'))
# Load tip3p molecule as OEMol.
tip3p_mol2_filepath = os.path.join(monomers_dir, 'tip3p_water.mol2')
tip3p_oemol = read_molecules(tip3p_mol2_filepath, verbose=False)[0]
# Create OpenMM-parametrized water ParmEd Structure.
tip3p_openmm_ff = openmm.app.ForceField('tip3p.xml')
tip3p_topology = generateTopologyFromOEMol(tip3p_oemol)
tip3p_positions = positions_from_oemol(tip3p_oemol)
tip3p_openmm_system = tip3p_openmm_ff.createSystem(tip3p_topology)
# TODO remove this line when ParmEd#868 is fixed
# Hack to make tip3p waters work with ParmEd Structure.createSystem.
add_water_bonds(tip3p_openmm_system)
tip3p_openmm_structure = parmed.openmm.topsystem.load_topology(
tip3p_topology, tip3p_openmm_system, tip3p_positions)
# Test case: monomer_filename
test_cases = [('smirnoff99Frosst.ffxml', 'ethanol.mol2'),
('smirnoff99Frosst.ffxml', 'methane.mol2')]
for ff_name, molecule_filename in test_cases:
# Load molecule as OEMol.
molecule_filepath = os.path.join(monomers_dir, molecule_filename)
molecule_oemol = read_molecules(molecule_filepath, verbose=False)[0]
# Create molecule parametrized ParmEd args.
molecule_ff = ForceField('forcefield/' + ff_name)
molecule_args = forcefield_utils.create_system_from_molecule(
molecule_ff, molecule_oemol)
molecule_structure = parmed.openmm.topsystem.load_topology(
*molecule_args)
_, _, molecule_positions = molecule_args
# Merge molecule and OpenMM TIP3P water.
structure = tip3p_openmm_structure + molecule_structure
structure.positions = np.append(tip3p_positions,
molecule_positions,
axis=0)
system = structure.createSystem()
energy_openmm = get_energy(system, structure.positions)
# Test the creation of system with multiple force field files.
ff = ForceField('forcefield/' + ff_name, 'forcefield/tip3p.ffxml')
system = ff.createSystem(structure.topology,
[tip3p_oemol, molecule_oemol])
# TODO remove this line when ParmEd#868 is fixed
# Add bonds to match ParmEd hack above.
add_water_bonds(system)
energy_smirnoff = get_energy(system, structure.positions)
# The energies of the systems must be the same.
assert np.isclose(energy_openmm,
energy_smirnoff), 'OpenMM: {}, SMIRNOFF: {}'.format(
energy_openmm, energy_smirnoff)
#!/bin/env python
from simtk.openmm import app
import openforcefield.utils as utils
from openeye import oechem
molecule = oechem.OEMol()
molpdb = utils.get_data_filename('molecules/toluene.pdb')
with oechem.oemolistream(molpdb) as ifs:
oechem.OEReadMolecule(ifs, molecule)
molecule_structure = utils.generateSMIRNOFFStructure(molecule)
print('Molecule:', molecule_structure)
pdbfile = utils.get_data_filename('proteins/T4-protein.pdb')
proteinpdb = app.PDBFile(pdbfile)
protein_structure = utils.generateProteinStructure(proteinpdb)
print('Protein:', protein_structure)
structure = utils.mergeStructure(protein_structure, molecule_structure)
print('Complex:', structure)
def test_read_ffxml_gbsa():
"""Test reading of ffxml files with GBSA support.
"""
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH_GBSA.ffxml'))
def test_component_combination():
"""Test that a system still yields the same energy after building it again out of its components."""
# We've had issues where subsequent instances of a molecule might have zero charges
# Here we'll try to catch this (and also explicitly check the charges) by re-building
# a system out of its components
forcefield = ForceField(
get_data_filename('forcefield/Frosst_AlkEthOH.ffxml'))
filename = get_data_filename(
os.path.join('systems', 'packmol_boxes',
'cyclohexane_ethanol_0.4_0.6.pdb'))
from simtk.openmm.app import PDBFile
pdbfile = PDBFile(filename)
mol2files = [
get_data_filename(os.path.join('systems', 'monomers', 'ethanol.mol2')),
get_data_filename(
os.path.join('systems', 'monomers', 'cyclohexane.mol2'))
]
flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
mols = []
mol = oechem.OEMol()
for mol2file in mol2files:
ifs = oechem.oemolistream(mol2file)
ifs.SetFlavor(oechem.OEFormat_MOL2, flavor)
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
oechem.OETriposAtomNames(mol)
mols.append(oechem.OEGraphMol(mol))
# setup system
system = forcefield.createSystem(pdbfile.topology,
mols,
chargeMethod='OECharges_AM1BCCSym')
# Make parmed structure
structure = parmed.openmm.topsystem.load_topology(pdbfile.topology, system,
pdbfile.positions)
# Split the system, then re-compose it out of its components
tmp = structure.split()
strs, nums = [], []
for s, n in tmp:
strs.append(s)
nums.append(n)
nums = [len(n) for n in nums]
# Re-compose system from components
new_structure = strs[0] * nums[0]
for idx in range(1, len(nums)):
new_structure += strs[idx] * nums[idx]
# Swap in coordinates again
new_structure.positions = structure.positions
# Create System
newsys = new_structure.createSystem(nonbondedMethod=app.NoCutoff,
constraints=None,
implicitSolvent=None)
# Cross check energies
groups0, groups1, energy0, energy1 = compare_system_energies(
pdbfile.topology,
pdbfile.topology,
system,
newsys,
pdbfile.positions,
verbose=False)
# Also check that that the number of components is equal to the number I expect
if not len(nums) == 2:
print("Error: Test system has incorrect number of components.")
raise Exception(
'Incorrect number of components in cyclohexane/ethanol test system.'
)
# Also check that none of residues have zero charge
for resnr in range(len(structure.residues)):
abscharges = [
abs(structure.residues[resnr].atoms[idx].charge)
for idx in range(len(structure.residues[resnr].atoms))
]
if sum(abscharges) == 0:
raise Exception(
'Error: Residue %s in cyclohexane-ethanol test system has a charge of zero, which is incorrect.'
% resnr)
mol_filename = 'mol1.mol2'
# Define a few simulation parameters
time_step = 2*unit.femtoseconds # simulation timestep
temperature = 300*unit.kelvin # simulation temperature
friction = 1/unit.picosecond # collision rate
num_steps = 1000000 # number of steps to run
trj_freq = 1000 # number of steps per written trajectory frame
data_freq = 1000 # number of steps per written simulation statistics
# Load molecule and create pdb object
pdb = PDBFile(pdb_filename)
# Load a SMIRNOFF forcefield
#forcefield = ForceField(get_data_filename('forcefield/Frosst_AlkEthOH_parmAtFrosst.offxml'))
forcefield = ForceField(get_data_filename('forcefield/smirnoff99Frosst.offxml'))
# Load molecule using OpenEye tools
mol = oechem.OEGraphMol()
ifs = oechem.oemolistream(mol_filename)
# LPW: I don't understand the meaning of these lines.
# flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_MOL2_Default | oechem.OEIFlavor_MOL2_Forcefield
# ifs.SetFlavor( oechem.OEFormat_MOL2, flavor)
oechem.OEReadMolecule(ifs, mol)
oechem.OETriposAtomNames(mol)
pdbatoms = list(pdb.topology.atoms())
labels = forcefield.labelMolecules([mol])[0]
for key, val in labels.items():
print(key)