def test_jacs_ligands(self):
"""Use template generator to parameterize the Schrodinger JACS set of ligands"""
from simtk.openmm.app import ForceField, NoCutoff
jacs_systems = {
#'bace' : { 'prefix' : 'Bace' },
#'cdk2' : { 'prefix' : 'CDK2' },
'jnk1' : { 'prefix' : 'Jnk1' },
'mcl1' : { 'prefix' : 'MCL1' },
#'p38' : { 'prefix' : 'p38' },
'ptp1b' : { 'prefix' : 'PTP1B' },
'thrombin' : { 'prefix' : 'Thrombin' },
#'tyk2' : { 'prefix' : 'Tyk2' },
}
for system_name in jacs_systems:
prefix = jacs_systems[system_name]['prefix']
# Load molecules
ligand_sdf_filename = get_data_filename(os.path.join('perses_jacs_systems', system_name, prefix + '_ligands.sdf'))
print(f'Reading molecules from {ligand_sdf_filename} ...')
from openforcefield.topology import Molecule
molecules = Molecule.from_file(ligand_sdf_filename, allow_undefined_stereo=True)
# Ensure this is a list
try:
nmolecules = len(molecules)
except TypeError:
molecules = [molecules]
print(f'Read {len(molecules)} molecules from {ligand_sdf_filename}')
#molecules = self.filter_molecules(molecules)
MAX_MOLECULES = len(molecules)
if 'TRAVIS' in os.environ:
MAX_MOLECULES = 3
molecules = molecules[:MAX_MOLECULES]
print(f'{len(molecules)} molecules remain after filtering')
# Create template generator with local cache
cache = os.path.join(get_data_filename(os.path.join('perses_jacs_systems', system_name)), 'cache.json')
generator = self.TEMPLATE_GENERATOR(molecules=molecules, cache=cache)
# Create a ForceField
forcefield = ForceField()
# Register the template generator
forcefield.registerTemplateGenerator(generator.generator)
# Parameterize all molecules
print(f'Caching all molecules for {system_name} at {cache} ...')
n_success = 0
n_failure = 0
for molecule in molecules:
openmm_topology = molecule.to_topology().to_openmm()
try:
forcefield.createSystem(openmm_topology, nonbondedMethod=NoCutoff)
n_success += 1
except Exception as e:
n_failure += 1
print(e)
print(f'{n_failure}/{n_success+n_failure} ligands failed to parameterize for {system_name}')
def setUp(self):
self.testsystems = dict()
for (system_name, prefix) in [
# TODO: Uncomment these after we fix input files
#('bace', 'Bace'),
#('cdk1', 'CDK2'),
('jnk1', 'Jnk1'),
#('mcl1', 'MCL1'),
#('p38', 'p38'),
#('ptp1b', 'PTP1B'),
#('thrombin', 'Thrombin'),
#('tyk2', 'Tyk2'),
]:
# Load protein
from simtk.openmm.app import PDBFile
pdb_filename = get_data_filename(os.path.join('perses_jacs_systems', system_name, prefix + '_protein_fixed.pdb'))
pdbfile = PDBFile(pdb_filename)
# Load molecules
from openforcefield.topology import Molecule
sdf_filename = get_data_filename(os.path.join('perses_jacs_systems', system_name, prefix + '_ligands.sdf'))
molecules = Molecule.from_file(sdf_filename, allow_undefined_stereo=True)
print(f'Read {len(molecules)} molecules from {sdf_filename}')
# Filter molecules as appropriate
molecules = self.filter_molecules(molecules)
n_molecules = len(molecules)
print(f'{n_molecules} molecules remain after filtering')
if n_molecules == 0:
continue
# Create structures
import parmed
protein_structure = parmed.load_file(pdb_filename)
molecules_structure = parmed.load_file(sdf_filename)
complex_structures = [ (protein_structure + molecules_structure[index]) for index in range(n_molecules) ]
# Store
testsystem = {
'name' : system_name,
'protein_pdbfile' : pdbfile,
'molecules' : molecules,
'complex_structures' : complex_structures
}
self.testsystems[system_name] = testsystem
# TODO: Create other test topologies
# TODO: Protein-only
# TODO: Protein-ligand topology
# TODO: Solvated protein-ligand topology
# TODO: Host-guest topology
# Suppress DEBUG logging from various packages
import logging
for name in ['parmed', 'matplotlib']:
logging.getLogger(name).setLevel(logging.WARNING)
def test_complex(self):
"""Test parameterizing a protein:ligand complex in vacuum"""
from openmmforcefields.generators import SystemGenerator
for name, testsystem in self.testsystems.items():
print(f'Testing parameterization of {name} in vacuum')
molecules = testsystem['molecules']
# Select a complex from the set
ligand_index = 0
complex_structure = testsystem['complex_structures'][ligand_index]
molecule = molecules[ligand_index]
openmm_topology = complex_structure.topology
cache = os.path.join(get_data_filename(os.path.join('perses_jacs_systems', name)), 'cache.json')
# Create a system in vacuum
generator = SystemGenerator(forcefields=self.amber_forcefields,
molecules=molecules, cache=cache)
system = generator.create_system(openmm_topology)
assert system.getNumParticles() == len(complex_structure.atoms)
# Create solvated structure
from simtk.openmm import app
from simtk import unit
modeller = app.Modeller(complex_structure.topology, complex_structure.positions)
modeller.addSolvent(generator.forcefield, padding=0*unit.angstroms, ionicStrength=300*unit.millimolar)
# Create a system with solvent and ions
system = generator.create_system(modeller.topology)
assert system.getNumParticles() == len(list(modeller.topology.atoms()))
with open('test.pdb', 'w') as outfile:
app.PDBFile.writeFile(modeller.topology, modeller.positions, outfile)
def test_barostat(self):
"""Test that barostat addition works correctly"""
# Create a protein SystemGenerator
generator = SystemGenerator(forcefields=self.amber_forcefields)
# Create a template barostat
from simtk.openmm import MonteCarloBarostat
from simtk import unit
pressure = 0.95 * unit.atmospheres
temperature = 301.0 * unit.kelvin
frequency = 23
generator.barostat = MonteCarloBarostat(pressure, temperature,
frequency)
# Load a PDB file
import os
from simtk.openmm.app import PDBFile
pdb_filename = get_data_filename(
os.path.join('perses_jacs_systems', 'bace',
'Bace_protein_fixed.pdb'))
pdbfile = PDBFile(pdb_filename)
# Delete hydrogens from terminal protein residues
# TODO: Fix the input files so we don't need to do this
from simtk.openmm import app
modeller = app.Modeller(pdbfile.topology, pdbfile.positions)
residues = [
residue for residue in modeller.topology.residues()
if residue.name != 'UNL'
]
termini_ids = [residues[0].id, residues[-1].id]
#hs = [atom for atom in modeller.topology.atoms() if atom.element.symbol in ['H'] and atom.residue.name != 'UNL']
hs = [
atom for atom in modeller.topology.atoms()
if atom.element.symbol in ['H'] and atom.residue.id in termini_ids
]
modeller.delete(hs)
from simtk.openmm.app import PDBFile
modeller.addHydrogens()
# Create a System
system = generator.create_system(modeller.topology)
# Check barostat is present
forces = {
force.__class__.__name__: force
for force in system.getForces()
}
assert 'MonteCarloBarostat' in forces.keys()
# Check barostat parameters
force = forces['MonteCarloBarostat']
assert force.getDefaultPressure() == pressure
assert force.getDefaultTemperature() == temperature
assert force.getFrequency() == frequency
def setUp(self):
# TODO: Harmonize with test_system_generator.py infrastructure
# Read test molecules
from openforcefield.topology import Molecule
filename = get_data_filename("minidrugbank/MiniDrugBank-without-unspecified-stereochemistry.sdf")
molecules = Molecule.from_file(filename, allow_undefined_stereo=True)
# Filter molecules as appropriate
self.molecules = self.filter_molecules(molecules)
# Suppress DEBUG logging from various packages
import logging
for name in ['parmed', 'matplotlib']:
logging.getLogger(name).setLevel(logging.WARNING)
def test_jacs_complexes(self):
"""Use template generator to parameterize the Schrodinger JACS set of complexes"""
# TODO: Uncomment working systems when we have cleaned up the input files
jacs_systems = {
#'bace' : { 'prefix' : 'Bace' },
#'cdk2' : { 'prefix' : 'CDK2' },
#'jnk1' : { 'prefix' : 'Jnk1' },
'mcl1' : { 'prefix' : 'MCL1' },
#'p38' : { 'prefix' : 'p38' },
#'ptp1b' : { 'prefix' : 'PTP1B' },
#'thrombin' : { 'prefix' : 'Thrombin' },
#'tyk2' : { 'prefix' : 'Tyk2' },
}
for system_name in jacs_systems:
prefix = jacs_systems[system_name]['prefix']
# Read molecules
ligand_sdf_filename = get_data_filename(os.path.join('perses_jacs_systems', system_name, prefix + '_ligands.sdf'))
print(f'Reading molecules from {ligand_sdf_filename} ...')
from openforcefield.topology import Molecule
molecules = Molecule.from_file(ligand_sdf_filename, allow_undefined_stereo=True)
try:
nmolecules = len(molecules)
except TypeError:
molecules = [molecules]
print(f'Read {len(molecules)} molecules from {ligand_sdf_filename}')
# Read ParmEd Structures
import parmed
from simtk import unit
protein_pdb_filename = get_data_filename(os.path.join('perses_jacs_systems', system_name, prefix + '_protein.pdb'))
from simtk.openmm.app import PDBFile
print(f'Reading protein from {protein_pdb_filename} ...')
#protein_structure = parmed.load_file(protein_pdb_filename) # NOTE: This mis-interprets distorted geometry and sequentially-numbered residues that span chain breaks
pdbfile = PDBFile(protein_pdb_filename)
protein_structure = parmed.openmm.load_topology(pdbfile.topology, xyz=pdbfile.positions.value_in_unit(unit.angstroms))
ligand_structures = parmed.load_file(ligand_sdf_filename)
try:
nmolecules = len(ligand_structures)
except TypeError:
ligand_structures = [ligand_structures]
assert len(ligand_structures) == len(molecules)
# Filter molecules
if 'TRAVIS' in os.environ:
MAX_MOLECULES = 3
else:
MAX_MOLECULES = 6
molecules = molecules[:MAX_MOLECULES]
ligand_structures = ligand_structures[:MAX_MOLECULES]
print(f'{len(molecules)} molecules remain after filtering')
# Create complexes
complex_structures = [ (protein_structure + ligand_structure) for ligand_structure in ligand_structures ]
# Create template generator with local cache
cache = os.path.join(get_data_filename(os.path.join('perses_jacs_systems', system_name)), 'cache.json')
generator = self.TEMPLATE_GENERATOR(molecules=molecules, cache=cache)
# Create a ForceField
from simtk.openmm.app import ForceField
forcefield = ForceField(*self.amber_forcefields)
# Register the template generator
forcefield.registerTemplateGenerator(generator.generator)
# Parameterize all complexes
print(f'Caching all molecules for {system_name} at {cache} ...')
for ligand_index, complex_structure in enumerate(complex_structures):
openmm_topology = complex_structure.topology
molecule = molecules[ligand_index]
# Delete hydrogens from terminal protein residues
# TODO: Fix the input files so we don't need to do this
from simtk.openmm import app
modeller = app.Modeller(complex_structure.topology, complex_structure.positions)
residues = [residue for residue in modeller.topology.residues() if residue.name != 'UNL']
termini_ids = [residues[0].id, residues[-1].id]
#hs = [atom for atom in modeller.topology.atoms() if atom.element.symbol in ['H'] and atom.residue.name != 'UNL']
hs = [atom for atom in modeller.topology.atoms() if atom.element.symbol in ['H'] and atom.residue.id in termini_ids]
modeller.delete(hs)
from simtk.openmm.app import PDBFile
modeller.addHydrogens(forcefield)
# Parameterize protein:ligand complex in vacuum
print(f' Parameterizing {system_name} : {molecule.to_smiles()} in vacuum...')
from simtk.openmm.app import NoCutoff
forcefield.createSystem(modeller.topology, nonbondedMethod=NoCutoff)
# Parameterize protein:ligand complex in solvent
print(f' Parameterizing {system_name} : {molecule.to_smiles()} in explicit solvent...')
from simtk.openmm.app import PME
modeller.addSolvent(forcefield, padding=0*unit.angstroms, ionicStrength=300*unit.millimolar)
forcefield.createSystem(modeller.topology, nonbondedMethod=PME)
def setUp(self):
self.testsystems = dict()
for (system_name, prefix) in [
# TODO: Uncomment these after we fix input files
('bace', 'Bace'),
#('cdk1', 'CDK2'),
#('jnk1', 'Jnk1'),
#('mcl1', 'MCL1'),
#('p38', 'p38'),
#('ptp1b', 'PTP1B'),
#('thrombin', 'Thrombin'),
#('tyk2', 'Tyk2'),
]:
# Load protein
from simtk.openmm.app import PDBFile
pdb_filename = get_data_filename(
os.path.join('perses_jacs_systems', system_name,
prefix + '_protein.pdb'))
pdbfile = PDBFile(pdb_filename)
# Load molecules
from openforcefield.topology import Molecule
sdf_filename = get_data_filename(
os.path.join('perses_jacs_systems', system_name,
prefix + '_ligands_shifted.sdf'))
molecules = Molecule.from_file(sdf_filename,
allow_undefined_stereo=True)
print(f'Read {len(molecules)} molecules from {sdf_filename}')
n_molecules = len(molecules)
# Limit number of molecules for testing
MAX_MOLECULES = 10 if not CI else 2
if (n_molecules > MAX_MOLECULES):
print(f'Limiting to {MAX_MOLECULES} for testing...')
n_molecules = MAX_MOLECULES
molecules = [molecules[index] for index in range(n_molecules)]
# Create structures
import parmed
# NOTE: This does not work because parmed does not correctly assign bonds for HID
#protein_structure = parmed.load_file(pdb_filename)
# NOTE: This is the workaround
protein_structure = parmed.openmm.load_topology(
pdbfile.topology, xyz=pdbfile.positions)
molecules_structure = parmed.load_file(sdf_filename)
molecules_structure = [
molecules_structure[index] for index in range(n_molecules)
]
complex_structures = [
(molecules_structure[index] + protein_structure)
for index in range(n_molecules)
]
complex_structures = [
molecules_structure[index] for index in range(n_molecules)
] # DEBUG
# Store
testsystem = {
'name': system_name,
'protein_pdbfile': pdbfile,
'molecules': molecules,
'complex_structures': complex_structures
}
self.testsystems[system_name] = testsystem
# DEBUG
for name, testsystem in self.testsystems.items():
from simtk.openmm import app
filename = f'testsystem-{name}.pdb'
print(filename)
structure = testsystem['complex_structures'][0]
#structure.save(filename, overwrite=True)
with open(filename, 'w') as outfile:
app.PDBFile.writeFile(structure.topology,
structure.positions, outfile)
testsystem['molecules'][0].to_file(
f'testsystem-{name}-molecule.sdf', file_format="SDF")
testsystem['molecules'][0].to_file(
f'testsystem-{name}-molecule.pdb', file_format="PDB")
# TODO: Create other test topologies
# TODO: Protein-only
# TODO: Protein-ligand topology
# TODO: Solvated protein-ligand topology
# TODO: Host-guest topology
# Suppress DEBUG logging from various packages
import logging
for name in ['parmed', 'matplotlib']:
logging.getLogger(name).setLevel(logging.WARNING)
