def _oemol_from_residue(res):
"""
Get an OEMol from a residue, even if that residue
is polymeric. In the latter case, external bonds
are replaced by hydrogens.
Parameters
----------
res : app.Residue
The residue in question
Returns
-------
oemol : openeye.oechem.OEMol
an oemol representation of the residue with topology indices
"""
import openeye.oechem as oechem
from openmoltools.forcefield_generators import generateOEMolFromTopologyResidue
external_bonds = list(res.external_bonds())
if external_bonds:
for bond in external_bonds:
res.chain.topology._bonds.remove(bond)
mol = generateOEMolFromTopologyResidue(res, geometry=False)
oechem.OEAddExplicitHydrogens(mol)
return mol
def test_generate_Topology_and_OEMol(self):
"""
Test round-trip from OEMol >> Topology >> OEMol
"""
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
from openeye import oechem, oeiupac
for molecule_name in IUPAC_molecule_names:
molecule1 = createOEMolFromIUPAC(molecule_name)
# Generate Topology from OEMol
topology = generateTopologyFromOEMol(molecule1)
# Check resulting Topology.
residues = [residue for residue in topology.residues()]
self.assertEqual(len(residues), 1)
self.assertEqual(residues[0].name, molecule1.GetTitle())
for (top_atom, mol_atom) in zip(topology.atoms(), molecule1.GetAtoms()):
self.assertEqual(top_atom.name, mol_atom.GetName())
for (top_bond, mol_bond) in zip(topology.bonds(), molecule1.GetBonds()):
self.assertEqual(top_bond[0].name, mol_bond.GetBgn().GetName())
self.assertEqual(top_bond[1].name, mol_bond.GetEnd().GetName())
# Generate OEMol from Topology
molecule2 = generateOEMolFromTopologyResidue(residues[0])
# Check resulting molecule.
self.assertEqual(molecule1.GetTitle(), molecule2.GetTitle())
for (atom1, atom2) in zip(molecule1.GetAtoms(), molecule2.GetAtoms()):
self.assertEqual(atom1.GetName(), atom2.GetName())
self.assertEqual(atom1.GetAtomicNum(), atom2.GetAtomicNum())
for (bond1, bond2) in zip(molecule1.GetBonds(), molecule2.GetBonds()):
self.assertEqual(bond1.GetBgn().GetName(), bond2.GetBgn().GetName())
self.assertEqual(bond1.GetEnd().GetName(), bond2.GetEnd().GetName())
def test_small_molecule_proposals():
"""
Make sure the small molecule proposal engine generates molecules
"""
from perses.rjmc import topology_proposal
from openmoltools import forcefield_generators
import openeye.oechem as oechem
list_of_smiles = ['CCCC','CCCCC','CCCCCC']
gaff_xml_filename = get_data_filename('data/gaff.xml')
stats_dict = {smiles : 0 for smiles in list_of_smiles}
system_generator = topology_proposal.SystemGenerator([gaff_xml_filename])
proposal_engine = topology_proposal.SmallMoleculeSetProposalEngine(list_of_smiles, system_generator)
initial_molecule = generate_initial_molecule('CCCC')
initial_system, initial_positions, initial_topology = oemol_to_omm_ff(initial_molecule, "MOL")
proposal = proposal_engine.propose(initial_system, initial_topology)
for i in range(50):
#positions are ignored here, and we don't want to run the geometry engine
new_proposal = proposal_engine.propose(proposal.old_system, proposal.old_topology)
stats_dict[new_proposal.new_chemical_state_key] += 1
#check that the molecule it generated is actually the smiles we expect
matching_molecules = [res for res in proposal.new_topology.residues() if res.name=='MOL']
if len(matching_molecules) != 1:
raise ValueError("More than one residue with the same name!")
mol_res = matching_molecules[0]
oemol = forcefield_generators.generateOEMolFromTopologyResidue(mol_res)
assert oechem.OEMolToSmiles(oemol) == proposal.new_chemical_state_key
proposal = new_proposal
def test_small_molecule_proposals():
"""
Make sure the small molecule proposal engine generates molecules
"""
from perses.rjmc import topology_proposal
from openmoltools import forcefield_generators
import openeye.oechem as oechem
list_of_smiles = ['CCCC','CCCCC','CCCCCC']
gaff_xml_filename = get_data_filename('data/gaff.xml')
stats_dict = {smiles : 0 for smiles in list_of_smiles}
system_generator = topology_proposal.SystemGenerator([gaff_xml_filename])
proposal_engine = topology_proposal.SmallMoleculeSetProposalEngine(list_of_smiles, system_generator)
initial_molecule = generate_initial_molecule('CCCC')
initial_system, initial_positions, initial_topology = oemol_to_omm_ff(initial_molecule, "MOL")
proposal = proposal_engine.propose(initial_system, initial_topology)
for i in range(50):
#positions are ignored here, and we don't want to run the geometry engine
new_proposal = proposal_engine.propose(proposal.old_system, proposal.old_topology)
stats_dict[new_proposal.new_chemical_state_key] += 1
#check that the molecule it generated is actually the smiles we expect
matching_molecules = [res for res in proposal.new_topology.residues() if res.name=='MOL']
if len(matching_molecules) != 1:
raise ValueError("More than one residue with the same name!")
mol_res = matching_molecules[0]
oemol = forcefield_generators.generateOEMolFromTopologyResidue(mol_res)
assert oechem.OEMolToSmiles(oemol) == proposal.new_chemical_state_key
proposal = new_proposal
def test_small_molecule_proposals():
"""
Make sure the small molecule proposal engine generates molecules
"""
list_of_smiles = ['CCCC','CCCCC','CCCCCC']
list_of_mols = []
for smi in list_of_smiles:
mol = smiles_to_oemol(smi)
list_of_mols.append(mol)
molecules = [Molecule.from_openeye(mol) for mol in list_of_mols]
stats_dict = defaultdict(lambda: 0)
system_generator = SystemGenerator(forcefields = forcefield_files, barostat=barostat, forcefield_kwargs=forcefield_kwargs, nonperiodic_forcefield_kwargs=nonperiodic_forcefield_kwargs,
small_molecule_forcefield = small_molecule_forcefield, molecules=molecules, cache=None)
proposal_engine = topology_proposal.SmallMoleculeSetProposalEngine(list_of_mols, system_generator)
initial_system, initial_positions, initial_topology, = OEMol_to_omm_ff(list_of_mols[0], system_generator)
proposal = proposal_engine.propose(initial_system, initial_topology)
for i in range(50):
#positions are ignored here, and we don't want to run the geometry engine
new_proposal = proposal_engine.propose(proposal.old_system, proposal.old_topology)
stats_dict[new_proposal.new_chemical_state_key] += 1
#check that the molecule it generated is actually the smiles we expect
matching_molecules = [res for res in proposal.new_topology.residues() if res.name=='MOL']
if len(matching_molecules) != 1:
raise ValueError("More than one residue with the same name!")
mol_res = matching_molecules[0]
oemol = generateOEMolFromTopologyResidue(mol_res)
smiles = SmallMoleculeSetProposalEngine.canonicalize_smiles(oechem.OEMolToSmiles(oemol))
assert smiles == proposal.new_chemical_state_key
proposal = new_proposal
def _oemol_from_residue(res):
"""
Get an OEMol from a residue, even if that residue
is polymeric. In the latter case, external bonds
are replaced by hydrogens.
Parameters
----------
res : app.Residue
The residue in question
Returns
-------
oemol : openeye.oechem.OEMol
an oemol representation of the residue with topology indices
"""
import openeye.oechem as oechem
from openmoltools.forcefield_generators import generateOEMolFromTopologyResidue
external_bonds = list(res.external_bonds())
if external_bonds:
for bond in external_bonds:
res.chain.topology._bonds.remove(bond)
mol = generateOEMolFromTopologyResidue(res, geometry=False)
oechem.OEAddExplicitHydrogens(mol)
return mol
def canonicalize_SMILES(smiles_list):
"""Ensure all SMILES strings end up in canonical form.
Stereochemistry must already have been expanded.
SMILES strings are converted to a OpenEye Topology and back again.
Parameters
----------
smiles_list : list of str
List of SMILES strings
Returns
-------
canonical_smiles_list : list of str
List of SMILES strings, after canonicalization.
"""
# Round-trip each molecule to a Topology to end up in canonical form
from openmoltools.forcefield_generators import generateOEMolFromTopologyResidue, generateTopologyFromOEMol
from openeye import oechem
canonical_smiles_list = list()
for smiles in smiles_list:
molecule = smiles_to_oemol(smiles)
topology = generateTopologyFromOEMol(molecule)
residues = [ residue for residue in topology.residues() ]
new_molecule = generateOEMolFromTopologyResidue(residues[0])
new_smiles = oechem.OECreateIsoSmiString(new_molecule)
canonical_smiles_list.append(new_smiles)
return canonical_smiles_list
def canonicalize_SMILES(smiles_list):
"""Ensure all SMILES strings end up in canonical form.
Stereochemistry must already have been expanded.
SMILES strings are converted to a OpenEye Topology and back again.
Parameters
----------
smiles_list : list of str
List of SMILES strings
Returns
-------
canonical_smiles_list : list of str
List of SMILES strings, after canonicalization.
"""
# Round-trip each molecule to a Topology to end up in canonical form
from openmoltools.forcefield_generators import generateOEMolFromTopologyResidue, generateTopologyFromOEMol
from perses.utils.openeye import smiles_to_oemol
from openeye import oechem
canonical_smiles_list = list()
for smiles in smiles_list:
molecule = smiles_to_oemol(smiles)
topology = generateTopologyFromOEMol(molecule)
residues = [ residue for residue in topology.residues() ]
new_molecule = generateOEMolFromTopologyResidue(residues[0])
new_smiles = oechem.OECreateIsoSmiString(new_molecule)
canonical_smiles_list.append(new_smiles)
return canonical_smiles_list
def generate_vacuum_hostguest_proposal(current_mol_name="B2", proposed_mol_name="MOL"):
"""
Generate a test vacuum topology proposal, current positions, and new positions triplet
from two IUPAC molecule names.
Parameters
----------
current_mol_name : str, optional
name of the first molecule
proposed_mol_name : str, optional
name of the second molecule
Returns
-------
topology_proposal : perses.rjmc.topology_proposal
The topology proposal representing the transformation
current_positions : np.array, unit-bearing
The positions of the initial system
new_positions : np.array, unit-bearing
The positions of the new system
"""
from openmoltools import forcefield_generators
from openmmtools import testsystems
from perses.tests.utils import createOEMolFromIUPAC, createSystemFromIUPAC, get_data_filename
host_guest = testsystems.HostGuestVacuum()
unsolv_old_system, pos_old, top_old = host_guest.system, host_guest.positions, host_guest.topology
ligand_topology = [res for res in top_old.residues()]
current_mol = forcefield_generators.generateOEMolFromTopologyResidue(ligand_topology[1]) # guest is second residue in topology
proposed_mol = createOEMolFromSMILES('C1CC2(CCC1(CC2)C)C')
initial_smiles = oechem.OEMolToSmiles(current_mol)
final_smiles = oechem.OEMolToSmiles(proposed_mol)
gaff_xml_filename = get_data_filename("data/gaff.xml")
forcefield = app.ForceField(gaff_xml_filename, 'tip3p.xml')
forcefield.registerTemplateGenerator(forcefield_generators.gaffTemplateGenerator)
solvated_system = forcefield.createSystem(top_old, removeCMMotion=False)
gaff_filename = get_data_filename('data/gaff.xml')
system_generator = SystemGenerator([gaff_filename, 'amber99sbildn.xml', 'tip3p.xml'], forcefield_kwargs={'removeCMMotion': False, 'nonbondedMethod': app.NoCutoff})
geometry_engine = geometry.FFAllAngleGeometryEngine()
proposal_engine = SmallMoleculeSetProposalEngine(
[initial_smiles, final_smiles], system_generator, residue_name=current_mol_name)
#generate topology proposal
topology_proposal = proposal_engine.propose(solvated_system, top_old, current_mol=current_mol, proposed_mol=proposed_mol)
#generate new positions with geometry engine
new_positions, _ = geometry_engine.propose(topology_proposal, pos_old, beta)
return topology_proposal, pos_old, new_positions
def write_xml(topology, file_name):
residues = [residue for residue in topology.residues()]
residue = residues[0]
molOE = generateOEMolFromTopologyResidue(residue,
geometry=False,
tripos_atom_names=True)
molOE.SetTitle('MOL')
ffxml = generateForceFieldFromMolecules([molOE])
f = open(file_name, 'w')
f.write(ffxml)
f.close()
def test_topology_molecules_round_trip():
"""
Test round-trips between OEMol and Topology
"""
# Create a test set of molecules.
molecules = [ createOEMolFromIUPAC(name) for name in IUPAC_molecule_names ]
# Test round-trips.
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
for molecule in molecules:
# Create topology from molecule.
topology = generateTopologyFromOEMol(molecule)
# Create molecule from topology.
residues = [residue for residue in topology.residues()]
molecule2 = generateOEMolFromTopologyResidue(residues[0])
# Create topology form molecule.
topology2 = generateTopologyFromOEMol(molecule2)
# Create molecule from topology with geometry.
residues2 = [residue for residue in topology2.residues()]
molecule3 = generateOEMolFromTopologyResidue(residues2[0], geometry=True)
# Create molecule from topology with Tripos atom names
molecule4 = generateOEMolFromTopologyResidue(residues2[0], tripos_atom_names=True)
def generate_ffxml(pdb_filename):
from simtk.openmm.app import PDBFile, Modeller
pdbfile = PDBFile(pdb_filename)
residues = [ residue for residue in pdbfile.topology.residues() ]
residue = residues[0]
from openmoltools.forcefield_generators import generateForceFieldFromMolecules, generateOEMolFromTopologyResidue
molecule = generateOEMolFromTopologyResidue(residue, geometry=False, tripos_atom_names=True)
molecule.SetTitle('MOL')
molecules = [molecule]
ffxml = generateForceFieldFromMolecules(molecules)
outfile = open('imatinib.xml', 'w')
outfile.write(ffxml)
outfile.close()
def test_atom_topology_index():
"""
Make sure that generateOEMolFromTopologyResidue adds the topology_index data
"""
# Create a test set of molecules.
molecules = [ createOEMolFromIUPAC(name) for name in IUPAC_molecule_names ]
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
topologies = [generateTopologyFromOEMol(molecule) for molecule in molecules]
for topology in topologies:
residue = list(topology.residues())[0] #there is only one residue
regenerated_mol = generateOEMolFromTopologyResidue(residue)
for i, top_atom in enumerate(topology.atoms()):
oeatom = regenerated_mol.GetAtom(oechem.OEHasAtomIdx(top_atom.index))
assert oeatom.GetData("topology_index")==top_atom.index
def test_topology_molecules_round_trip():
"""
Test round-trips between OEMol and Topology
"""
# Create a test set of molecules.
molecules = [createOEMolFromIUPAC(name) for name in IUPAC_molecule_names]
# Test round-trips.
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
for molecule in molecules:
# Create topology from molecule.
topology = generateTopologyFromOEMol(molecule)
# Create molecule from topology.
residues = [residue for residue in topology.residues()]
molecule2 = generateOEMolFromTopologyResidue(residues[0])
# Create topology form molecule.
topology2 = generateTopologyFromOEMol(molecule2)
# Create molecule from topology with geometry.
residues2 = [residue for residue in topology2.residues()]
molecule3 = generateOEMolFromTopologyResidue(residues2[0],
geometry=True)
# Create molecule from topology with Tripos atom names
molecule4 = generateOEMolFromTopologyResidue(residues2[0],
tripos_atom_names=True)
def generate_ffxml(pdb_filename):
from simtk.openmm.app import PDBFile, Modeller
pdbfile = PDBFile(pdb_filename)
residues = [residue for residue in pdbfile.topology.residues()]
residue = residues[0]
from openmoltools.forcefield_generators import generateForceFieldFromMolecules, generateOEMolFromTopologyResidue
molecule = generateOEMolFromTopologyResidue(residue,
geometry=False,
tripos_atom_names=True)
molecule.SetTitle('MOL')
molecules = [molecule]
ffxml = generateForceFieldFromMolecules(molecules)
outfile = open('imatinib.xml', 'w')
outfile.write(ffxml)
outfile.close()
def test_atom_topology_index():
"""
Make sure that generateOEMolFromTopologyResidue adds the topology_index data
"""
# Create a test set of molecules.
molecules = [createOEMolFromIUPAC(name) for name in IUPAC_molecule_names]
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
topologies = [
generateTopologyFromOEMol(molecule) for molecule in molecules
]
for topology in topologies:
residue = list(topology.residues())[0] #there is only one residue
regenerated_mol = generateOEMolFromTopologyResidue(residue)
for i, top_atom in enumerate(topology.atoms()):
oeatom = regenerated_mol.GetAtom(
oechem.OEHasAtomIdx(top_atom.index))
assert oeatom.GetData("topology_index") == top_atom.index
def test_generate_Topology_and_OEMol(self):
"""
Test round-trip from OEMol >> Topology >> OEMol
"""
from openmoltools.forcefield_generators import generateTopologyFromOEMol, generateOEMolFromTopologyResidue
from openeye import oechem, oeiupac
for molecule_name in IUPAC_molecule_names:
molecule1 = createOEMolFromIUPAC(molecule_name)
# Generate Topology from OEMol
topology = generateTopologyFromOEMol(molecule1)
# Check resulting Topology.
residues = [residue for residue in topology.residues()]
self.assertEqual(len(residues), 1)
self.assertEqual(residues[0].name, molecule1.GetTitle())
for (top_atom, mol_atom) in zip(topology.atoms(),
molecule1.GetAtoms()):
self.assertEqual(top_atom.name, mol_atom.GetName())
for (top_bond, mol_bond) in zip(topology.bonds(),
molecule1.GetBonds()):
self.assertEqual(top_bond[0].name, mol_bond.GetBgn().GetName())
self.assertEqual(top_bond[1].name, mol_bond.GetEnd().GetName())
# Generate OEMol from Topology
molecule2 = generateOEMolFromTopologyResidue(residues[0])
# Check resulting molecule.
self.assertEqual(molecule1.GetTitle(), molecule2.GetTitle())
for (atom1, atom2) in zip(molecule1.GetAtoms(),
molecule2.GetAtoms()):
self.assertEqual(atom1.GetName(), atom2.GetName())
self.assertEqual(atom1.GetAtomicNum(), atom2.GetAtomicNum())
for (bond1, bond2) in zip(molecule1.GetBonds(),
molecule2.GetBonds()):
self.assertEqual(bond1.GetBgn().GetName(),
bond2.GetBgn().GetName())
self.assertEqual(bond1.GetEnd().GetName(),
bond2.GetEnd().GetName())
def generate_vacuum_hostguest_proposal(current_mol_name="B2",
proposed_mol_name="MOL"):
"""
Generate a test vacuum topology proposal, current positions, and new positions triplet
from two IUPAC molecule names.
Parameters
----------
current_mol_name : str, optional
name of the first molecule
proposed_mol_name : str, optional
name of the second molecule
Returns
-------
topology_proposal : perses.rjmc.topology_proposal
The topology proposal representing the transformation
current_positions : np.array, unit-bearing
The positions of the initial system
new_positions : np.array, unit-bearing
The positions of the new system
"""
from openmoltools import forcefield_generators
from openmmtools import testsystems
from perses.utils.openeye import smiles_to_oemol
from perses.utils.data import get_data_filename
host_guest = testsystems.HostGuestVacuum()
unsolv_old_system, old_positions, top_old = host_guest.system, host_guest.positions, host_guest.topology
ligand_topology = [res for res in top_old.residues()]
current_mol = forcefield_generators.generateOEMolFromTopologyResidue(
ligand_topology[1]) # guest is second residue in topology
proposed_mol = smiles_to_oemol('C1CC2(CCC1(CC2)C)C')
initial_smiles = oechem.OEMolToSmiles(current_mol)
final_smiles = oechem.OEMolToSmiles(proposed_mol)
gaff_xml_filename = get_data_filename("data/gaff.xml")
forcefield = app.ForceField(gaff_xml_filename, 'tip3p.xml')
forcefield.registerTemplateGenerator(
forcefield_generators.gaffTemplateGenerator)
solvated_system = forcefield.createSystem(top_old, removeCMMotion=False)
gaff_filename = get_data_filename('data/gaff.xml')
system_generator = SystemGenerator(
[gaff_filename, 'amber99sbildn.xml', 'tip3p.xml'],
forcefield_kwargs={
'removeCMMotion': False,
'nonbondedMethod': app.NoCutoff
})
geometry_engine = geometry.FFAllAngleGeometryEngine()
proposal_engine = SmallMoleculeSetProposalEngine(
[initial_smiles, final_smiles],
system_generator,
residue_name=current_mol_name)
#generate topology proposal
topology_proposal = proposal_engine.propose(solvated_system,
top_old,
current_mol=current_mol,
proposed_mol=proposed_mol)
#generate new positions with geometry engine
new_positions, _ = geometry_engine.propose(topology_proposal,
old_positions, beta)
return topology_proposal, old_positions, new_positions