def writeParameters(mol,
parameters,
qm,
method,
netcharge,
outdir,
original_coords=None):
paramDir = os.path.join(outdir, 'parameters', method.name,
_qm_method_name(qm))
os.makedirs(paramDir, exist_ok=True)
typemap = None
extensions = ('mol2', 'pdb', 'coor')
if method == FFTypeMethod.CGenFF_2b6:
extensions += ('psf', 'rtf', 'prm')
# TODO: remove?
f = open(os.path.join(paramDir, "input.namd"), "w")
tmp = '''parameters mol.prm
paraTypeCharmm on
coordinates mol.pdb
bincoordinates mol.coor
temperature 0
timestep 0
1-4scaling 1.0
exclude scaled1-4
outputname .out
outputenergies 1
structure mol.psf
cutoff 20.
switching off
stepsPerCycle 1
rigidbonds none
cellBasisVector1 50. 0. 0.
cellBasisVector2 0. 50. 0.
cellBasisVector3 0. 0. 50.
run 0'''
print(tmp, file=f)
f.close()
elif method in (FFTypeMethod.GAFF, FFTypeMethod.GAFF2):
# types need to be remapped because Amber FRCMOD format limits the type to characters
# writeFrcmod does this on the fly and returns a mapping that needs to be applied to the mol
# TODO: get rid of this mapping
frcFile = os.path.join(paramDir, 'mol.frcmod')
typemap = getAtomTypeMapping(parameters)
writeFRCMOD(mol, parameters, frcFile, typemap=typemap)
logger.info('Write FRCMOD file: %s' % frcFile)
tleapFile = os.path.join(paramDir, 'tleap.in')
with open(tleapFile, 'w') as file_:
file_.write('loadAmberParams mol.frcmod\n')
file_.write('A = loadMol2 mol.mol2\n')
file_.write('saveAmberParm A structure.prmtop mol.crd\n')
file_.write('quit\n')
logger.info('Write tleap input file: %s' % tleapFile)
# TODO: remove?
f = open(os.path.join(paramDir, "input.namd"), "w")
tmp = '''parmfile structure.prmtop
amber on
coordinates mol.pdb
bincoordinates mol.coor
temperature 0
timestep 0
1-4scaling 0.83333333
exclude scaled1-4
outputname .out
outputenergies 1
cutoff 20.
switching off
stepsPerCycle 1
rigidbonds none
cellBasisVector1 50. 0. 0.
cellBasisVector2 0. 50. 0.
cellBasisVector3 0. 0. 50.
run 0'''
print(tmp, file=f)
f.close()
else:
raise NotImplementedError
def remapAtomTypes(mol):
tmpmol = mol
if typemap is not None:
tmpmol = mol.copy()
tmpmol.atomtype[:] = [
typemap[atomtype] for atomtype in mol.atomtype
]
return tmpmol
tmpmol = remapAtomTypes(mol)
for ext in extensions:
file_ = os.path.join(paramDir, "mol." + ext)
if ext == 'prm':
writePRM(mol, parameters, file_)
elif ext == 'rtf':
writeRTF(mol, parameters, netcharge, file_)
else:
tmpmol.write(file_)
logger.info('Write %s file: %s' % (ext.upper(), file_))
if original_coords is not None:
molFile = os.path.join(paramDir, 'mol-orig.mol2')
tmpmol.coords = original_coords
tmpmol.write(molFile)
logger.info(
'Write MOL2 file (with original coordinates): {}'.format(molFile))
def main_parameterize(arguments=None):
args = getArgumentParser().parse_args(args=arguments)
if not os.path.exists(args.filename):
raise ValueError('File %s cannot be found' % args.filename)
method_map = {
'GAFF': FFTypeMethod.GAFF,
'GAFF2': FFTypeMethod.GAFF2,
'CGENFF': FFTypeMethod.CGenFF_2b6
}
methods = [method_map[method] for method in args.forcefield]
# Get RTF and PRM file names
rtfFile, prmFile = None, None
if args.rtf_prm:
rtfFile, prmFile = args.rtf_prm
# Create a queue for QM
if args.queue == 'local':
queue = LocalCPUQueue()
elif args.queue == 'Slurm':
queue = SlurmQueue(_configapp=args.code.lower())
elif args.queue == 'LSF':
queue = LsfQueue(_configapp=args.code.lower())
elif args.queue == 'PBS':
queue = PBSQueue() # TODO: configure
elif args.queue == 'AceCloud':
queue = AceCloudQueue() # TODO: configure
queue.groupname = args.groupname
queue.hashnames = True
else:
raise NotImplementedError
# Override default ncpus
if args.ncpus:
logger.info('Overriding ncpus to {}'.format(args.ncpus))
queue.ncpu = args.ncpus
if args.memory:
logger.info('Overriding memory to {}'.format(args.memory))
queue.memory = args.memory
# Create a QM object
if args.code == 'Psi4':
qm = Psi4()
elif args.code == 'Gaussian':
qm = Gaussian()
else:
raise NotImplementedError
# This is for debugging only!
if args.fake_qm:
qm = FakeQM2()
logger.warning('Using FakeQM')
# Get rotatable dihedral angles
mol = Molecule(args.filename)
mol = canonicalizeAtomNames(mol)
mol, equivalents, all_dihedrals = getEquivalentsAndDihedrals(mol)
netcharge = args.charge if args.charge is not None else int(
round(np.sum(mol.charge)))
if args.list:
print('\n === Parameterizable dihedral angles of {} ===\n'.format(
args.filename))
with open('torsions.txt', 'w') as fh:
for dihedral in all_dihedrals:
dihedral_name = '-'.join(mol.name[list(dihedral[0])])
print(' {}'.format(dihedral_name))
fh.write(dihedral_name + '\n')
print()
sys.exit(0)
# Set up the QM object
qm.theory = args.theory
qm.basis = args.basis
qm.solvent = args.environment
qm.queue = queue
qm.charge = netcharge
# Select which dihedrals to fit
parameterizable_dihedrals = [list(dih[0]) for dih in all_dihedrals]
if len(args.dihedral) > 0:
all_dihedral_names = [
'-'.join(mol.name[list(dihedral[0])]) for dihedral in all_dihedrals
]
parameterizable_dihedrals = []
for dihedral_name in args.dihedral:
if dihedral_name not in all_dihedral_names:
raise ValueError(
'%s is not recognized as a rotatable dihedral angle' %
dihedral_name)
parameterizable_dihedrals.append(
list(
all_dihedrals[all_dihedral_names.index(dihedral_name)][0]))
# Print arguments
print('\n === Arguments ===\n')
for key, value in vars(args).items():
print('{:>12s}: {:s}'.format(key, str(value)))
print('\n === Parameterizing %s ===\n' % args.filename)
for method in methods:
print(" === Fitting for %s ===\n" % method.name)
printReport(mol, netcharge, equivalents, all_dihedrals)
parameters, mol = fftype(mol,
method=method,
rtfFile=rtfFile,
prmFile=prmFile,
netcharge=args.charge)
if isinstance(qm, FakeQM2):
qm._parameters = parameters
# Copy the molecule to preserve initial coordinates
orig_coor = mol.coords.copy()
# Update B3LYP to B3LYP-D3
# TODO: this is silent and not documented stuff
if qm.theory == 'B3LYP':
qm.correction = 'D3'
# Update basis sets
# TODO: this is silent and not documented stuff
if netcharge < 0 and qm.solvent == 'vacuum':
if qm.basis == '6-31G*':
qm.basis = '6-31+G*'
if qm.basis == 'cc-pVDZ':
qm.basis = 'aug-cc-pVDZ'
logger.info('Changing basis sets to %s' % qm.basis)
# Minimize molecule
if args.minimize:
print('\n == Minimizing ==\n')
mol = minimize(mol, qm, args.outdir)
# Fit ESP charges
if args.fit_charges:
print('\n == Fitting ESP charges ==\n')
# Set random number generator seed
if args.seed:
np.random.seed(args.seed)
# Select the atoms with fixed charges
fixed_atom_indices = getFixedChargeAtomIndices(
mol, args.fix_charge)
# Fit ESP charges
mol, _, esp_charges, qm_dipole = fitCharges(
mol, qm, args.outdir, fixed=fixed_atom_indices)
# Print dipoles
logger.info('QM dipole: %f %f %f; %f' % tuple(qm_dipole))
mm_dipole = getDipole(mol)
if np.all(np.isfinite(mm_dipole)):
logger.info('MM dipole: %f %f %f; %f' % tuple(mm_dipole))
else:
logger.warning(
'MM dipole cannot be computed. Check if elements are detected correctly.'
)
# Fit dihedral angle parameters
if args.fit_dihedral:
print('\n == Fitting dihedral angle parameters ==\n')
# Set random number generator seed
if args.seed:
np.random.seed(args.seed)
# Invent new atom types for dihedral atoms
mol, originaltypes = inventAtomTypes(mol,
parameterizable_dihedrals,
equivalents)
parameters = recreateParameters(mol, originaltypes, parameters)
parameters = createMultitermDihedralTypes(parameters)
if isinstance(qm, FakeQM2):
qm._parameters = parameters
# Fit the parameters
fitDihedrals(mol,
qm,
method,
parameters,
all_dihedrals,
parameterizable_dihedrals,
args.outdir,
geomopt=args.optimize_dihedral)
# Output the FF parameters
print('\n == Writing results ==\n')
writeParameters(mol,
parameters,
qm,
method,
netcharge,
args.outdir,
original_coords=orig_coor)
# Write energy file
energyFile = os.path.join(args.outdir, 'parameters', method.name,
_qm_method_name(qm), 'energies.txt')
printEnergies(mol, parameters, energyFile)
logger.info('Write energy file: %s' % energyFile)
def _fit_charges(mol, args, qm, atom_types):
from htmd.charge import fitGasteigerCharges, fitChargesWithAntechamber, fitESPCharges, symmetrizeCharges
from htmd.parameterization.util import guessBondType, getFixedChargeAtomIndices, getDipole, _qm_method_name
from htmd.parameterization.detect import detectEquivalentAtoms
logger.info('=== Atomic charge fitting ===')
logger.info('Method: {}'.format(args.charge_type))
if args.charge_type == 'None':
# TODO move to argument validation
if len(args.fix_charge) > 0:
logger.warning('Flag --fix-charge does not have effect!')
logger.info('Atomic charges are taken from {}'.format(args.filename))
elif args.charge_type == 'Gasteiger':
# TODO move to argument validation
if len(args.fix_charge) > 0:
logger.warning('Flag --fix-charge does not have effect!')
# TODO move to _prepare_molecule
if np.any(mol.bondtype == "un"):
logger.info('Guessing bond types')
mol = guessBondType(mol)
mol = fitGasteigerCharges(mol, atom_types=atom_types)
charge = int(round(np.sum(mol.charge)))
if args.charge != charge:
raise RuntimeError(
'Molecular charge is set to {}, but Gasteiger atomic charges add up to {}.'
.format(args.charge, charge))
elif args.charge_type == 'AM1-BCC':
# TODO move to argument validation
if len(args.fix_charge) > 0:
logger.warning('Flag --fix-charge does not have effect!')
mol = fitChargesWithAntechamber(mol, type='bcc', molCharge=args.charge)
mol = symmetrizeCharges(mol)
elif args.charge_type == 'ESP':
# Detect equivalent atom groups
logger.info('Equivalent atom groups:')
atom_groups = [
group for group in detectEquivalentAtoms(mol)[0] if len(group) > 1
]
for atom_group in atom_groups:
logger.info(' {}'.format(', '.join(mol.name[list(atom_group)])))
# Select the atoms with fixed charges
fixed_atom_indices = getFixedChargeAtomIndices(mol, args.fix_charge)
# Create an ESP directory
espDir = os.path.join(args.outdir, "esp", _qm_method_name(qm))
os.makedirs(espDir, exist_ok=True)
charge = int(round(np.sum(mol.charge)))
if args.charge != charge:
logger.warning(
'Molecular charge is set to {}, but atomic charges add up to {}'
''.format(args.charge, charge))
if len(args.fix_charge) > 0:
raise RuntimeError(
'Flag --fix-charge cannot be used when atomic charges are inconsistent with passed '
'molecular charge {}'.format(args.charge))
mol.charge[:] = args.charge / mol.numAtoms
# Set random number generator seed
if args.seed:
np.random.seed(args.seed)
# Fit ESP charges
mol, extra = fitESPCharges(mol, qm, espDir, fixed=fixed_atom_indices)
# Print QM dipole
logger.info(
'QM dipole: {:6.3f} {:6.3f} {:6.3f}; total: {:6.3f}'.format(
*extra['qm_dipole']))
else:
raise ValueError()
# Print MM dipole
mm_dipole = getDipole(mol)
logger.info('MM dipole: {:6.3f} {:6.3f} {:6.3f}; total: {:6.3f}'.format(
*mm_dipole))
# Print the new charges
logger.info('Atomic charges:')
for name, charge in zip(mol.name, mol.charge):
logger.info(' {:4s}: {:6.3f}'.format(name, charge))
logger.info('Molecular charge: {:6.3f}'.format(np.sum(mol.charge)))
return mol
def main_parameterize(arguments=None):
args = getArgumentParser().parse_args(args=arguments)
if not os.path.exists(args.filename):
raise ValueError('File %s cannot be found' % args.filename)
method_map = {'GAFF': FFTypeMethod.GAFF, 'GAFF2': FFTypeMethod.GAFF2, 'CGENFF': FFTypeMethod.CGenFF_2b6}
methods = [method_map[method] for method in args.forcefield]
# Get RTF and PRM file names
rtfFile, prmFile = None, None
if args.rtf_prm:
rtfFile, prmFile = args.rtf_prm
# Create a queue for QM
if args.queue == 'local':
queue = LocalCPUQueue()
elif args.queue == 'Slurm':
queue = SlurmQueue(_configapp=args.code.lower())
elif args.queue == 'LSF':
queue = LsfQueue(_configapp=args.code.lower())
elif args.queue == 'PBS':
queue = PBSQueue() # TODO: configure
elif args.queue == 'AceCloud':
queue = AceCloudQueue() # TODO: configure
queue.groupname = args.groupname
queue.hashnames = True
else:
raise NotImplementedError
# Override default ncpus
if args.ncpus:
logger.info('Overriding ncpus to {}'.format(args.ncpus))
queue.ncpu = args.ncpus
if args.memory:
logger.info('Overriding memory to {}'.format(args.memory))
queue.memory = args.memory
# Create a QM object
if args.code == 'Psi4':
qm = Psi4()
elif args.code == 'Gaussian':
qm = Gaussian()
else:
raise NotImplementedError
# This is for debugging only!
if args.fake_qm:
qm = FakeQM2()
logger.warning('Using FakeQM')
# Get rotatable dihedral angles
mol = Molecule(args.filename)
mol = canonicalizeAtomNames(mol)
mol, equivalents, all_dihedrals = getEquivalentsAndDihedrals(mol)
netcharge = args.charge if args.charge is not None else int(round(np.sum(mol.charge)))
if args.list:
print('\n === Parameterizable dihedral angles of {} ===\n'.format(args.filename))
with open('torsions.txt', 'w') as fh:
for dihedral in all_dihedrals:
dihedral_name = '-'.join(mol.name[list(dihedral[0])])
print(' {}'.format(dihedral_name))
fh.write(dihedral_name+'\n')
print()
sys.exit(0)
# Set up the QM object
qm.theory = args.theory
qm.basis = args.basis
qm.solvent = args.environment
qm.queue = queue
qm.charge = netcharge
# Select which dihedrals to fit
parameterizable_dihedrals = [list(dih[0]) for dih in all_dihedrals]
if len(args.dihedral) > 0:
all_dihedral_names = ['-'.join(mol.name[list(dihedral[0])]) for dihedral in all_dihedrals]
parameterizable_dihedrals = []
for dihedral_name in args.dihedral:
if dihedral_name not in all_dihedral_names:
raise ValueError('%s is not recognized as a rotatable dihedral angle' % dihedral_name)
parameterizable_dihedrals.append(list(all_dihedrals[all_dihedral_names.index(dihedral_name)][0]))
# Print arguments
print('\n === Arguments ===\n')
for key, value in vars(args).items():
print('{:>12s}: {:s}'.format(key, str(value)))
print('\n === Parameterizing %s ===\n' % args.filename)
for method in methods:
print(" === Fitting for %s ===\n" % method.name)
printReport(mol, netcharge, equivalents, all_dihedrals)
parameters, mol = fftype(mol, method=method, rtfFile=rtfFile, prmFile=prmFile, netcharge=args.charge)
if isinstance(qm, FakeQM2):
qm._parameters = parameters
# Copy the molecule to preserve initial coordinates
orig_coor = mol.coords.copy()
# Update B3LYP to B3LYP-D3
# TODO: this is silent and not documented stuff
if qm.theory == 'B3LYP':
qm.correction = 'D3'
# Update basis sets
# TODO: this is silent and not documented stuff
if netcharge < 0 and qm.solvent == 'vacuum':
if qm.basis == '6-31G*':
qm.basis = '6-31+G*'
if qm.basis == 'cc-pVDZ':
qm.basis = 'aug-cc-pVDZ'
logger.info('Changing basis sets to %s' % qm.basis)
# Minimize molecule
if args.minimize:
print('\n == Minimizing ==\n')
mol = minimize(mol, qm, args.outdir)
# Fit ESP charges
if args.fit_charges:
print('\n == Fitting ESP charges ==\n')
# Set random number generator seed
if args.seed:
np.random.seed(args.seed)
# Select the atoms with fixed charges
fixed_atom_indices = getFixedChargeAtomIndices(mol, args.fix_charge)
# Fit ESP charges
mol, _, esp_charges, qm_dipole = fitCharges(mol, qm, args.outdir, fixed=fixed_atom_indices)
# Print dipoles
logger.info('QM dipole: %f %f %f; %f' % tuple(qm_dipole))
mm_dipole = getDipole(mol)
if np.all(np.isfinite(mm_dipole)):
logger.info('MM dipole: %f %f %f; %f' % tuple(mm_dipole))
else:
logger.warning('MM dipole cannot be computed. Check if elements are detected correctly.')
# Fit dihedral angle parameters
if args.fit_dihedral:
print('\n == Fitting dihedral angle parameters ==\n')
# Set random number generator seed
if args.seed:
np.random.seed(args.seed)
# Invent new atom types for dihedral atoms
mol, originaltypes = inventAtomTypes(mol, parameterizable_dihedrals, equivalents)
parameters = recreateParameters(mol, originaltypes, parameters)
parameters = createMultitermDihedralTypes(parameters)
if isinstance(qm, FakeQM2):
qm._parameters = parameters
# Fit the parameters
fitDihedrals(mol, qm, method, parameters, all_dihedrals, parameterizable_dihedrals, args.outdir, geomopt=args.optimize_dihedral)
# Output the FF parameters
print('\n == Writing results ==\n')
writeParameters(mol, parameters, qm, method, netcharge, args.outdir, original_coords=orig_coor)
# Write energy file
energyFile = os.path.join(args.outdir, 'parameters', method.name, _qm_method_name(qm), 'energies.txt')
printEnergies(mol, parameters, energyFile)
logger.info('Write energy file: %s' % energyFile)
def writeParameters(mol, parameters, qm, method, netcharge, outdir, original_coords=None):
paramDir = os.path.join(outdir, 'parameters', method.name, _qm_method_name(qm))
os.makedirs(paramDir, exist_ok=True)
typemap = None
extensions = ('mol2', 'pdb', 'coor')
if method == FFTypeMethod.CGenFF_2b6:
extensions += ('psf', 'rtf', 'prm')
# TODO: remove?
f = open(os.path.join(paramDir, "input.namd"), "w")
tmp = '''parameters mol.prm
paraTypeCharmm on
coordinates mol.pdb
bincoordinates mol.coor
temperature 0
timestep 0
1-4scaling 1.0
exclude scaled1-4
outputname .out
outputenergies 1
structure mol.psf
cutoff 20.
switching off
stepsPerCycle 1
rigidbonds none
cellBasisVector1 50. 0. 0.
cellBasisVector2 0. 50. 0.
cellBasisVector3 0. 0. 50.
run 0'''
print(tmp, file=f)
f.close()
elif method in (FFTypeMethod.GAFF, FFTypeMethod.GAFF2):
# types need to be remapped because Amber FRCMOD format limits the type to characters
# writeFrcmod does this on the fly and returns a mapping that needs to be applied to the mol
# TODO: get rid of this mapping
frcFile = os.path.join(paramDir, 'mol.frcmod')
typemap = getAtomTypeMapping(parameters)
writeFRCMOD(mol, parameters, frcFile, typemap=typemap)
logger.info('Write FRCMOD file: %s' % frcFile)
tleapFile = os.path.join(paramDir, 'tleap.in')
with open(tleapFile, 'w') as file_:
file_.write('loadAmberParams mol.frcmod\n')
file_.write('A = loadMol2 mol.mol2\n')
file_.write('saveAmberParm A structure.prmtop mol.crd\n')
file_.write('quit\n')
logger.info('Write tleap input file: %s' % tleapFile)
# TODO: remove?
f = open(os.path.join(paramDir, "input.namd"), "w")
tmp = '''parmfile structure.prmtop
amber on
coordinates mol.pdb
bincoordinates mol.coor
temperature 0
timestep 0
1-4scaling 0.83333333
exclude scaled1-4
outputname .out
outputenergies 1
cutoff 20.
switching off
stepsPerCycle 1
rigidbonds none
cellBasisVector1 50. 0. 0.
cellBasisVector2 0. 50. 0.
cellBasisVector3 0. 0. 50.
run 0'''
print(tmp, file=f)
f.close()
else:
raise NotImplementedError
def remapAtomTypes(mol):
tmpmol = mol
if typemap is not None:
tmpmol = mol.copy()
tmpmol.atomtype[:] = [typemap[atomtype] for atomtype in mol.atomtype]
return tmpmol
tmpmol = remapAtomTypes(mol)
for ext in extensions:
file_ = os.path.join(paramDir, "mol." + ext)
if ext == 'prm':
writePRM(mol, parameters, file_)
elif ext == 'rtf':
writeRTF(mol, parameters, netcharge, file_)
else:
tmpmol.write(file_)
logger.info('Write %s file: %s' % (ext.upper(), file_))
if original_coords is not None:
molFile = os.path.join(paramDir, 'mol-orig.mol2')
tmpmol.coords = original_coords
tmpmol.write(molFile)
logger.info('Write MOL2 file (with original coordinates): {}'.format(molFile))
