OPLSPotentialBuilder

class OPLSPotentialBuilder(use_bonds=None, include_bond_stretching=None, include_bond_angles=None, include_torsions=None, include_inversions=None, include_lennard_jones=None, include_electrostatic=None, lennard_jones_cutoff=None, lennard_jones_smoothing_length=None, electrostatic_pairwise_cutoff=None, electrostatic_smoothing_length=None, electrostatic_spme_cutoff=None, electrostatic_accuracy=None, electrostatic_scale=None, use_electrostatic_direct_sum=None, ignore_missing_types=None)

This object builds a OPLS potential using TremoloX.

Parameters:

• use_bonds (bool) – Controls if any bonding (stretching, angle, torsion and inversion) terms are to be included.
  Default: True.

• include_bond_stretching (bool) – Controls if bonding stretching terms are included.
  Default: use_bonds.

• include_bond_angles (bool) – Controls if angle bending terms are included.
  Default: use_bonds.

• include_torsions (bool) – Controls if bond torsion terms are included.
  Default: use_bonds.

• include_inversions (bool) – Controls if bond inversion torsion terms are included.
  Default: use_bonds.

• include_lennard_jones (bool) – Controls if Lennard-Jones terms are included.
  Default: use_bonds.

• include_electrostatic (bool) – Controls if electrostatic terms are included.
  Default: use_bonds.

• lennard_jones_cutoff (PhysicalQuantity of type distance) – Distance beyond which the Lennard-Jones terms are truncated.
  Default: 10.0 * Angstrom.

• lennard_jones_smoothing_length (PhysicalQuantity of type distance) – Distance over which the Lennard-Jones terms are brought to zero at the cutoff.
  Default: 2.0 * Angstrom.

• electrostatic_pairwise_cutoff (PhysicalQuantity of type distance) – Distance at which the pairwise electrostatic terms are brought to zero in pairwise summation.
  Default: 12 * Angstrom

• electrostatic_smoothing_length (PhysicalQuantity of type distance) – Distance over which the electrostatic terms are brought to zero at the cutoff when using the direct electrostatic sum.
  Default: 2.0 * Angstrom

• electrostatic_spme_cutoff (PhysicalQuantity of type distance) – Distance at which the pairwise electrostatic terms are brought to zero in SPME summation.
  Default: 7.5 * Angstrom

• electrostatic_accuracy (float) – Sets the relative accuracy of the SPME summation.
  Default: 0.0001

• electrostatic_scale (float) – Set the relative strength of the electrostatic terms.
  Default: 1.0

• use_electrostatic_direct_sum (bool) – Use the direct Coulomb sum to calculate the electrostatic terms in MoleculeConfigurations.
  Default: False

• ignore_missing_types (bool) – Whether or not molecules missing OPLS types are ignored.
  Default: False

assignAtomTypes(configuration, search_types=None, overwrite=False)

Automatically assign OPLS atom tags to a configuration.

Parameters:

• configuration (AtomicConfiguration) – The configuration on which to assign tags.

• search_types (list of str | None) – List of types to assign onto the configuration. If specified only these types will be assigned, with OPLS_UNKNOWN assigned to atoms that don’t match any of the given types. The default argument None means that all available types are assigned.
  Default: None.

• overwrite (bool) – Whether or not to assign new types if the configuration already has a valid set of types.
  Default: False

checkAtomTypes(configuration, add_missing_terms=False)

Function that tests that each atom has only one valid OPLS type.

Parameters:

• configuration (AtomicConfiguration) – The configuration that is being tested.

• add_missing_terms (bool) – Whether or not to add terms for missing OPLS types.
  Default: False

createCalculator(configuration, check_types=True, verbose=False, set_improper_torsion_indices=True, atomic_charges=None, charge_averaging=None)

Create and return a calculator with the OPLS potential.

Parameters:

• configuration (AtomicConfiguration) – The configuration the potential is to be used with.

• check_types (bool) – Whether or not the types are checked for basic consistency.

• verbose (bool) – Whether or not to print the terms assigned in the potential set.
  Default: False

• set_improper_torsion_indices (bool) – Whether or not to add improper torsions onto the input configuration.
  Default: True

• atomic_charges (PhysicalQuantity of type charge) – Custom atomic charges to be used in the potential.

• charge_averaging (NLFlag | None) – The type of averaging of charges to be used, if any. Accepted values are None, AtomConnectivity to average over atoms bonded to the same elements and AtomTypes to average over atoms of the same forcefield type.
  Default: None

Returns:

The calculator containing the potential.

Return type:

TremoloXCalculator

electrostaticAccuracy()

Returns:

The relative accuracy of the electrostatic summation for SPME.

Return type:

float

electrostaticPairwiseCutoff()

Returns:

The cutoff used in pairwise electrostatic summation.

Return type:

PhysicalQuantity of type length

electrostaticScale()

Returns:

Scaling factor for the electrostatic terms.

Return type:

float

electrostaticSmoothingLength()

Returns:

Distance over which electrostatic terms are splined when using direct summation.

Return type:

PhysicalQuantity of type length

electrostaticSpmeCutoff()

Returns:

The cutoff used in SPME electrostatic summation.

Return type:

PhysicalQuantity of type length

ignoreMissingTypes()

Returns:

Whether or not missing types are ignored on molecules.

Return type:

bool

includeBondAngles()

Returns:

Whether or not bond angle terms are included in the potential set.

Return type:

bool

includeBondStretching()

Returns:

Whether or not bond stretching terms are included in the potential set.

Return type:

bool

includeElectrostatic()

Returns:

Whether or not electrostatic terms are included in the potential set.

Return type:

bool

includeInversions()

Returns:

Whether or not bond inversion terms are included in the potential set.

Return type:

bool

includeLennardJones()

Returns:

Whether or not Lennard Jones terms are included in the potential set.

Return type:

bool

includeTorsions()

Returns:

Whether or not bond torsion terms are included in the potential set.

Return type:

bool

isOplsType(type_label)

Function that tests if a type is in the potential.

Parameters:

type_label (str) – The type being tested.

Returns:

Whether or not the tag is defined in the potential.

Return type:

bool

lennardJonesCutoff()

Returns:

Distance beyond which Lennard Jones terms are truncated.

Return type:

PhysicalQuantity of type length

lennardJonesSmoothingLength()

Returns:

Distance over which Lennard Jones terms are splined.

Return type:

PhysicalQuantity of type length

oplsType(index, configuration, add_missing_terms=False)

Gives the OPLS type on an atom in a configuration, ignoring other tags.

Parameters:

• index (int) – Index of the atom for which the type is sought.

• configuration (AtomicConfiguration) – The configuration containing the atom.

• add_missing_terms (bool) – Whether or not to add terms for missing OPLS types.
  Default: False

Returns:

The type of the atom.

Return type:

str

printTotalTypes(type_class=None)

Print a table of available atom types. This provides a list of types and descriptions which can be useful when doing manual type assignment.

Parameters:

type_class (str | None) – Display only atom types with this class label. None displays all types.
Default: None

printUsedTypes(configuration)

Print a table of the types used on the configuration.

Parameters:

configuration (AtomicConfiguration) – The configuration whose types are being displayed.

totalCharge(configuration, verbose=False)

Return the total charge of the configuration according to the atom types on the molecule.

Parameters:

• configuration (AtomicConfiguration) – The configuration whose charges are being reported.

• verbose (bool) – Flag as to whether or not additional information about charges is printed.
  Default: False

Returns:

The total charge of the configuration according to the atom types on the molecule.

Return type:

PhysicalQuantity with units of charge.

typePartialCharge(type_label)

Return the atomic partial charge associated with the given OPLS type label.

Parameters:

type_label (str) – The type label for the OPLS atom type.

Returns:

The atom charge associated with the type label.

Return type:

PhysicalQuantity of type charge

uniqueString()

Return a unique string representing the state of the object.

useBonds()

Returns:

Whether or not bond terms are included in the potential set.

Return type:

bool

useElectrostaticDirectSum()

Returns:

Whether or not direct Coulomb sums are used for molecules.

Return type:

bool

Usage Examples

Build a bulk configuration of poly(vinyl chloride) plasticized with dimethyl sebacate using the OPLS potential.

# ---------------------------------------------
# Monomer And Included Molecules
# ---------------------------------------------
PVC_monomer = nlread('PVC_Monomer.hdf5')[-1]
dimthyl_sebacate = nlread('Dimethyl_Sebacate.hdf5')[-1]

# -------------------------------------------------------------
# OPLS Potential Definition
# -------------------------------------------------------------
opls_potential_builder = OPLSPotentialBuilder()

# Display the tags and charges placed on the dimethyl sebacate molecule
opls_potential_builder.printUsedTypes(dimthyl_sebacate)
opls_potential_builder.totalCharge(dimthyl_sebacate, verbose=True)

# ---------------------------------------------
# Polymer Sequence
# ---------------------------------------------
polymer_sequence = PolymerSequence(
    monomer_configurations=[PVC_monomer],
    number_of_monomers=20,
    number_of_chains=20,
    tactic_ratio=0.5,
)

# -------------------------------------------------------------
# Polymer Monte Carlo Builder
# -------------------------------------------------------------
polymer_builder = PolymerMonteCarloBuilder(
    polymer_sequence=[polymer_sequence],
    density=1.4000*gram/cm**3,
    included_molecules=[(dimthyl_sebacate, 4)],
    monte_carlo_temperature=300.0*Kelvin,
    angle_sampling_points=20,
    repack_molecules=False,
)
bulk_configuration = polymer_builder.buildPolymerConfiguration()

# -------------------------------------------------------------
# Add OPLS Potential To Polymer Configuration
# -------------------------------------------------------------
calculator = opls_potential_builder.createCalculator(bulk_configuration)
bulk_configuration.setCalculator(calculator)
nlsave('PVC_DMS.hdf5', bulk_configuration)

# -------------------------------------------------------------
# Force Capped Equilibration
# -------------------------------------------------------------
equilibration_method = ForceCappedEquilibration(
    temperature=300.00*Kelvin,
    md_steps_per_force_capped_simulation=40000,
    md_time_step=0.50*fs,
    force_capped_simulations=4,
    starting_factor=1.040,
    ending_factor=0.800,
    fixed_indices=None,
)
bulk_configuration = equilibration_method.runEquilibration(bulk_configuration)
nlsave('PVC_DMS.hdf5', bulk_configuration)

OPLS_PVC_DMS_Example.py PVC_Monomer.hdf5 Dimethyl_Sebacate.hdf5

The given example above demonstrates both how to create a calculator that uses the OPLS-AA potential and also how to assign tags for the dimethyl sebacate molecule that is included in the polymer melt. Both the atom types and charges are printed. This shows both the appropriate types that have been selected and also that the overall charge of the molecule is zero.

Notes

The OPLSPotentialBuilder class enables the creation of a TremoloXCalculator that implements the OPLS-AA potential of Mayo, Olafson and Goddard[1]. This is a bonded valence forcefield that represents the potential energy of a molecule as a collection of simple functions based on bond lengths, bond angles, torsion angles, inversion angles and inter-atomic distances. The overall mathematical form of the OPLS-AA potential can be given as:

\[ \begin{align}\begin{aligned}E(\mathbf{x}) = \sum_i^{bonds} k_i(r_i-r_0)^2 + \sum_i^{angles} k_i(\theta_i-\theta_0)^2 + \sum_i^{torsions} \sum_{n=1}^3 k_{i,n}(1-\cos(n\phi_i-\delta_{i,n}))\\+ \sum_i^{inversion} k_i(1-\cos(n\chi_i)) + \sum_{ij}^{atoms} 4\varepsilon_{ij}\left[\left(\frac{\sigma_{ij}}{r_{ij}}\right)^{12} - \left(\frac{\sigma_{ij}}{r_{ij}}\right)^6\right] + \sum_{ij}^{atoms} \frac{q_iq_j}{4\pi \epsilon_0 r_{ij}}\end{aligned}\end{align} \]

Here \(r\), \(\theta\), \(\phi\) and \(\chi\) represent inter-atomic and bond distances, angles, torsions and inversions respectively in the configuration.

To create a OPLS-AA calculator, an instance of the OPLSPotentialBuilder must first be created. The constructor for this class has a number of options that controls the way in which the calculator built. Individual types of interactions can be excluded by setting the include_ options to False By default all possible terms are included. All valence terms can also be excluded by setting the argument use_bonds to False. This is appropriate for configurations consisting of single atoms that are not bonded to each other.

As the OPLSPotentialBuilder class uses the bonding in the configuration to determine appropriate potentials, bonds should be set on the configuration. This can be done with the findBonds() method in either the MoleculeConfiguration or BulkConfiguration class. This automatically assigns bonds between atoms that are within the covalent radii of each other. Specific bonds can also be assigned with the setBonds(), which takes a list of the indices of pairs of atoms that are to be considered bonded.

There are then a number of options that control the non-bonding interactions. The parameter lennard_jones_cutoff sets the distance beyond which Lennard-Jones interactions are truncated. To avoid integration problems in molecular dynamics, Lennard-Jones interactions are brought to zero at the cutoff using a spline function. The width of this spline can be set using the lennard_jones_smoothing_length parameter. There are no specific hydrogen hydrogen bonding terms defined within the OPLS-AA potential.

Electrostatic interactions are a little more complicated. By default the OPLSPotentialBuilder adds a CoulombSPME electrostatic solver for calculators to be used with a bulk configuration, and a CoulombDSF electrostatic solver to calculators to be used with a molecule configuration. When adding an CoulombSPME solver the cutoff used for calculating the real-space interactions can be set using the electrostatic_spme_cutoff argument. Likewise the accuracy of the smoothed particle mesh Ewald solver can be set using the electrostatic_accuracy argument. In the case where a CoulombDSF solver is added to the potential, the cutoff used can be set using the electrostatic_pairwise_cutoff argument. In cases where molecular electrostatics are to be compared with values calculated in a bulk configuration, an explicit splined Coulomb potential can be used for molecules instead of the damped shifted force potential. In this case a CoulombN2Spline electrostatic solver can be added by using use_electrostatic_direct_sum argument. As with the damped shifted force potential the cutoff is controlled using the electrostatic_pairwise_cutoff argument, and like the Lennard-Jones potential is brought to zero with a spline whose length is defined by the electrostatic_smoothing_length argument. For all Coulomb solvers the total electrostatic energies and forces can be scaled up or down using the electrostatic_scale argument.

Once the OPLSPotentialBuilder object is instantiated, a calculator is created using the createCalculator method. This takes a configuration and returns the appropriate OPLS-AA calculator for that configuration. To use electrostatic potentials in the calculator, atomic partial charges have to be assigned to each atom. The OPLS-AA defines appropriate charges for each atom type, which by default are added to the created calculator. It is also possible to add pre-calculated charges from different sources using the atomic_charges argument. This argument takes a list of charges that have the same ordering as the atoms in the configuration. The partial atomic charges can also be averaged according to a few different schemes. This is controlled by the charge_averaging parameter, which takes an NLFlag which indicates the type of averaging. The flag AtomConnectivity averages charges on each atom based on the element and the other elements it is bonded to. The flag AtomTypes averages according to the assigned OPLS-AA type. For the calculator to recognize appropriate inversion terms, improper torsion indices also need to be set on the configuration. By default this is done automatically, however existing improper torsions can be preserved by setting the set_improper_torsion_indices to False. As the OPLS-AA potential contains a number of different potentials, details of each potential used to create the calculator can be printed out using the verbose argument, which is False by default.

When creating a OPLS-AA calculator for a configuration, the configuration must have the appropriate tags for each atom type. Atom types in the OPLS-AA forcefield are grouped into classes. Atom types in a class all have the same valence potential terms and individual Lennard-Jones and electrostatic potentials. This reduces the number of parameters required to construct a transferable forcefield while still allowing specific tailoring of the potentials to different molecules. Tags used by the OPLS-AA forcefield all begin with the prefix OPLS_. The next section of tag gives the label of the class of the atomic type, and the final section gives the number of the type in its class. Appropriate OPLS tags can be assigned automatically using the method assignAtomTypes. This method uses the bonding environment of each atom to estimate appropriate types. These tags are then added to the input configuration. Type definitions have been added for atom types common to polymers, although not all atom types have a definition. In cases where the OPLSPotentialBuilder is unable to recognize the type of an atom, the tag OPLS_UNKNOWN is added to that atom.

The OPLSPotentialBuilder contains a few methods to assist in manually assigning the correct types to configurations in cases where the types cannot be automatically assigned. Atom types in the OPLS-AA potential are generally defined for specific functional groups. As a result, atomic partial charges for a given functional group usually add to zero or an integer charge for charged groups. To discover the appropriate types for a configuration, the method printTotalTypes prints a table of the available atomic types, with a short description of the kinds of atoms this type represents. As they are defined together, types for different functional groups are often grouped together. Types for just one class can also be printed using the type_class argument. The types assigned on a specific configuration can also be displayed using the printUsedTypes method. As functional groups generally have zero overall charge, correctly typed configurations should also have zero overall charge. Calculating the total assigned charge to the configuration is therefore a useful check to make sure the correct types have been added to the configuration. The total assigned charge to a configuration using the OPLS forcefield can be calculated using the totalCharge method.

[1]

William L. Jorgensen, David S. Maxwell, and Julian Tirado-Rives. Development and testing of the opls all-atom force field on conformational energetics and properties of organic liquids. J. Am. Chem. Soc., 118:11225, November 1996. URL: https://pubs.acs.org/doi/10.1021/ja9621760, doi:10.1021/ja9621760.