perses.annihilation.HybridTopologyFactory

class perses.annihilation.HybridTopologyFactory(topology_proposal, current_positions, new_positions, use_dispersion_correction=False, functions=None, softcore_method='amber', softcore_alpha=None, softcore_beta=None, bond_softening_constant=1.0, angle_softening_constant=1.0, soften_only_new=False)

This class generates a hybrid topology based on a perses topology proposal. This class treats atoms in the resulting hybrid system as being from one of four classes:

unique_old_atom : these atoms are not mapped and only present in the old system. Their interactions will be on for

lambda=0, off for lambda=1

unique_new_atom : these atoms are not mapped and only present in the new system. Their interactions will be off

for lambda=0, on for lambda=1

core_atom : these atoms are mapped, and are part of a residue that is changing. Their interactions will be those

corresponding to the old system at lambda=0, and those corresponding to the new system at lambda=1

environment_atom : these atoms are mapped, and are not part of a changing residue. Their interactions are always

on and are alchemically unmodified.

Attributes:

hybrid_positions

The positions of the hybrid system.

hybrid_system

The hybrid system.

hybrid_topology

An MDTraj hybrid topology for the purpose of writing out trajectories.

new_to_hybrid_atom_map

Give a dictionary that maps new system atoms to the hybrid system.

old_to_hybrid_atom_map

Give a dictionary that maps old system atoms to the hybrid system.

omm_hybrid_topology

An OpenMM format of the hybrid topology.

Methods

handle_harmonic_angles()	This method adds the appropriate interaction for all angles in the hybrid system.
handle_harmonic_bonds()	This method adds the appropriate interaction for all bonds in the hybrid system.
handle_periodic_torsion_force()	Handle the torsions in the hybrid system in the same way as the angles and bonds.
new_positions(hybrid_positions)	Get the positions corresponding to the new system.
old_positions(hybrid_positions)	Get the positions corresponding to the old system

handle_nonbonded

__init__(topology_proposal, current_positions, new_positions, use_dispersion_correction=False, functions=None, softcore_method='amber', softcore_alpha=None, softcore_beta=None, bond_softening_constant=1.0, angle_softening_constant=1.0, soften_only_new=False)

Initialize the Hybrid topology factory.

Parameters:

topology_proposal : perses.rjmc.topology_proposal.TopologyProposal object

TopologyProposal object rendered by the ProposalEngine

current_positions : [n,3] np.ndarray of float

The positions of the “old system”

new_positions : [m,3] np.ndarray of float

The positions of the “new system”

use_dispersion_correction : bool, default False

Whether to use the long range correction in the custom sterics force. This is very expensive for NCMC.

functions : dict, default None

Alchemical functions that determine how each force is scaled with lambda. The keys must be strings with names beginning with lambda_ and ending with each of bonds, angles, torsions, sterics, electrostatics. If functions is none, then the integrator will need to set each of these and parameter derivatives will be unavailable. If functions is not None, all lambdas must be specified.

softcore_method : str, default ‘default’

The softcore method to use. The options are: default: as an atom is being disappeared, increase softcore strength. For core atoms, don’t use softcore at endpoints, but interpolate to full softcore at lambda=0.5 amber: same as default, but core is excluded from softcore classic: original scheme used by this code. All alchemical atoms interpolate to 0.25 * softcore at lambda=0.5, but don’t use softcore at endpoints.

softcore_alpha: float, default None

“alpha” parameter of softcore sterics. If None is provided, value will be set to 0.5

softcore_beta: unit, default None

“beta” parameter of softcore electrostatics. If None is provided, value will be set to 12*unit.angstrom**2

bond_softening_constant : float

For bonds between unique atoms and unique-core atoms, soften the force constant at the “dummy” endpoint by this factor. If 1.0, do not soften

angle_softening_constant : float

For bonds between unique atoms and unique-core atoms, soften the force constant at the “dummy” endpoint by this factor. If 1.0, do not soften

Methods

__init__(topology_proposal, …[, …])	Initialize the Hybrid topology factory.
handle_harmonic_angles()	This method adds the appropriate interaction for all angles in the hybrid system.
handle_harmonic_bonds()	This method adds the appropriate interaction for all bonds in the hybrid system.
handle_nonbonded()
handle_periodic_torsion_force()	Handle the torsions in the hybrid system in the same way as the angles and bonds.
new_positions(hybrid_positions)	Get the positions corresponding to the new system.
old_positions(hybrid_positions)	Get the positions corresponding to the old system

Attributes

hybrid_positions	The positions of the hybrid system.
hybrid_system	The hybrid system.
hybrid_topology	An MDTraj hybrid topology for the purpose of writing out trajectories.
new_to_hybrid_atom_map	Give a dictionary that maps new system atoms to the hybrid system.
old_to_hybrid_atom_map	Give a dictionary that maps old system atoms to the hybrid system.
omm_hybrid_topology	An OpenMM format of the hybrid topology.

handle_harmonic_angles()

This method adds the appropriate interaction for all angles in the hybrid system. The scheme used, as with bonds, is:

1. If the three atoms are all in the core, then we add to the CustomAngleForce and interpolate between the two

   parameters

2. Otherwise, we add the angle to a regular angle force.

handle_harmonic_bonds()

This method adds the appropriate interaction for all bonds in the hybrid system. The scheme used is:

1. If the two atoms are both in the core, then we add to the CustomBondForce and interpolate between the two

   parameters

2. Otherwise, we add the bond to a regular bond force.

handle_nonbonded()

handle_periodic_torsion_force()

Handle the torsions in the hybrid system in the same way as the angles and bonds.

hybrid_positions

The positions of the hybrid system. Dimensionality is (n_environment + n_core + n_old_unique + n_new_unique) The positions are assigned by first copying all the mapped positions from the old system in, then copying the mapped positions from the new system.

Returns:

hybrid_positions : [n, 3] Quantity nanometers

hybrid_system

The hybrid system.

Returns:

hybrid_system : openmm.System

The system representing a hybrid between old and new topologies

hybrid_topology

An MDTraj hybrid topology for the purpose of writing out trajectories. Note that we do not expect this to be able to be parameterized by the openmm forcefield class.

Returns:

hybrid_topology : mdtraj.Topology

new_positions(hybrid_positions)

Get the positions corresponding to the new system.

Parameters:

hybrid_positions : [n, 3] np.ndarray with unit

The positions of the hybrid system

Returns:

new_positions : [m, 3] np.ndarray with unit

The positions of the new system

new_to_hybrid_atom_map

Give a dictionary that maps new system atoms to the hybrid system.

Returns:

new_to_hybrid_atom_map : dict of {int, int}

The mapping of atoms from the new system to the hybrid

old_positions(hybrid_positions)

Get the positions corresponding to the old system

Parameters:

hybrid_positions : [n, 3] np.ndarray with unit

The positions of the hybrid system

Returns:

old_positions : [m, 3] np.ndarray with unit

The positions of the old system

old_to_hybrid_atom_map

Give a dictionary that maps old system atoms to the hybrid system.

Returns:

old_to_hybrid_atom_map : dict of {int, int}

The mapping of atoms from the old system to the hybrid

omm_hybrid_topology

An OpenMM format of the hybrid topology. Also cannot be used to parameterize system, only to write out trajectories.

Returns:

hybrid_topology : simtk.openmm.app.Topology