HarmonicBondForce giving energy of zero

Hi OpenFF - question, not necessarily a bug.

I was re-writing the force groups (in OpenMM) as Custom*Forces and noticed the HarmonicBondForce from an OpenFF-parameterized ligand has zero bond energies. Can that be correct? Perhaps the bonds were all at their minima - but even after a few simulation steps the energy is still zero. Below is a smallish example using Nemonapride. Thanks for your time!

from simtk.openmm.app import *
from simtk.openmm import *
from simtk import *

from rdkit import Chem
from rdkit.Chem import AllChem

from openff.toolkit.topology import Molecule
from openmmforcefields.generators import SMIRNOFFTemplateGenerator


#get a ligand (AQD - nemonapride):
smi = 'C[C@@H]1[C@@H](CCN1Cc2ccccc2)NC(=O)c3cc(c(cc3OC)NC)Cl'
aqd_mol = Chem.MolFromSmiles(smi)
aqd_molH = Chem.AddHs(aqd_mol)
AllChem.EmbedMolecule(aqd_molH)
Chem.MolToPDBFile(aqd_molH, 'AQD_ligand.pdb')

# Create an OpenFF Molecule object
molecule = Molecule.from_rdkit(aqd_molH)
molecule.assign_partial_charges('gasteiger')

gen = SMIRNOFFTemplateGenerator(molecules=molecule)

## Create an OpenMM ForceField object
forcefield = app.ForceField('amber/protein.ff14SB.xml')
# Register the template generator
forcefield.registerTemplateGenerator(gen.generator)

# load PDB file
pdb = PDBFile('AQD_ligand.pdb')


system = forcefield.createSystem(pdb.topology, nonbondedMethod=app.CutoffNonPeriodic,
         nonbondedCutoff=0.9*unit.nanometer,constraints=None)
for c, f in enumerate(system.getForces()):
    f.setForceGroup(c)
    
# make a simulation and print energies:
integrator = openmm.LangevinIntegrator(50*unit.kelvin, 1/unit.picosecond, 1*unit.femtosecond)
platform = openmm.Platform.getPlatformByName('CPU')
simulation = app.Simulation(pdb.topology, system, integrator, platform)
simulation.context.setPositions(pdb.positions)


forces = system.getForces()
for c, f in enumerate(forces):
    original_energy = simulation.context.getState(getEnergy=True,groups=c).getPotentialEnergy()
    print(original_energy, f)

output:

0.0 kJ/mol <simtk.openmm.openmm.HarmonicBondForce; proxy of <Swig Object of type 'OpenMM::HarmonicBondForce *' at 0x1611c3390> >
51.81715162906145 kJ/mol <simtk.openmm.openmm.HarmonicAngleForce; proxy of <Swig Object of type 'OpenMM::HarmonicAngleForce *' at 0x1611c3300> >
340.93114607291744 kJ/mol <simtk.openmm.openmm.PeriodicTorsionForce; proxy of <Swig Object of type 'OpenMM::PeriodicTorsionForce *' at 0x1611c3270> >
392.7482977019789 kJ/mol <simtk.openmm.openmm.NonbondedForce; proxy of <Swig Object of type 'OpenMM::NonbondedForce *' at 0x1611c3090> >
38.06694470156162 kJ/mol <simtk.openmm.openmm.CMMotionRemover; proxy of <Swig Object of type 'OpenMM::CMMotionRemover *' at 0x1611c30f0> >

run some steps and print energies again:

simulation.step(50)
for c, f in enumerate(forces):
    original_energy = simulation.context.getState(getEnergy=True,groups=c).getPotentialEnergy()
    print(original_energy, f)

output:

0.0 kJ/mol <simtk.openmm.openmm.HarmonicBondForce; proxy of <Swig Object of type 'OpenMM::HarmonicBondForce *' at 0x1611c3390> >
26.704057535960047 kJ/mol <simtk.openmm.openmm.HarmonicAngleForce; proxy of <Swig Object of type 'OpenMM::HarmonicAngleForce *' at 0x1611c3300> >
363.67554671367213 kJ/mol <simtk.openmm.openmm.PeriodicTorsionForce; proxy of <Swig Object of type 'OpenMM::PeriodicTorsionForce *' at 0x1611c3270> >
390.37960424963217 kJ/mol <simtk.openmm.openmm.NonbondedForce; proxy of <Swig Object of type 'OpenMM::NonbondedForce *' at 0x1611c3090> >
45.48346996107692 kJ/mol <simtk.openmm.openmm.CMMotionRemover; proxy of <Swig Object of type 'OpenMM::CMMotionRemover *' at 0x1611c30f0> >

[ljmartin]

Ah yes, that solves it thanks! Appreciate you taking a look. Closing

[Lnaden]

The bitmask behavior on groups is exactly what is happening. If you are ever going to be getting a single force group or set of force groups from the getState function, you should always cast your int/ints into an iterable object, unless you work in bitmask math.

Does anyone know whether CMMotionRemover should be showing nonzero energies?

If the Center of Mass is in the center of the box at step 0, its energy will be 0, but afterwords it can move a bit. The CMMotionRemover only applies a force to keep the Center of Mass in the Center of Volume loosely. Its a uniform force acting on every particle. See here for the actual operation.