implicit solvent with pbc condition?

[slweng0321]

Dear all, I am running the implicit solvent simulation to relax my configuration. I use the code below:

pdb = app.PDBFile('protein.pdb') ​ forcefield = app.ForceField('amber99sbildn.xml', 'amber99_obc.xml') ​ system = forcefield.createSystem(pdb.topology, nonbondedMethod=app.CutoffNonPeriodic, constraints=app.HBonds) integrator = mm.LangevinMiddleIntegrator(300unit.kelvin, 1.0/unit.picoseconds, 2.0unit.femtoseconds) platform = mm.Platform.getPlatformByName('CUDA') prop = dict(CudaPrecision='mixed') # Use mixed single/double precision simulation = app.Simulation(pdb.topology, system, integrator, platform, prop) ​ simulation.context.setPositions(pdb.positions) simulation.context.setVelocitiesToTemperature(300*unit.kelvin) ​ state = simulation.context.getState(getEnergy=True, getForces=True)

simulation.step(500000) positions = simulation.context.getState(getPositions=True).getPositions() app.PDBFile.writeFile(simulation.topology, positions, open('protein_relaxed.pdb', 'w'))

I have defined the box vector in the pdb file but when I checked the generated pdb with VMD, there were some overlap between periodic copies. Does this imply that the simulation didn't run with pbc condition or it ran correctly and just needed to run a longer time to be equilibrium? Thanks for any help.

[peastman]

In the call to createSystem() you specified nonbondedMethod=app.CutoffNonPeriodic. That means it does not use periodic boundary conditions, and the box vectors are ignored. This is the normal way to run implicit solvent simulations.

[slweng0321]

Thanks for replying. So if I want to run implicit solvent relaxation with pbc condition, I just turn CutoffNonPeriodic to CutoffPeriodic and leave others the same?

[peastman]

Correct. Though it's unusual to use periodic boundary conditions with implicit solvent. Not that it's necessarily wrong, if you really do want to simulate a periodic system, rather than a molecule in bulk water.

I notice that in your call to createSystem(), you don't specify a cutoff, which means you get the default value of 1 nm. For implicit solvent, you generally need a longer cutoff. 2 nm tends to be a reasonable choice.

[slweng0321]

Thanks for suggestion. I will try them. Now I think we can close this issue. Thanks again for your help.