Advice on best practices for simulating small molecules in OpenMM

[Steve Ramsey]

Hello all,

I am interested in using OpenMM to simulate protein systems with small molecule ligands present, using openff to parameterize the small molecule.

I see in the documentation protocol examples for creating a protein-ligand topology with interchange, but also directly using the openff toolkit to create parameters for the ligand separately. Given that there are a few tutorial/documentation pages that cover this subject with slightly different protocols/methods, I was hoping to get some feedback from more experienced users on what is the best practices for parameterizing a small molecule+protein system in explicit water.

For instance is my best bet to follow the protocol outlined in the toolkit showcase here which uses interchange: https://github.com/openforcefield/openf ... case.ipynb

Or this protocol which does not use interchange to parameterize the ligand and then to create the protein-ligand system using those parameters (in ambertools for example): https://github.com/openforcefield/openf ... rmed.ipynb

Do the parameters assigned to the small molecule vary depending on whether interchange is used?

I apologize if these questions have been addressed elsewhere I did try to search the forums for threads related to small molecules before posting. Thanks for your time and help!

--Steve

[Peter Eastman]

So far as I know, Interchange is a recently added class that just provides a different interface to the same parametrization code. For protein-ligand systems, the recommended approach is to use OpenMM's ForceField class for the protein, and add a SMIRNOFFTemplateGenerator for the ligand. See https://github.com/openmm/openmmforcefi ... parameters.

I suggest asking on the OpenFF Toolkit github. The developers can tell you if there are any differences in the results from the different approaches.

[Steve Ramsey]

Thanks for the quick reply! Appreciate the help!