2022-02-16 MT/HJ Check-in meeting notes

HJ – As I recall from offline comms, I can’t roundtrip GROMACS / combine a GROMACS file with an OpenFF ligand, right? Will that come with the biopolymer toolkit release?

• MT – GROMACS importing won’t be ready until May at the earliest. Priority list is biopolymer stuff, then parameterhandler plugins and GBSA. Exporting is WAY easier than importing.

HJ – So, I want to have a plan for when I can combine a protein with an OFF ligand and simulate it. I want to make sure that I start testing/adopting interchange at a time when my testing can extend into our implementation.

• MT – To lay out some options - It’s currently possible to:

  • Load a protein PDB into the development version of the OpenFF toolkit, parameterize it using the ff14SB port, and then combine it with an OpenFF ligand.

• HJ – The thing I need is to replace a workflow that takes

  • Protein: An OE Design Unit (OEMol)

  • Ligand: OEMol

  • ForceField: Uses openmm.ForceField class to parameterize

  • Combiner: Uses parmed to combine systems

  • Engine: Gromacs

• JW – What we can do today is:

  • Protein: OEMol

  • Ligand: OEMol

  • ForceField: SMIRNOFF FFs and OpenFF’s port of AMBER FF14SB

  • Combiner: Interchange

  • Engine: OpenMM, GROMACS, others

• JW – What we can do in the future:

  • ForceField:

    • I don’t know of plans to pull other protein FFs into SMIRNOFF format

    • OpenFF is going to release Rosemary, probbaly late 2022 (but not guaranteed)

    • If we wanted to spend some time validating it, we could support loading proteins from openmm

MT – Can make interchange using OFF toolkit, have an unvalidated implementation of from_openmm (but information contentwise there will be certain aspects of this that are fully impossible)

JW – It’d be helpful to clarify if the Interchange.from_gromacs request is really a requirement. It’s possible that if you’re just using OpenMM to assign the parameters, then we may only need to support Interchange.from_openmm_system.

• HJ – I’ll keep working on defining requirements and looking into which importers are needed.

• MT – And to clarify - Loading .gro files is easy and we can do it right now. But if you need to load .top or .itp files, that’s much harder and it will be a while before we can do it

JW – So, maybe we want to define whether this is a good time for you to start using Interchange, or if you’ll need additional functionality before you can start testing it.

• MT – In Interchange.from_openmm_system, I'd be concerned about Topology.from_openmm,

HJ – I tested MT’s script, it worked perfectly. Do we have a function that returns the total energy of the system? The current functions all give me decomposed energies, is there a method that returns the totals?

• MT – There isn’t a method that returns the total energy, you’ll need to sum it up yourself.

Next week, JW and MT will give updates on the status/timeline for

• loading proteins from OEMols