2023-10-11 Mitchell/Wagner Check-in meeting notes

General updates

• Welcome back!

• JW – I’ll be out basically all of November, want to make sure you’re on a good footing before that happens. I’ve assigned you some tickets that are more documentation/instruction-adjacent.

• JW – Any suggestions for changes to how we coordinate/plan work?

  • JM – OpenFF is pretty well organized. Would be good to start planning API breaks ahead of time and committing to them, like starting a milestone for a 1.0 release and keeping track of breaks going into it.

  • JW – I’m becoming more OK with rolling deprecations - Maybe we could have all of them land every 18 months or so. A little nervous right now about committing to anything 1.0.

  • JW + JM – Making new functionality private by default except for well-considered user-facing functionality is a good idea.

• JW – JHorton is owner for bespokefit for the next 9 months. We’re responsible for deployment/keeping up with upstream breaks but he’s driver+gatekeeper for major changes and new features. User support will probably be shared.

• JW – I’d like OpenFF to run a few followup workshops in the coming months - similar to the PTM workshop but on other topics of interest. Thoughts for topics?

  • JM – ZB and I just chatted about how to get folks excited about OMSF project. We had the idea of teaching people MD with OMSF tools - Like setting up systems in OpenFF and running them in OpenMM. So it could be cool to do a “have you never done MD before?” type intro.

  • JM – How’s Rosemary looking? Any chance it’s a vsite FF and thyme is protiens?

  • JW – 3+ months, possibly as much as 12. Yes, it’s possible vsites will come first.

  • JM – Nagl could be a topic. Maybe PTMs with Nagl.

    • JW – Good call.Also, we have a substructure spec for Topology.from_pdb - example here. I’d like ot have a tool that makes it possible for users to make these for their inputs

    • JM – Could make a tool that creates substructures from SMILES.

    • JW – That could work, I’ve got some early funcitonality that creates substructures from OFFMols where all atoms are labeled with metadata['substructure_atom']=True or `False`.

    • JM – Could do both. How do we load PDB now?

    • JW – OpenMM reads the PDB, populates CONECTS by residue template+explicit CONECTs. Then we add bond orders and formal charges using substurctures. But we want to eventually be compatible with matierlas with no concept of residues. So maybe it’s OK to guess connecticity from distance sometimes.

    • JM – I’m super hesitant about guessing from distance. But maybe in the materials world that’s inevitable. But we should have tools ready to handle inputs with residues/explicit CONECT and non-explicit connectivity cases.

    • JW – OpenFE is working on pdbinf which is a custom PDB-loader-that-can-accept-custom-residues.

    • JW – This is useful - Let’s come back to this.

Trello

To dos

1. (high) LOI for CZI

2. (high) Propose 3 workshops for Dec-Feb, including topic and format.

3. (medium) Add NaglToolkitWrapper to OpenFF Toolkit docs

4. (medium) Topology loading cookbook PR

5. (medium) Add autogenerated API docs (with no/minimal use of __all__, and it’s fine if things like ForceField are super deeply nested)

6. (medium) RNA (openmmforcefields) + small mol/lipid (Sage) example (nucleic acid formulations are a hot topic, could get us funding/users)

7. (medium) Take a swing at getting bond orders into Topology.visualize (and check whether it’s in Molecule.visualize)