2024-02-16 Mitchell/Wagner Check-in meeting notes

Participants

  • @Josh Mitchell

  • @Jeffrey Wagner

Discussion topics

Item

Notes

Item

Notes

General updates

  • JW –

    • I’ll send workshop invites soon

    • Please share workshop materials somewhere for bus factor, even if they’re incomplete.

      • JM – Will do, I’ll invite you and MT

    • FYI: We’ll be doing a joint-ish demo at the annual meeting with OpenFold and OpenFE (90 minutes total). This will target running on colab and will NOT require using output from one ecosystem as input to the other. We’ll just all use the same protein-ligand system, and the OpenFF part can just aim to do vanilla MD. I’m going to task you with this in late March and I’ll take point on coordination before then, so just FYI.

  • JM – I did clockify after the last meeting.

    • JW – Excellent. Thanks!

  •  

  •  

Workshop prep

https://openforcefield.atlassian.net/wiki/spaces/COMMS/pages/2719055920

  • JW –

    • Stay positive - Don’t diss other FFs

    • In the introduction, consider mentioning that you’re showing a very roundabout way to do parameter assignment and showing deep inspection options in this tutorial, and that folks that just

    • Nomenclature:

      • Parameter sections (Electronstatics, Bonds)

      • Parameters (Bond) have SMIRKS and physical values

    • SMIRKS explanaiton picture: https://docs.google.com/presentation/d/1q9_fhglzvLK6Rtr5v0x_Vnq1GztPvVT548Db8xZANk4/edit#slide=id.g2312c8758a7_0_303

    • May be good to practice talk with the notebook at the size you’ll have it for the workshop

    • If you need to take more time, when talking about parameter assignment, consider showing the interchange matrix formulation for parameter assignment, and say something about how ML is cool.

      • JM – This would be kinda off-topic, this isn’t something you could do interactively/as a part of parameter fitting. Also kinda want to

      • JW – Might be relevant for folks who want to do fitting in ML frameworks?

      • JM – I’ll think about it.

    • AMBER energy mismatch - It’s just a thing with how we’re calling AMBER’s vdW switching to calculate that energy. We can switch settings to get it more spot-on, but it’s kinda complicated, and there’s no guarantee that users will use the correct vdW switching cutoff.

      • JM – Similar stuff with Gromacs, I think?

      • JW – I think so, so maybe good to mention that we guarantee correctness in OPenFF, but taht in other engines folks may need to watch out for stuff like nonbonded settings.

    • Make the nglview window big when you look at it, be conscious that right now there’s a 5ish second lag between audio and video, and I’m getting about 0.5fps, so focus on talking about static views.

    • Have bespokefit outputs in the workshop materials.

    • Do simulation and load into MDAnalysis to highlight interop?

    • Point people to examples, show how to unpack with examples helper and/or point to docs.openforcefields online examples.

    • Install instructions?

      • Both local and colab

      • Env will have last OFFTK and Interchange that’s compatible with current bespokefit (OFFTK 0.14.5, Interchange 0.3.18, Bespokefit 0.2.3)

    • JW – Feel free to ping MT and I to test install instructions

    • …

    • JW – I’ll try to get a bespokefit build with a new version of OFFTK out ASAP. But please prepare install instructions as if this won’t work, in case I forgot something.

    •  

    •  

Action items

Decisions

Â