2022-11-07 Core Developers meeting notes

General updates

• JW – I assume Thanksgiving holiday is Thurs+Friday off for Us employees.

• JW – Daylight savings in the USA. Watch out - we tried to get the major meetings into a single time zone but some meetings may have come into conflict. Contact the meeting organizer if you’re no longer able to make a meeting.

• JW – JM is cleaning up + uploading followup workshops - First one is at https://www.youtube.com/watch?v=lZ4UUgHQpWg

• JW – DD is skipping most meetings this week to focus on F@H infrastructure

Individual updates

• DN

  • Basically almost entirely consumed by OpenFE election last week.

• MT

  • Spent a lot of time trying to hunt down regressions with OpenMM 8 beta. Found a lot of issues relating to energies computed using linux builds. Plausibly related to nonbonded kernels, but couldn’t find root cause. Made some other improvements to interchange’s nonbonded driver, issues magically went away. Let me know if anyone else has issues with OMM 8 beta.

    • JW – I’m having issues with OMM 8 beta - Toolkit showcase simulation crashes https://github.com/openforcefield/openff-toolkit/actions/runs/3372822615/jobs/5596700366#step:13:157

    • JS – What’s new in OMM 8?

    • MT – I haven’t found a discrete changelog. I’d guess it’s largely about OpenMM’s ML functionality. Not a ton of changes to the API/backend that we interact with.

    • CC – One big thing is that Modeller can now build non-triclinic boxes.

    • MT – That’s just new to modeller, right? The other parts of OpenMM have been able to handle those for a while, right?

    • CC - Right

  • Pint minor version fiasco from two weeks ago got wrapped up - Pint made an API breaking release. We updated openff-units to 0.2.0 to follow these changes. Toolkit needed small changes and made 0.11.3 changes. This was also fortuitous since the new toolkit release with a higher version will lead to cleaner env solves.

  • Worked on learning some stuff about QCArchive - Found a lot of the stumbling blocks in the user experience. And the ecosystem that we rely on (QCEl, QCNg) isn’t well-maintained.

• JS

  • Migrated pAPRika repo to Gilson Lab github (from Dave Slochower’s).

  • I released pAPRika version 1.2. Some API changes, broke Evaluator. I’ll open a PR soon.

  • I’ll also need to update TapRoom. At the moment it belong to Dave Slochower, thinking of moving it to OpenFF or GilsonLab github org.

    • JW – I recommend gilson lab.

  • I’ll also be updating evaluator, could use a reviewer

    • MT – I’d be happy to review!

• CC

  • Parameter fits for the protein null model have converged. So we have at least a first version of the protein FF.

  • Beginning to run the validation benchmarks on the null model this week.

  • Also beginning to do a fit with some protein-specific torsions.

  • PB told me that my delocalized charge torsion fits look different than his. Going to check into that further this week.

  • Setting up calcs for validating against NMR observables

• PB

  • Not much to report, was working a lot with BSwope on experimental fits last week.

    • JW – More details?

    • PB – Not new experimental data, trying different ways to do fitting to our existing data.

  • Some minor edits to Sage manuscript. Pushed for coauthor review.

• JW

  • Toolkit 0.11.3 release - Largely made to also work with new openff-units release.

  • Working on FB update, openmm 8 issue, protein related issues on OFFTK issue tracker

  • Working on partner survey, WBS dictionary for Interchange

  • I’ll be meeting with Yuanqing Wang tomorrow to make an importable EspalomaWrapper for the OFF toolkit. Partial charge method name will be XYZ-espaloma.

    • PB – This is a graph charge method?

      • JW – Yes

    • MT – Are we putting this in the toolkit repo?

      • JW – No, it’ll be in a different repo. YQ will be responsible for maintenance.

  •