2022-10-17 Core Developers meeting notes

General updates

• DN will be offline for the next few hour

• In-person meeting

  • We are working on a date, but it is probable that we’ll hold it in 2023.

Updates

DN

• OpenFF governing board meeting

  • OpenFF governing board meeting notes editing

  • OpenFF governing board meeting follow up email

    • DN - we talked about the decision making process about new items and features. The GB is okay with us making the day to day decisions, but we should always bring major things to them. Features prioritization is our prerogative.

      • DD - Should we come up with a “code of conduct”?

      • DN - I can talk about that with Karmen.

• OpenFF Confluence sanitization

• OpenFF meeting invites sanitization

• OpenFF Zenodo update (failed, but took me a lot of time)

• OpenFF lead team meeting

• OpenFF website updates (industry partners and how to join tab)

• OpenFF ff-release meeting (meeting, notes and follow up)

MT

• Partially offline last week

• Interchange 0.2.2

  • Small improvements to Amber export

    • A bug with zero-residue system fixed by somebody from Treeline Biosciences

    • Atom names now passed through from information provided by toolkit, or atom names if Atom.name is empty

  • Includes openmmforcefields and from_openmm examples

• Possible bug with circular imports in toolkit

  • Does not appear to happen with conda builds

  • Difficult to reproduce with pip-based installs

  • The fix is in, but hesitant to make a new release if this is not an issue to users

• DD

  • Protein-ligand benchmarks

    • working on compute system, test suite for fah-alchemy

    • CI is up on fah-alchemy, including standing up Docker instance of neo4j

    • worked with Ivan on perses protocol; got protocol working, now he's putting it into a nice form for my usage in fah-alchemy

    • worked with Mike on settings feedback in gufe

    • merged protein-ligand-benchmark#52 with Irfan; working on fixes to Python lib to accommodate in protein-ligand-benchmark#78

  • QCArchive

    • new prod environments PR up on qca-dataset-submission#316; waiting for review from Pavan before merge/deploy

      • Updating deployment environments, including geomeTRIC 1.0

• CC

  • Some issues with protein fits

    • Persistent issues with carboxylate parameters, same as what was brought up a few weeks ago. Considering punting on this, not fitting new parameters and just removing the failing optimization datasets from this training - this seems to be a different issue, not needed in order to get an initial protein parameter set out.

    • Generating validation plots (ddE, RMSD, etc. histograms) for proposed carboxylate split