2024-09-08 DM/MO/LW

Need for recording PL input inputs

• DM: needs

  • files, environment, details, etc, with version control

  • Should not live under MO

  • OpenFF best to look after it, but MO could contribute

• LW: don’t want MO to spend significant effort or time on reformatting, just put it somewhere

• MO: currently inputs are formatted the same way IAlibay formats his, so should be fairly standardized already

  • OpenFE already has PLB test systems as part of installed modules

  • LW: I think I’ve seen this before, sounds good

• LW: are the inputs small enough to upload to GitHub?

  • MO: looks like for 2.1, it’s 11 MB

• DM: suggests plan:

  • OpenFF creates repo, adds MO as maintainer, MO uploads data

• LW: JW and I talked about qca-dataset-submission-style repo, but automation should happen later when things are more standardised

• DM: Might be best to work in Zenodo when there’s a clear process – e.g. could see a future where for each FF release, we run benchmarks, and need to release benchmarks on Zenodo

• MO: will this just be for F@H/Alchemiscale-type benchmarks?

  • LW: that’s what we were thinking

  • MO: should I go back to septop benchmarks?

  • DM: no

• MO: IA has same files structure for 2.0 and 2.1 benchmark. I could potentially upload through my Google Drive link

  • DM: sounds like a good plan

• DM: eventually we might ask MO to run some SFEs to validate OpenFE infrastructure

  • DM: potentially compare gmx and openfe energies on some small molecules