2022-07-12 Meeting notes

Updates from MolSSI and infrastructure advances

• BP – Not much from me. I’ve been working on some ancillary things like conda envs and docker images. Doing some cleanup of the next release. Changed the way that torsiondrives and grid optimizations work in the next release.

• BP – I have a rough plan for contacting people about using the next branch.

  • JW – MThompson and I are running openff toolkit+interchange RCs right now on conda-forge. Let me know if you’d like our release checklists and other materials for your beta/rc period.

• BP – For the next month-ish I’m totally booked with other things.

• BP – Ran regular monthly backups over the weekend

• DD – I’m working on the data storage for protein-ligand-benchmark project. Wondering how to do good hashing of structures/objects, but I’m wondering if you have advice about which pitfalls to avoid. Like, in the next branch, does the cleint do the hashing, or is it done server-side?

  • BP – The hashing is done server-side, since you should never trust the client. Also in the next release, ONLY the molecule will be hashed.

  • DD – Why only molecules?

  • BP – Hashing other things is fragile - The fields that you hash might change, so you need to figure out which items get serialized to dict before hashing. And that could change with version.

  • DD – For the F@H project, the thing that will be hashed is an alchemical network.

Infrastructure advances

• JW – OE license lapsed, I’ll have this fixed by noon pacific time today.

• DD – For PRP, we can’t submit new pods until we update our publication lists.

  • JW – I’ll update this. May need some help with my credentials in case they’ve lapsed, will let you know.

Throughput status

New dataset: RNA dataset from Chodera lab. LGTM!

• Their submission is a PR from an external fork and fails validation due to bad credentials, it works if I made a PR from it, maybe we need to add them to our repo or we can work off the new PR.

• DD – Hmm, validation should be possible to run from a fork (that is, it doesn’t need to talk to QCArchive, so it doesn’t need a QCA token)

• PB – How should we proceed? I’ll need to mirror any changes that the submitter makes to their fork in my branch

  • DD – I can make it a low-priority to-do to not require the QCArchive token for submission validation. But until then PB should continue mirroring the changes to the fork on his branch.

  • (General) – Do we actually need wavefunctions?

    • PB – I mentioned that he shouldn’t request wavefunctons unless he has a defined need

    • BP – It seems like it’ll be about 50-100GB, we could hold that if needed, but let’s make sure it’s actually needed.

• PB – Version 2 of benchmark ligands - JH, do you still want this to run?

  • JH – Let’s end-of-life this

• OpenFF Protein Capped 3-mer Backbones v1.0 - 0/54 TDs complete. 196000 from 160065 opts in last week.

•