2021-10-18 Core Developers meeting notes

General updates

• JW – I’m working 8 AM - 12 PM US Eastern time Mon-Thurs of this week, offline Fri.

• JW – MT is also working half-days. Unsure which hours.

• DD – Out starting Thursday

  • JW – Want me to take over PRP management?

  • DD – I could do it, doesn’t take much time. Largely judgement calls on which resources to request and what’s in the queue. So it’s good to follow the dataset tracking board (Error Cycling label) and spin up workers accordingly.

  • JW – If you’d like me to take over, I should still have credentials. Just send me the “playbook” for the kubectl commands you usually use.

Updates

• JW

  • VU collab meeting. Decided we’re going to make a cool proof of concept that mixes small organic molecules (openFF) with surfaces and nanoparticles (MoSDeF)

  • Roadmap review part 1. Part 2 coming next all-hands. Feel free to propose new items in the meantime.

  • Drafting new fellowship terms. Specific feedback? Good idea in the first place? Comment on this doc: https://docs.google.com/document/d/1Um4PSdx61yNGm73UfoleXzBJA0ut-Bk2Byu0-FI8BiQ/edit

  • Debugged #1106 – Tricky OpenMM forcegroup usage led to scary energy decompositions. I audited our examples to make sure we weren’t making this mistake.

  • Working on Molecule.from_pdb – Lots of progress today on generating structures with all permutations of leaving atoms gone, but output substructure file is really large. Will meet with IPulido tomorrow to figure out how to proceed.

• CC

  • Short week – Mostly working on LiveCOMS review, coming up on xtal section review. Pushing people to finish on time.

  • Dipeptide dataset – Some weird input dataset geometries. Now doing a sage minimization ahead of time.

  • I saw that DD pinged me about QCF restraints – I’ll look at that. Action needed?

    • DD – The needed changes aren’t merged yet (QCF #693). Also requires QCSubmit #172. Those will provide the prerequisites.

    • CC – Anything else needed form me?

    • DD – No.

  • This week I’ll be finalizing librarycharges that will become rosemary.

    • JW – The librarycharges from the ff14SB offxml should be good to go and high coverage. I’ll work on approving the PR that removes duplicate parameters soon (ideally before the end of the year)

• IP

  • Working with JW on Molecule.from_pdb. Earlier today I realized that we may have an issue where some bonds have unspecified bond orders.

    • JW – I think this error is good – The peptide bonds aren’t covered in any substructure so they’re listed as unspecified. So we’ll want to explicitly add the peptide bond substructure with the appropriate bond orders.

  • Other than that, looking into removing use of unnecessary RDKit molecules from our from_pdb workflow.

• DD

  • On the QCA front – Resubmitted WWang’s RESP polarizability dataset. Had to resubmit because a string wasn’t being sent through appropriately.

  • On Friday I added high-memory compute for PB’s dataset. However I noticed a typo that prevented things from running.

  • PRP hasn’t been spinning up all of our requested resources, so things have been somewhat slow recently.

  • Created PRs for dipeptide torsiondrive support in QCF and QCSubmit (see CC’s update).

  • Spoke with BPritchard on upcoming plans for QCF, will push on QCF release and production server update.

  • Worked with LD’Amore to have him take over partner announcements. We have Sage results from about half of the partners, and OPLS from many. This is timely because GTresadern is working on manuscript.

  • Drafted retrospective survey for benchmarking project, to get feedback from partners.

  • On PLBenchmarks, I have no updates.

• PB

  • Not much to update on – Presented at Mobley lab meeting last week.

  • Some fitting-related work.

  • Worked with Mobley lab grad student on evaluator. Rotation students are working on replicating solvation free energies from an old paper, comparing different FFs. DMobley wants to replicate that and show improvement using new FFs. Simulating a 25x25 grid of solute/solvent permutations. Still deciding on which compute resources will be appropriate.

• MT –

  • Great working session with JW on updating Interchange to use the topology refactor. Most needed functionality is already there, API changes weren’t too painful, and most of the work was due to hacky stuff I put into Interchange.

  • Interchange now has a prmtop writer that writes all sections and works for some (very) small molecules, but is missing some key functionality (impropers) and is not trustworthy for general use yet. Major focus of next two weeks is hammering out more test cases; focus is on energy comparisons to existing OpenMM/GROMACS exports (mostly single-molecule systems but some liquid boxes) and open to input of other key test cases.

  • Worked w/ VU on a “Foyer + Sage” system of physisorbed small molecule on an alkylsilane monolayer, later to be used for de-Parm(Ed)-ifying a VU grad student’s tribology studies. Unclear what value there is for this audience, but reach out to me if you are interested in getting a (non-biological) polymer running alongside Sage small molecules.

  • Wanting to get starting on psi4 conda-forge, no movement on libint2 issues and need input from psi4 devs on which package(s) can be ported while that’s unresolved

  • Packages on conda-forge · Issue #2300 · psi4/psi4

• (JW and IP working session on Molecule.from_pdb)

  •