2021-10-04 Core Developers meeting notes

Participants

  •  

Discussion topics

Item

Notes

Item

Notes

General updates

Updates

  • SB

    • Rewritten bespokefit executor - About 6000 changes – It’s way simpler now, basically a bunch of microservices, distributed over celery, much expanded API, much friendlier CLI. Overall it’s much closer to production-readiness.

      • DD – I’d love to read over this and see how everything is architected

  • CC

    • Still working with JHorton to submit peptide torsiondrive. Big blocker is that we need to constrain torsions that aren’t being driven. So I’m continuing to work with JH to get that supported

    • Working on getting amino acid library charges that LWang developed to interface with the rest of the FF. Still need to decide what the best set of SMILES string is.

    • Submitted abstract to BPS meeting. Thanks everyone who helped with revision. Should we formalize the conference abstract process?

      • JW – I don’t think we should formalize process, I generally put something together quickly for conferences, and the talk itself is the polished artifact.

      • MT – Kinda agree that we don’t need to formalize process. It could be helpful for new researchers like grad students, though it could also be kinda intimidating/bureaucratic.

  • MT

  • DD

    • Protein-Ligand Benchmarks

      • collected feedback; reviewing and addressing this week

      • meeting with folks from OpenFE on Wednesday to determine possibly-shared scope

        • JW – I’m concerned about looseness/uncertainty around input spec. We can chat later this week

    • QCArchive

      • rebalanced PRP deployments to try to advance Pavan's submission.

      • worked with Willa Wang on submission creation, user quickstart guide

      • QCEngine 0.20.0 is out

        • SB – Thanks for this, DD! It works locally for me.

    • Benchmarking

      • putting together partner retrospective survey for Season 1

  • PB

    • Done some fitting experiments with WBOs, amide sets.

    • Looked into in-ring torsions and whether we have QC data for them

    • Opened up a few issues on QCA-dataset-submission. One was about a split of the functional and dispersion calculation. QCFractal allows this, but psi4 doesn’t have a valid functional for that.

    • Volunteered for some of the OpenMM QM dataset submissions. Looked at the datasets prepared by PEastman and the submissions were having trouble. JHorton is working on some fixes for this that could get released this week

    • Worked on some more AM1 stuff. JW helped me speed it up.

      • JW – I saw that AM1 WBO calculations may do the optimization twice, but more likely it just starts a calc form the converged structure

      • SB – It may take more than one step in the second run

    • I’m having trouble retrieving collections from QCF. Says “ResultCollection can only be created from dataset object”

      • SB – This will depend on a lot of details. Probably best to post a reproducing example on GH.

      • PB – I’ve had this issue with both a basic dataset and a torsiondrive dataset. I only have this issue with local QCF instances/datasets, not the global QCA dataset.

        • DD – I could work with you on this.

        • SB – (posted some queries to run locally to help debug) Please post these on GH issue as well.

        • SB – Make sure to use QCPortal instead of QCFractal.

  • JW

    • Some toolkit/chemical perception debugging from Eastman issue (#1109)

    • TopologyMolecule deprecation/cleanups.

    • Planning some coordination with JMitchell and others. Trying to figure out how to asychronously make users guide.

      • SB – Please put me in contact with JMitchell. Bespokefit will be largely industry facing so we could use good docs there, and docs for OpenFF users isn’t really JHorton’s job. So I’ve already written docs for many packages and hould be able to provide starting material for JMitchell.

    • Some Q&A with MolSSI on QCA role(s), who is hiring, who is in charge

    • Chatted with AAM about MMIC/MMElemental. Will assist in a very limited capacity on this (~15 hours for the rest of the year)

    •  

    •  

Action items

Decisions