2021-01-08 QCA Submission Meeting notes

Owned by David Dotson, created
Last updated: Apr 20, 2021
2 min read

Date

Jan 8, 2021

Participants

  • @David Dotson

  • @Pavan Behara

  • @Trevor Gokey

  • Ben Pritchard

Goals

  • New advancements

  • New submissions

  • Upcoming infrastructure improvements

    • STANDARDSv3 submission machinery in QCSubmit

    • STANDARDSv3 submission machinery in qca-dataset-submission

    • Dataset index automation on qca-dataset submission

  • Upcoming science support

    • Enforced c1 symmetry in psi4 is ready

    • Hessians support

  • Larger advances

    • Automated FF coverage gap identification, torsion prioritization, submission generation

    • Benchmarking (dashboard, etc.)

Discussion topics

Item

Presenter

Notes

Item

Presenter

Notes

Other users are starting up with compute

Ben

  • BP: if there are problematic interactions with use of the server, let us know

    • DD: managers without compute tags will grab tasks from all compute tags; important that we ensure all managers have at least the openff tag

  • BP: Doa’s working on a refactor of the REST interface; this is including fine-grained permissions, including on compute tags

Protomers/tautomers

Pavan

  • PB: need to resubmit protomers/tautomers dataset with molecule indexing fix

  • DD+TG: should create a v1.1 with a fix to the notebook, regenerated dataset.json; make a new PR with this change

Hessians

Trevor

  • TG: planning to add hessians to lots of datasets to support my work; think it will be easy, but not sure yet

Genentech sets

Pavan

  • PB: waiting for Josh to cut a new release of QCSubmit, then preparing additional Genentech sets

STANDARDSv3

David

  • TG: Reviews from Parsley paper could be addressed at least partially by STANDARDSv3 for future release manuscripts; will have to think more about how the standards can help us in future reviews

  • DD: we’ll pick up STANDARDSv3 implementation discussion next week

Dataset index automation

David

  • DD: will take on automating this; may have to wait until after benchmarking work with partners largely complete (Feb)

Automated FF coverage

Trevor

  • TG: been working on typing; working on procedural SMARTS generation

Benchmarking

David

  • TG: Re: Swope’s question - managers fail to deliver an optimization, but local runs succeed

  • BP: Can pull down the input schema from the DB, run with QCEngine (TG’s approach is similar, but skips the QCEngine layer and injects straight into geomeTRIC).

    • DD: will take a snippet from Ben as a start (posted in Slack)

Action items

@David Dotson will send out a reminder to all managers during spin up that compute tags must at least include openff
@Pavan Behara will create a v1.1 of the the protomers/tautomers dataset with indexes generated from preparation notebook fix
@David Dotson will take on automating dataset INDEX.md status page in Feb
@David Dotson will use Ben and Trevor’s suggestion for executing optimizations with QCEngine from a server task

Decisions