/
2022-02-02 Mitchell/Wagner check in

2022-02-02 Mitchell/Wagner check in

Participants

  • @Josh Mitchell

  • @Jeffrey Wagner

Discussion topics

Item

Notes

Item

Notes

General updates/discussion

  • JW – Bespokefit status (I forgot)

    • JM – Still kinda under development, it should probably have the docs PR finished before it goes into production. Not sure about status. It’s amazing when it works, but gives mysterious errors kinda frequently.

  • JW – Working on improving team transparency - Could I put you on Trello?

    • JM – Sure

    • (JW added JM to trello)

  • JM – How do you find molecules of interest in QCArchive/using QCSubmit?

    • JW – You kinda don’t. Only our datasets on QCA have SMILES, and QCSubmit has some limited search functionality for these SMIELS if you know the dataset you’re looking in. We’d initiated some efforts to systematically index QCArchive based on chemistry, and let people search/slice/visualzie contents, but they’ve all kinda fallen flat, eg https://openforcefield.atlassian.net/wiki/pages/createpage.action?spaceKey=IN&title=Benchmark%20dashboard

    • JM – I’d be interested in improving things on this front, especially if it proceeds all the way to a benchmarking frontend.

    • JW – As you work on the QCSubmit user guide/docs, think about how the stages of this sort of frontend could be completed in a way that adds value/functionality at each step, even if the whole process doesn’t complete.

  • JM – What’s the difference between QCFractal and QCArchive? Who controls them?

    • JW – QCFractal is the codebase to run a QCFratal server. “QCArchive” is the largest instance of a QCFractal server. Both of them are controlled by MolSSI (a large grant-funded computational molecular sciences institute at Virginia Tech).

  • JM – Why do we have QCSubmit?

    • JW – Early on, we’d submit molecule datasets to QCArchive, but then when we’d try to pull them out, we’d have no idea what their graphs were, so we’d be unable to use them in force field fitting. We also got into a mess of bookkeeping, where datasets were named/tagged/labeled inconsistently. So QCSubmit and QCA-dataset-submission are repos that try to reduce entropy on those fronts.

Todos

  1. (high) Bespokefit users guide (in progress)

  2. (high) QCSubmit users guide (in progress)

  3. (high) ELF10 COOH conformer rotation issue (in progress)

  4. (medium) Toolkit API desire paths

  5. (medium) Toolkit docs cleanup

  6. (medium) Toolkit revised user guide

  7. (medium) Review+approve+merge OFFTK #1176

  8. (medium) Come up with “milestones” for making first of three videos this year (like, “first video on covid spike protein and small molecule, filmed this day, edited in that range, etc…”)

  9. (low) “How to cite” section on website (do not start, pending feedback)

  10. (low) Check main website for broken links

  11. (low) migrate docs to be under http://openforcefield.org

  12. domain (openforcefield/status would be base project repo)

  13. (low) Conda env yamls for each release (automated inside of Toolkit’s single-file-installer action)



Action items

Decisions