2022-10-18 Mitchell/Wagner Check-in meeting notes

Participants

  • @Josh Mitchell

  • @Jeffrey Wagner

Discussion topics

Item

Notes

Item

Notes

JW notes

  • Reviewed the notebooks+instructions on the train. VERY nice work!

  • Thanks for remembering to post the materials and announce the workshop on general.

  • Conda solve for the first env takes a long time - Maybe even make 3 envs (mamba, 0.10.6, and 0.11.1)

  • It was smart to use a tagged version of the protein ligand benchmark set

  •  

  • “launching the bespoke executoror”

  • You should show how to start the executor on binder

    • JM – That’s in my notes, will do

  • Including OE in the conda envs is making bespoke caching machinery try to run oe variants of some molecule processing steps, which is causing my local build to crash

    • JM – I can’t reproduce this

    • (JW reproduces on binder)

    • JM – I’ll remove OE from the environments and say “optionally, you can install OE if you have a license”.

  • The protein.pdb file isn’t readable by rdkittoolkitwrapper (but t4 has a proline? This is weird)

    • JM – I can reproduce this and have an awesome one-liner to fix it

    • JW – I’m still curious about the root cause - Optimally we could make a small change to the input PDB, and then raise an issue once we understand what’s going wrong.

  • The toolkit notebook doesn’t load the bespoke force field

    • JM – No, it does, you should read better

    • JW – I should read better





Action items

Decisions