2021-06-03 SB/DM/JW Workshop Sync up

2021-06-03 SB/DM/JW Workshop Sync up

Participants

  • @Simon Boothroyd

  • @David Mobley

  • @Jeffrey Wagner

Discussion topics

Item

Notes

Item

Notes

Presentation plan

  • JW’s slides (in progres) https://docs.google.com/presentation/d/1T-IgneIQ6LJqYl4mKqglHaFnt8d0KD1rb2zx4VYYlt8/edit#slide=id.gdd8c60fd7c_0_250

  • Schedule

    • Mobley – 5 min

      • What today’s going to be about

      • Where we came from (

        • first year = automation and FF release,

        • second year = dataset design and automation and another FF release,

        • third year = trying lots of new science, reaping benefits of automation in FF optimization+benchmarking)

      • Preview for rest of talk

        • Call out one or two of the most impressive things that Simon is presenting, and that we can now do routinely

        • Here are things we’ve tried and what we’ve learned – Preview of one or two things from subsequent parts.

        • Industry benchmarking call-out

          • Which plot to show? From one company or from public set? Would eb godo to answer whether all the plots look the same

          • JW will make plots for all the datasets that we’ve received from pharma and report whether they’re all qualitatively the same, and if so provide guidance on which to use.

          • If he has time, SB will expand the public set to have the GAFF results and upload those plots.

    • Wagner – 15 min

      • “Here is what we see our big objectives as, and how infrastructure has contributed to each of them”

      • Maybe rearrange talk to better transition from mobley – If he ends with industry benchmarking, maybe begin with the “COMAPRE” slide series

    • Boothroyd – 40 min

      • Will start work over weekend

      • Historical perspective on what we’ve tried, what did/didn’t work

      • WBO review, organizational review / science/infrastructure progress

      • Sage progress, what we’re trying

      • Responsiveness to issues from pharma partners – eg amide issue, sulfonamide

      • What’s coming: sage, WBOs, protein-ligand FF, self-consistent fitting of FFs.

    • JW – For science part, do we want to focus more on the “amazing amount of things that Simon can do with automation”, even if it’s at the expense of more practical information?

      • DM – Kinda agree. People who want lots of details will be able to join the followup sessions.

      • SB – Agree. People who want more dry details are free to join FF release meetings.

Followup session planning

JW and KCJ started preparing a poll for followups here: https://openforcefield.atlassian.net/wiki/spaces/MEET/pages/1828192257

  • SB + JW pruned the presentation topics list to make the bespokefit session more of a presentation, since the software isn’t ready for release. Also, removed the system session, since it’s also pre-alpha and hte ML interface isn’t clean.

Action items

Decisions