2024-05-7 BW/LW Check-in

General updates and discussion on projects

• Torsion splitting in 2.2

  • Median looks a lot better – haven’t benchmarked on additional dataset yet

  • LW – might be relevant here

  • Possibly redo benchmarking

  • Present at FF fitting meeting tomorrow

  • Filter results by moeity with Checkmol or SMIRKS/SMARTS by mattwthompson · Pull Request #44 · openforcefield/yammbs – for filtering benchmarks by chemistries

• Fragment dataset curation code

  • BW – have database of 10M fragments

• Torsion shapes

  • BW – not had a chance to look at these yet

• slides for OMSF talk

  • LW – could I please get the graphs (RMSD/TFD/DDE) you show, or the data

  • BW – data is all

• Zach asked to moderate panel

• Dashboard link:

• idivf benchmark

  • LW – what are benchmarks called?

    • idiv-sage-2.1.0: without re-fit (probably orange line)

    • nc-tm: “nc” means “no connectivity”, so specific connectivity removed from parameters

    • r-nc-tm: re-fit version of nc

  • BW: I’ll check the plotting script commit history for mapping of graph to force field