2022-04-13 Industry Benchmarking meeting ntoes

Protein-ligand benchmarks repo

• DD – OpenFE is interested in taking over maintenance of protein-ligand-benchmarks. I’d previously asked LD to take over a few topics on that.

  • DH – I’ve been less involved there, working with LD on that but we both have a lot of other work to do.

  • JW – One issue there is that everyone’s probably “scared” to make changes, they’re not sure what has to be consistent or what elements of structure need to be preserved.

• DH – LD is on vacation this week. He’s generally been busy with torsion studies, the manuscript, and other things.

• (General) – The protein-ligand benchmarks repo is really good for what it is, and it was made before we had a good understanding of data repo practices (and specifically how to aplpy them to this case). So it was an imporant first iteration, but the structure may change in the future.

• DD – I’ve started tagging issues on the repo for the next release. JChodera will re-prepare some files and resolve several structural problems using Spruce.

• DH – Worth noting that many structures don’t meet the current LiveCOMS paper requirements (low binding affinity range, sparse data points, bad crystal resolution) - I’ve opened an issue for these.

•  

Benchmarking manuscript

• DH – BSwope and XLucas wanted to add something from their analysis. And Lorenzo wanted to add an outlier analysis. We’ll know more when he comes back next Tuesday.

• LD – Been working closely with BSwope;

  • Some molecules have more conformer passing coverage report than inputs

    • LD – This is probably just a mistake on my part (later confirmed it was just a transcription issue)

  • Some GNT and MRK molecules have more conformers passing for openff-2.0.0 than B3LYP and openff-1.3.0

    • LD – This could have happened because BSwope ran openff-2.0.0 on some inputs BEFORE filtering was applied.

      • LD – I’ll work with Bill to look directly at some of the differing molecules and see if there’s something that they have in common

    • JW + LD – It could be that some jobs finished between when they ran the first MM, and when they ran the Sage optimizations.

    •  

•  

•  

• 
  
  

Lorenzo reporting

• JW – LD, since you’re reporting to three people (DH, DD, DM) now, and we have a new science lead coming online, I want to make sure that you have a single OpenFF person to report to in the future in case your priorities are ever in conflict. So I’ll talk with everyone else, but I’ll probably become your “primary manager” during the transition. That won’t mean that you need to stop what you’re doing, but rather that, if you’re ever not sure what you should be doing, you have a single person to talk to (me). I’ll talk with the other folks that you’re working with and confirm later if I should become your manager.