Updates | | JW: haven’t done much on the benchmarking front unlikely to be able to test Schrodinger would need to request access via Gilson lab; will probably be a “no” one discussion point: when do we want to jump to openff-toolkit usage in the deployed envs
DH: gave demo on 3/26; same as partner call, but with more detail Xavier, Bill, and Kaushik present not sure if they’ve given it a try on their own infrastructure Lorenzo starting to work on analysis expansions
LD: been talking with Xavier, Bill in particular, Xavier’s suggestion on finding global QM minimum, take the index of that conformer, check MM energy of that conformer, see if that is the MM global minimum (or not). If MM and QM global minimum conformers not the same, give the RMSD between the conformers. DH: Xavi is more interested in the optimized geometry MM evaluation, not the initial geometry optimization; this is basically the match-minima analysis DH: could add a column that gives you which conformer was the reference DD: let’s collect the suggestions from Bill, Alberto, and Xavier for this analysis; we’ll identify what they ultimately want (likely the same thing), and then build an approach (if it doesn’t already exist as e.g. match-minima) JW: I wonder if there’s some way we can provide them with composable components I see openff-benchmark as three stages in near term future stage 1: finish out season 1, highest priority stage 2: refactor into a more coherent set of workflow components stage 3: a season 2, new results, etc.
LD: once we finish season 1, all data is there, can do all analysis you want
JW: for season 1, need 2 PRs from JH, in review Schrodinger pathway, need a reviewer with access to Schrodinger DD: could engage Bill directly; I can facilitate this, worked with him to do live testing many times
JH: 2 PRs, SMARTS filter and fix to coverage
|
