QM fits with Amber RESP charges | @Chapin Cavender | CC will link slides here JW: comparing the fits on slide 9, is it correct that QAmber charges should only hit proteins vs small molecules, except for a few cases. Shouldn’t the small mol TD fits mostly be the same as the Specific FFs? PB: is it inconsistent to mix ELF10 and AMBER protein charges? MS: will you investigate AMBER L-J? CC: yes, next step. I am trying just charges now. Looking at the values, L-J terms don’t seem to change much between FFs, to the 100th decimal place. Charges seem to differ much more
DM – Do you think this difference is “big”? CC – Unknown, I don’t have a good sense for what to expect without running the benchmarks. MG – Yeah, we’ve seen lots of times when things looked good and failed on folded protein benchmarks. MS – We have plenty of compute available here, if we want to leapfrog benchmarks
LS – Are the torsion profiles becoming more similar to AMBER protein ff torsion profiles? CC – Is the second release candidate for nagl in a state where i can use it for benchmarks? And would this be scientifically useful? LW – Q1 I think so, planning to upload second release candidate soon. Q2 I’m not sure what to expect since the nagl and librarycharges were so similar CC – Even thought charge RMSE is similar, I think the small difference in NAGL charges might lead to a significantly different charge distribution. I’ll file this as “interesting question/low priority”
MS – One question that came up was the plan for coming up with alternate conformations from the PDB, and how to turn that into a molecule fragment suitable for QC calculation. CC – Right, if we excise a small folded peptide from the PDB and optimize in QM, will it just move to a different minimum. I think AF was looking into it. AF – Those aren’t finished, so I still don’t have much info on whether they’re staying in the same basin.
|
