2020-04-01 Bespoke Fitting Meeting notes

General check in

• Huafeng – No protomer enumeration

• DLM – We should ask the community for tools to do protomer state enumeration

• JDC – Could do using OETK

• (General) – We’ll do protomer enumeration in OETK for now, but hope/advertise for a OSS solution later

• JH – For fragmenter migration, is AmberTools WBO ready for prime-time? Do we want to complete migration into OFFTK?

• JDC – We have many different options

• JW – AmberTools WBO should be around the same as

• JH – What specific functionality should I aim for when pulling data from QCA

• JDC – Look at Yudong’s work

• JH – Look at getting trajectory back as final results, etc

• JH – Has anyone tried fitting to torsion gradients, instead of just energies? Tachenko’s work looked at this.

• JDC – We should be able to fit to both simultaneously. You lose longer-range info about relative energies of conformer basins if you only use gradients.

• JA – I recall seeing some work that you can’t fit to both energies and gradients at the same time

• JDC – This may be due to the fact that ABSOLUTE energies in MM mean nothing, but in fitting, you need to control for all the factors that contribute to the absolute energy of a system. We may be able to get a lot more accurate using class 2 force fields, but for now

Other topics

• JW – Jamshed – Confluence onboarding

  • We’ll do this right after call

• JW – Thompson is being assigned to support benchmarking efforts, JH should coordinate with him

• DLM – HX, have you used optimization trajectories during fitting?

• HX – We use transition/saddle points, not just minima

• DLM – To clarify, I mean, geometries +energies from iterations

• HX – We’re not using this, and this ins’t something I’d thought about. Can you share data?

• DLM – Yes, we’ll keep you in the loop about how this turns out.

• HX – I’ll ask my team about what they think of this. I’m concerned that our DFT theory isn’t high-level enough to capture fine energy differences, especially when you’re far from the optimal geometry. JDC has heard about our work doing energy min using low-level DFT, and then optimize using a high-level method.

• DLM – Speaking of fitting to minima/quantum theory level, OPLS3 does a really good job of identifying lowest-energy conf, but not so good on the actual

• JA – Looking at absolute free energy of ligand, and solubility calculations. What’s best way to talk about this?

• (General) – Will schedule another meeting for this