[AMBER-Developers] Running TI and MBAR with GB

From: Tyler Luchko (Lists) <"Tyler>
Date: Sun, 12 Jan 2020 14:56:30 -0800

Hello everybody,

Should it be possible to run TI and MBAR with GB? We have been trying to run TI and MBAR using GB as an implicit solvent for a couple of projects but have run into a couple of problems.

For the electrostatic potential with GB, we can use TI and MBAR with pmemd, only TI with pmemd.MPI and neither TI nor MBAR with pmemd.cuda. For pmemd.MPI, the TI output looks fine but the numbers for the MBAR output don’t correspond to the actual potential energy, typically being larger and often *******. For pmemd.cuda, the entire simulation blows up after just a few steps. Is this behavior the result of a bug or an unsupported feature?

I know that soft-core VDW is not supported at this time but I was wondering what the technical issues are. How difficult is this to implement and is anybody interested?

Thank you,


Tyler Luchko
Associate Professor
Department of Physics & Astronomy
California State University, Northridge
Northridge, CA

AMBER-Developers mailing list
Received on Sun Jan 12 2020 - 15:00:02 PST
Custom Search