[AMBER] Replica Exchange with Solute Tempering implementation in AMBER14/16/18 on GPUs

From: Alan Hicks <ah14k.my.fsu.edu>
Date: Thu, 12 Jul 2018 17:38:15 +0000

Dear Amber Users and Developers,

Is it possible to implement Replica Exchange with Solute Tempering (DOI: 10.1021/jp204407d) using the current H-REMD GPU implementation in AMBER14 and later versions? Are there already plans to implement this method into the AMBER MD suite? Parmed can be used to generate the scaled topology files ( like https://arxiv.org/abs/1307.5144 or Terakawa et al, JCC 2011 ), similar to a REFEP simulation. Previous implementations of REST2 in GROMACS, NAMD, and DESMOND exchange between neighboring residues, H-REMD in AMBER implements a double-ended scheme, as I understand it based off reading the AMBER16 manual and cited papers. Would adding a neighbor exchange option in the H-REMD GPU code be possible for REST2 simulations?

I have run some preliminary tests using AMBER14 H-REMD on GTX 980 cards with a couple small test systems. When I compare the exchange frequencies between previous simulations in GROMACS using the same scaling factors, they don't match after exchange frequencies have equalized after 10 ns of simulation time. I have tested the scaling by comparing energies, which match the scaling factors, so using a neighbor exchange instead of the double ended exchange might fix this discrepancy.

Thanks and Regards,

Alan Hicks

Graduate Research Assistant

Florida State University -- Department of Physics

Tallahassee, FL 32304

AMBER mailing list
Received on Thu Jul 12 2018 - 11:00:02 PDT
Custom Search