REMD simulation in water and vacuum environment

User discussions
1

GROMACS version: 2023
GROMACS modification: Yes/No
Here post your question

Hello all, I am doing REMD simulation, I’ve executed REMD simulations for a protein in both vacuum and water environments for the time period of 2ns with 6 replicas based on the tutorial (http://dx.doi.org/10.13140/RG.2.2.22373.96482). In the vacuum condition, the average exchange probability is 0.50, while water simulations show an average exchange probability of 0.01. how to enhance the average exchange probabilities in the water environment.?

2

Hi, the relative energy fluctuations (ones that enable overlap in energy distribution between replicas, and thus exchanges) generally become smaller with the root of the number of atoms in the system. For that reason the more atoms in your system, the harder it is to obtain frequent exchanges at a fixed temperature difference.

There are two typical ways one would circumvent this:

  • add more replicas with closer temperature spacing (means more computational cost) - preferably use this calculator; or
  • switch to solute tempering so that the “effective” temperature change affects only a small subset (REST2 or even recent REST3; for the former, both gromologist and plumed have scripts that help prepare modified topologies).

Note that in REST, you only get correct sampling for the lowest temperature window, so if you want to get unbiased distributions for all temperatures, stick to option 1 (or try some more efficient and more recent approaches, such as OPES with the expanded ensemble target).