[gmx-users] Odd temperature spikes during T-REMD?

[Smith, Micholas D.]

Good morning,

I am experiencing a very bizarre problem during my T-REMD simulations of a (single-glycosylated residue) protein in water. During my T-REMD simulations, I find that my temperatures jumps from the expected range (my replicas are spaces between 310 to 350K) to 1000K or higher for 10 to 15 steps at a time and then fall back to the expected temperatures, and this occurs for all of the replicas.

The system contains ~25k atoms, so it isn't that large, and exchanges are occurring with -replex 500 (and I did test at -replex 1000 as well, same behavior).

I've tried using both Nose-Hoover and V-Rescale for the termostats, and I get the same behaviour. I ran additional relaxation simulations at each replicate independently without replex (where I don't see this spiking behavior) and tried again after the relaxations, and still no luck. Energy minimization of the original system looks fine, and simulating a similar sized system (almost the same protein actually) without the glycosylation didn't have this problem.

I am using gromacs2018.3 and my mdp options are the defaults you would obtain from CHARMM-GUI for a standard NPT simulation (with modified temperatures for the different replicas).

Any ideas? I am still trying to get to the bottom of why it is doing this, but was wondering if anyone else has had this happen them before.

===================
Micholas Dean Smith, PhD. MRSC