Error in starting a new simulation starting from configuration obtained with cluster

GROMACS version: 2020.4
GROMACS modification: No

Dear all,

I have successfully used the cluster tool to obtain different conformations of a simple solute+solvent system, previously obtained with the REMD methodology.

The problem occurs when running a simulation starting from a structure obtained with the cluster. With the cluster tool, I get different configurations of the solute, and from these configurations, I start a new simulation. During energy minimization, errors such as:
Steepest Descents converged to machine precision in 15 steps,
but did not reach the requested Fmax < 0.01.
Potential Energy = 3.7985155e+16
Maximum force = inf on atom 682
Norm of force = inf

This problem occurs in all configurations that I submit to a new simulation.
The solution that I am using is to increase the size of the simulation box, centralizing the configuration obtained via cluster and filling the empty space with the same solvent. The central portion of the new simulation box is similar to the configuration obtained with the cluster. It doesn’t make sense to me that I can’t perform an energy minimization on any frame of the simulation.

Does anyone have any idea how to work around this problem without having to modify the simulation box?


Most likely you are packing box too densely with solvent molecules, which results in steric hindrance and bad contacts. If you wish to increase box size, why not simply using larger box…

Dear Masrull, thank you

The original simulation where the configurations were selected has negative potential energy. This simulation performed very well, without any problem. In my point of view, the cluster analysis just selects few configurations from this simulation. In my case, I ask to cluster to select conformations with solute RMSD larger than 0.3 nm.

I don’t want to increase or modify the simulation box, because I need to perform a simulation in the conformation obtained with cluster analysis. I told that if I increase the simulation box I was able to perform the simulation using the same conformation from cluster (kernel).
However, that doesn’t make sense and the problem remains.
If I have a configuration taken from a simulation, why can’t I just use it as a starting point for a new simulation? And why increasing the simulation box solves the problem?


I suggest to visualize the starting structure and to look at 682.
That may help you to understand what is going on.
It may occur that 2 solvent molecules (one of your cluster and other from the new solvation) are too close to each other.