[gmx-users] question on diffusion during MD

[Mark Abraham]

Ramon Crehuet wrote:
> Dear gmx users,
> 
> I am running MD simulations of a tetrameric protein and as I want to
> compare with neutron scattering data I need to take into account protein
> diffusion implicitly. If I use /nstcomm = 1/ (default value) the
> simulation (1 ns) leads to a minimal random motion (traslational and
> rotational), but when using/ nstcomm = total_steps_of_the_simulation/
> the protein is moving much more towards a quite defined direction. Am I
> actually introducing traslational diffusion or is it an artifact?

It's an artefact. Look at comm_mode, or set nstcomm = 0 to avoid using 
these algorithms.

Mark