[gmx-users] Separate removal of COM motion

[Ángel Piñeiro (R)]

Dear all
I am involved in a project where the target systems are lipid bilayers 
with and without different macromolecules embedded. In the past I did 
quite a lot of MD simulations of membrane systems always removing the 
COM motion of lipids and the solvent in separate groups. There are 
several discussions about this in the gmx users list (see for instance: 
https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-users/2016-September/108590.html) 
but I do not understand clearly the reason to do this. I am very 
interesting in determining diffusion coefficients, so the right removal 
of COM is key for my analysis. Can anyone explain why the global removal 
of COM for all the system might produce a lateral relative displacement 
of the bilayer with respect to the solvent in opposite directions (thus 
keeping the COM)? Is it guaranteed that this is solved by removing the 
COM separately for lipids and the solvent? When I have macromolecules 
embedded in the membrane, should I remove the COM of the macromolecules 
together with the lipids?

Thank you in advance for any help,

Ángel.