[gmx-users] Restrain on principal axis to inhibit protein rotation

[Soeren von Buelow]

Dear all,

I would like to simulate an elongated ('stretched-out', rod-like)
protein in an orthorhombic or tetragonal box to simulate as few solvent
particles as possible and save computation time. The longest axis of the
protein would be aligned to the box axis with largest side-length.

The problem that I have with this approach is that the protein
(obviously) tends to rotate and see its periodic images, once the
longest axes of protein and box are no longer aligned. I'm not
interested in the rotational diffusion of the protein, so I'm fine with
imposing a soft collective restraint on the orientation of the protein
to keep it from rotating.

I don't want to restrain each atom individually, because that would mess
up the dynamics of the protein, but would rather like to impose an
energy penalty (harmonic or flat-bottom potential or anything actually),
if the vector of the largest principal axis of the inertia tensor
deviates by too much from the starting structure. This would allow
internal fluctuations to continue happily, as long as the protein does
not rotate too much globally.

Do you have any idea how to do that in gromacs? I only found an
implementation in a hacked older gromacs version from this paper

dx.doi.org/10.1021/ct100666v | J. Chem. Theory Comput. 2011, 7, 1381–1393,

but I would like to stick with some of the recent gromacs implementations.

Thanks and best regards,

-- 
Sören