[gmx-users] umbrella sampling of angles

[David Mobley]

Dear all,

Does anyone know if it possible to somehow do umbrella sampling in
angles (rather than distances) using the pull code? The stuff I turn
up in the manual all only seems to involve distances.

Alternatively, is there any way to make angle restraints use only
simple harmonic springs rather than the cosine function that is
typically used?

My current problem is that I need to compute the PMF for rotating a
small molecule through a particular angle. Using the angle_restraints
cosine functional form gives me two zeros of the force when I only
want one. If the energy landscape were smooth, this would be no
problem, because one of the zeros is an unstable fixed point (the
potential energy is high) and only the other is stable. But since the
energy landscape is pretty rough, my molecule can actually get trapped
at the unstable zero when it happens to coincide with a minimum in the
energy landscape. This would obviously be solved by using simple
harmonic restraints (as done with distance and dihedral restraints,
but not with angle restraints, strangely enough).

Is there a work-around? Or do I have to implement the harmonic form myself?

Thanks,
David