Re: Colvar orientation and CoG

From: Jérôme Hénin (
Date: Fri Jun 17 2016 - 04:43:06 CDT

Hi Nicolas,

The orientation coordinates are defined as purely rotational, that is, they
are orthogonal to overall translation of the object. For that reason,
forces applied to those coordinates have zero sum, and will never cause
center of mass motion directly. However, they do not prevent center of mass
motion: if the object is tethered by a hinge at one end, it will tend to
rotate around the hinge.

You may want to apply a torque "around the hinge" so that it is not
translation-neutral, but then I don't know what coordinate that would be
the gradient of. Then it would be simplest to forgo collective variable
biases entirely, and write your own scripted forces that apply a torque
around the axis of your choice.


On 16 June 2016 at 16:02, Nicolas Martin <> wrote:

> Dear users,
> I am currently trying to restrain a complex motion in a protein. I have a
> starting and an end structure that I use to compute a quaternion with NAMD.
> The rotation I get is described by an angle inferior to 90 deg and for
> which the center is not the center of geometry of my selection. You can see
> the motion I'm trying to describe more as an hinge motion than a self
> rotation.
> In a second time I tried to use this quaternion and a harmonic restrain on
> my starting structure to reproduce the end structure. Since the axis of the
> rotation described by the quaternion previously computed is translated
> toward the cog of my selection (I think I saw that into the source code of
> spinangle and orientation) the rotation applied is different from the one I
> initially computed since the axis / center of rotation are different.
> So here are my two questions. Can one just comment the lines corresponding
> to the translation in the code ? And if no why?
> Thanks a lot for the advice, bests
> NM

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:16 CST