From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Thu Jul 18 2013 - 11:41:55 CDT
I think it boils down to whether you can assume that restraining the
orientation of the molecule carries a systematic error or not. If it does,
then applying the restraint to all the atoms of the molecule, or only to a
subset of it, won't fix the problem.
The ideal would be to use a cubic or truncated octahedral box, to
*not*apply any orientational restraint, and define instead all
collective
variables in the rotated frame of the bigger molecule by using the
refPositionsGroup as in:
https://github.com/colvars/colvars/blob/master/examples/8_relative_rotation.colvars.in?raw=true
Then, of course, I do understand that it saves quite a few atoms with a
non-isotropic unit cell. See also Axel's reply that I just saw popping up.
Giacomo
On Thu, Jul 18, 2013 at 12:21 PM, Thomas Albers <talbers_at_binghamton.edu>wrote:
> Hello!
>
> > Hello Thomas, the bias is only on the overall rotation, not on the
> internal
> > degrees of freedom: any force applied to the orientation colvar doesn't
> have
> > a projection along vectors that would change the internal structure of
> the
> > protein.
>
> Yes, but aren't the forces that are applied through the harmonic
> restrains acting only on the backbone?
>
> OK, one can argue that the random forces that cause the protein to
> diffuse away from its original orientation act only on the exterior,
> and the force applied to the backbone is of the same order of
> magnitude.
>
> I'm just slightly concerned that the colvar applies forces only to the
> backbone, and that might introduce systematic error.
>
> Regards,
> Thomas
>
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:27 CST