Re: Trouble using rotateReference, centerRefrence and refPositionsGroup in the colvar module

From: Giacomo Fiorin (
Date: Wed Feb 01 2012 - 19:31:53 CST

Hum, this is a very peculiar situation... I wish you had defined the
variable in terms of atoms that move together with the peptide (such as
certain side chains) instead of dummy atoms if you were planning on
allowing the protein to move later.

In any case, I don't believe you need to rotate the dummy atoms. The dummy
atoms won't drift or rotate with the protein. Instead, it's the protein
atoms that you should fit using the initial structure to go back to the
frame where the protein didn't move. So you only apply the fitting options
to group1.

Regarding why the PDB file doesn't work, you need to match the number of
coordinates of the atoms that you use for fitting (refPositionsGroup), not
the number of atoms of group1. The problem in providing a file with the
whole system is that the PDB reader will by default read the coordinates of
group1, not those of refPositionsGroup.

If you write a PDB file that just contains the five positions of the
refPositionsGroup atoms, it should work.


On Wed, Feb 1, 2012 at 4:44 AM, Ajasja Ljubetič

> Dear Giacomo,
>> Hello Ajasja, first of all, the centerReference and rotateReference
>> options only affect the group they are defined into. If you need to center
>> and rotate all groups of a colvar component (in this case, the dihedral
>> component), you should copy those options inside each group. A future
>> version of the colvars module will probably include "reusable groups" for
>> this purpose, but in the end you achieve what you need with some cutting
>> and pasting of the keywords (centerReference, rotateReference,
>> refPositionsGroup, refPositionsFile) inside group1 through group4.
>> Oh, copy-pasting is no problem. I just misread this<> error
> message.
> I don't know about the different number of atoms, since I can't see the
>> PDB file you provided.
> Here <> is the PDB file. It contains all
> the atoms of my system (and hope that coordinates are selected by index as
> stated in the user guide)
> So to answer your final question, you don't need a relative reference
>> coordinate system for a dihedral.
>> Or perhaps you do, but then I have to ask you what you were trying to do
>> when you modified the original configuration?
> I use the dummy atoms to set the reference for the spherical
> coordinates. So the phi is defined by a dihedral component as per your
> excellent suggestion of almost one year ago :)
>> Hello Asjasja, have you tried defining a dihedral, using the four atom
>> groups you mention? Perhaps I'm missing a small detail, but it seems to me
>> that it would provide you what you need, at least when you define the two
>> axes dynamically, based on the centers of mass of those groups.
>> If the difference is only adding the two axes as fixed in space (with
>> axis and planeAxis), based on the gradients and system forces of dihedral,
>> either Jérôme or I can derive fairly easily an alternate code.
> (here <>is
> an example -- the first three atoms from bottom to top are the dummy atoms,
> the fourth is the COM of the spin label ring). This works nicely as long as
> I keep the protein fixed.
> Now I would like to release the protein. This is why I would like to keep
> the dummy atoms in the same position *relative* to the heavy backbone atoms
> of the spin label.
> How difficult would it be to implement the rotation an centring of the
> dummy atoms (when used with a valid ref positions group)?
> Thank you & best regards,
> Ajasja

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:10 CST