From: Ajasja Ljubetič (ajasja.ljubetic_at_gmail.com)
Date: Thu Feb 09 2012 - 12:05:19 CST
Dear Jrme and Giacomo,
Yes, you can use these two variables to find out the difference between
> colvars and VMD fitting. However, I don't suppose there is an issue with
> the vector distance between the two centers of geometry, right? :-)
>
No, no, but this data is still needed to align the two structures.
The function in the tcl file attached (sorry, I forgot to upload it in the
> VMD script library some time ago) prints you the rotation matrix that you
> can use with "atomselect moveby" in VMD. It is a 4x4 matrix as per VMD's
> format.
No problem, I managed to find a wrong formula for the conversion at
http://www.j3d.org/matrix_faq/matrfaq_latest.html (Q54). This is why I had
some problems. I should have just listened to wikipedia
http://en.wikipedia.org/wiki/Quaternions_and_spatial_rotation#Conversion_to_and_from_the_matrix_representation
PS: It's just "atomselect move".
My guess is that if you have a well-defined structure (read: thermal
> fluctuations are much smaller than the radius of gyration of the reference
> coordinates), the colvars' matrix and that computed by VMD via the Kabsch
> method should be pretty much the same. If you're not in those conditions,
> they may differ, but in that case they are both not very meaningful.
>
The RMSD as given from the colvars or calculated in VMD is exactly the same
(when using the correct weight as you pointed out). (shown
here<http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S9.GIF>).
So hopefully all my configuration files are OK.
Using a more defined structure is very good advice. So when I use the whole
backbone as a reference I obtain the correct
PMF<http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S8.GIF> as
opposed to what I had
before<http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S6.GIF>,
when using just the backbone atoms and the CB atom of the spin label.
Taking a closer look at the orientation colvar in both cases (in terms of
the rotation angle and the unit vector of rotation) was also very
educational (shown
here<http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S10.GIF>)
for me. Basically I expected that the five backbone atoms determine the
orientation quite well, but this obviously is not the case. Perhaps if one
would use a moving average of the atom positions the situation would
improve...
Anyway I'm just happy that the PMFs obtained
with<http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S8.GIF>
refPositionsGroup
and without <http://lbf.ijs.si/ajasja/mails/ABF/relative-coords/S7.GIF> are
the same. So the neglected derivatives of the optimal rotation don't appear
to cause any problems.
Thank you and best regards,
Ajasja
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:11 CST
