Dear Peter,
This simply should not be happening.
The actual value of the weight term btw does not say much, what we
would need to know is the resulting rmsd, or the associated z-score.
For what you describe, weighting should not matter really - same goes
for the torsion restraint, that will not help to keep rotamers fixed.
As for what you say, I am wondering if you are dealing with double
conformations, where the positive
peaks suggest alternative conformers and the current position with
negative peaks has less occupancy.
You should have a few f these at 1.8 A.
I am also wondering if you do a simple logistics error on the model
you are refining and the data -
REFMAC might not move things, but move them in negative density peaks
is hard to believe.
As for the idea to fix some atoms is not a good one, they should be
refining in the right places
as soon as you get everything sorted.
Good luck -
Tassos
On 30 Sep 2009, at 18:22, Peter J Stogios wrote:
> Hi,
>
> I'm working with a 1.8 A structure with Coot and Refmac, and there are
> many sidechain rotamers that show very clear difference density peaks
> for setting their correct positions. However, Refmac continuously
> moves the rotamers back into negative density peaks. It's really
> quite silly because often there is an obvious positive density peak
> near to a negative density peak.
>
> I have tried using automatic geometry weighting and manually setting
> the weighting term to a very tight 0.025, but each has no effect. I
> have also tried increasing the torsion angle restraint term to 2.0 but
> this also has no effect.
>
> Does anyone have any suggestions? Is there any way to "fix" atom
> positions for Refmac?
>
> Thanks in advance,
>
> ~
> Peter J Stogios, Ph.D.
> Postdoctoral Research Fellow
> e: [log in to unmask]
> p: 416-978-4033
> w: http://www.uhnres.utoronto.ca/centres/proteomics/
>
> Structural Proteomics in Toronto Research Group, University Health
> Network
> C.H. Best Institute
> 112 College Street, Room 70
> Toronto, Ontario, Canada M5G 1L6
|