Hi, Mousheng and Michael,
If you haven't tried this already, using your most recent refined models, calculate maps and rebuild the entire structures with ARP/wARP. That might help. Good luck,
Hidong
| Raji Edayathumangalam <[EMAIL PROTECTED]>
Sent by: [EMAIL PROTECTED] 08/28/2006 08:18 AM |
|
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
There has been a "stuck refinement" discussion going on over the ccp4bb
for the past few days.
Your unit cell values make me think you might not have twinning.
Twinning in P.O. s.g. is not very common except when two unit cell
lengths are very similar. Folks correct me if I am fibbing here.
If you haven't seen the discussion yet, you are probably going to hear
the foll:
1. look for sharp increase in rfac/rfree with resolution
2. so your data truncated at resolution where the i/sig is 9?? how about
data beyond/higher than 2.1 Ang resolution?
3. not clear from your email if you have already have added waters, ions
etc. if not, this might help
4. how much of protein is accounted for in the model? are there regions
with terrible bfactors?
5. i do not know if "optimize wa" to optimize the x-ray term is a
helpful suggestion for your case. say what, others?
Sorry for this laundry list. Good luck.
Raji
Michael Hothorn wrote:
> *** For details on how to be removed from this list visit the ***
> *** CCP4 home page http://www.ccp4.ac.uk ***
>
>
> Dear list members,
>
> I am currently trying to refine a structure in primitive orthorhombic
> space-group P212121 with the cell dimensions:
>
> 46.24 66.63 73.79 90 90 90
>
> tha data process nicely in the primitive orthorhombic setting: Rmeas
> overall 0.067 and 0.17 in the highstet shell, 100% completeness and
> I/SIGMA 22 and 9 in the highest shell. Based on the systematic absences
> along H,0,0 0,K,0 0,0,L, the space-group is P212121 and I found a nice
> solution with two molecules in the AU using phaser. Clear difference
> density shows the location of two 10-residue sized peptides that I build
> in. However, at 2.1 A resolution refinement in either Refmac (including
> TLS refinement) or CNS gets stuck at a R/Rfree of 0.27/0.33. The geometry
> looks very resonable and I only find two sensible Ramachandran outliers
> in the peptide region.
>
> As there is obvioulsy a problem in refinement, I have tried to solve the
> structure in either P222, P21212 or P2221 without success. I then
> processed the data in P2/P21. The data processing statistics look very
> similar to the oP case with a beta angle very close to 90 deg (90.010).
> Of course I can also solve the structure in P21 and in P1, however I do
> not see a clear improvment in the refinement.
>
> As there is no strong difference density or geometry problems, I am
> wondering what might be going on.
> Do the cell constants and SG settings allow for any type of twinning? I
> had a look at the cumulative intensity distribution in scala without
> spotting anything strange. Also, the packing rules out the presence of a
> third molecule... What other problems could I test for?
>
> any help/hint/suggestion is highly appreciated
> thank you!
> michael
>
>
> Michael Hothorn, Ph.D.
>
> Structural & Computational Biology Unit
> European Molecular Biology Laboratory (EMBL)
>
> Meyerhofstrasse 1
> 69117 Heidelberg
> Phone 0049(0)6221 387 268
> Fax 0049(0)6221 387 519
> http://www.hothorn.de
>
>
