[ccp4bb]: High vs. low symmetry

[Wulf Blankenfeldt <wb6@st-andrews.ac.uk>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Dear all,

I am struggling with the final steps of a refinement. This protein 
apparently crystallises in spacegroup p21212 with a=47.5, b=71.9 and c=82.2 
with one monomer in the asymmetric unit. Initial phases to 2.3 A have been 
determined by MAD and I am refining the structure in REFMAC5. We have a 
couple of other data sets, going to about 2.6 A.

However hard I try I can't get R-free below 27 %, with parts of the density 
being not interpretable and other parts really fluffy. TLS-refinement in 
REFMAC5 shows that the protein moves around quite a bit. This had already 
been indicated by a "serious anisotropy" warning in TRUNCATE and high 
overall B-factors of about 70 A*A. I should maybe say that the images are 
very "smeary" with very big spots in the middle and a lot of background 
that looks like faint spots from the crystal (mosaicity refinement in 
MOSFLM doesn't take these into account).

Here comes the weird part: I recently had the chance to collect a highly 
redundant 2.2 A data set of over 300 degrees on ID14/1 at the ESRF (dinner 
time). Again the data looks like p21212 (R-merge 7.2 %). Systematic 
absences are there with some exceptions. This time I did not get an 
anisotropy warning from TRUNCATE and the overall B is 47 A*A. When I put my 
"fully refined" model in it behaves as follows in REFMAC5:

			R		R-free
Start			34.9		38.4
After TLS		27.4		31.6
After restrained	24.7		29.1


HOWEVER, when I work up my data in p1 with the same cell now having 4 
monomers in the ASU (R-merge 6.0 %) the refinement looks much better 
(monomers positioned with MOLREP):

			R		R-free
Start			35.8		34.6
After TLS		29.7		28.7
After restrained	23.4		24.7



The density in p1 seems marginally better, but the bad parts are still not 
really traceable (yet). The TLS parameters don't come down by this 
treatment. I should also mention that I assigned a new Free-R set for p1 
which is maybe not justified given the relationship between the two 
lattices, but I don't think this will account for 4.5 % (?), especially if 
you look at the numbers from the restrained refinement.

My question is this: How would you proceed with this refinement? p1 or 
p21212? It is my feeling that the fourfold number of parameters gives me 
more "dump-terms" to account for the not so good data, but then I would 
feel really bad to submit a structure to the PDB for which I could only 
brake the 30 % R-factor mark thanks to TLS-refinement in REFMAC5.


Thanks for any suggestion!




Wulf

-----------------
Dr. Wulf Blankenfeldt

Structural Biology Group
Biomolecular Sciences Building
University of St. Andrews

North Haugh
St. Andrews, Fife
KY16 9ST
UK

Tel:    +44 (0) 13 34 - 46 72 82
Fax:    +44 (0) 13 34 - 46 25 95
e-mail: wb6@st-andrews.ac.uk
-----------------