Re: [ccp4bb]: b-factors, tls-refinement

[Ganapathy Sarma <sarma@odysseus.als.orst.edu>]

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


Dear all,

My apologies, first, for sending the blank email by mistake.

Dirk, I have also experienced a similar behavior in the refinement of a
structure that came out of our lab recently. In our case, it was a 2.6 A
structure with the R and Rfree at 27.3 and 31.5 % respectively after CNS
refinements. The average B-factors were 65 A*2. The CNS refinements
indicated crystal anisotropy, because of which we tried REFMAC with TLS
refinement. Using TLS, the R and Rfree dropped to 25.1 and 30.2
respectively with the average B-factors increasing to 80 A*2! 

At this point, to assess the validity of the B-factors, CNS refinement was
carried out on the model. The B-factors dropped to 65 A*2. Combining the
overall anisotropic scale factors with the average B-factor lead us to the
insight that the effective average B-factors are different in the three
directions (75, 75 and 47 A*2 along the x, y and z-directions
respectively). The average B-factors are complex and its 
calculation/definition, I would say, is somewhat program dependent.

In your case, I am curious about the drop of 5%. During your CNS
refinements, did you have anisotropic scaling for the B-factor refinement.
Also, since you have atleast three chains in the asu, did you have NCS
restraints for B-factors that led to a lower average B-factor of 45 A*2
for all the chains?

Cheers,
Sarma 
 

On Fri, 11 Apr 2003, Dirk Reinert wrote:

> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
> 
> hi all,
> i am trying out tls-refinement after refinement in cns remained static. (the 
> statistics for the dataset i am using are excellent and the resolution limit 
> is 2.1 a.) the r-factors (in cns) were about 24 (r) and 28 (rfree) and the 
> b-factors for the different chains in asu were around 45. after doing 
> tls-refinement in refmac5 r and rfree dropped by 5%, respectively, and the 
> total bs for two of the chains are now about 80 (!). has anyone experienced 
> such an increase of the b-factors or has any comments on that, as i am a bit 
> worried on the correctness of what i am doing.
> thanks in advance, dirk.
> 
> -- 
> Dirk Reinert
> Universität Freiburg (FRG)
> Institut für Organische Chemie und Biochemie
> Alberstr. 21
> D-79104 Freiburg
> 
> Phone:  +49 761 2036088
> Fax:    +49 761 2036161
> 
> 

_________________________________________

Ganapathy N. Sarma.                      
Department of Biochemistry and Biophysics
Oregon State University                  
Tel: (541)737-3196
_________________________________________