grouped B-factors too low??

Zbyszek Otwinowski zbyszek at mix.swmed.edu
Wed Nov 6 15:16:38 EST 1996


Pirkko Heikinheimo BTK wrote:

> we are refining several structures from data collected in cryo temperatures
> and have trouble refining the bfactors. We are starting off with the grouped
> B-factor refinement at 2.3-2.5 A resolution. Currently my structure is at
> 2.3 A, with 27671 reflections used for 4620 atoms (560 residues and 179
> waters). The problem is that if I refine the B-factors I end up with lots
> of B-factors of 2.0 (the minimum) and at some places, lower B-factors for
> the main chain than for the side chain portion.
> 
> Is there some tricks that I should try, something related to cryo data
> compared to the one collected at room temperature? Or is this just a
> reflection of underfitting the data still, too many well ordered waters
> that I just haven't modelled in yet? Solvent flattening?
> 

1.) Do not apply sigma cutoff after integration of diffraction peaks.
Defaults in many programs are completely wrong and result in artificialy
larger (than correct) values of structure factors at high resolution.
If you apply 1-2 sigma cutoff to random data of high redundancy
merging will produce mostly 2-4 sigma (on intensity) hkls,
or in structure factor terms (Fobs) 4-8 sigma hkl. Larger than correct
Fobs -> lower B factors in refinement, also larger errors of B-factors.

2.) Lack of solvent correction in refinement will dramatically lower
refined B-values relative to their true values.

3.) At 2.3-2.5A errors of B-factors are quite large. Low values may
be simply statistical fluctuations.

4.) Restrain or contrain B-factors. unrestrained B factor refinement
requires higher resolution data.


-- 
  Zbyszek Otwinowski                     |        zbyszek at chop.swmed.edu
  University of Texas                    |        tel : (214)-648-5098 
  Southwestern Medical Center            |        fax : (214)-648-5095 
  5323 Harry Hines Boulevard  Dallas, Texas  75235-9038
[EOB]



More information about the X-plor mailing list