[phenixbb] High B Factors

Dale Tronrud det102 at uoxray.uoregon.edu
Wed Dec 2 22:50:25 PST 2009

    The Wilson B is not "an average B caluclated from the reflection".
It is an estimate of what the B factor of your structure would be if
all the individual B factors were equal to each other - and of course
they are not.  While it is an interesting property of a set of
diffraction intensities it is not an estimate of the average of the
individual B factors in your PDB file.

    The main difference, in effect, is that the largest B factors for
atoms in your crystal are ignored in the Wilson B calculation.  This
omission results in the Wilson B always being lower than the mean of
the individual B factors.  The pattern of B factors you describe is
quite normal and expected.  I would worry if they were equal.

Dale Tronrud

Miles Pufall wrote:
> Hello All - 
> I am working on a kinase:inhibitor complex with some interesting 
> properties.  First, it is a long, thin unit cell (45x45x300) that has 
> highly anisotropic B-factors (60 in two dimensions and 35 in the other 
> according to xtriage).  We had a fair amount of trouble collecting data 
> for these complexes due to somewhat high mosaicity (1.5) and overlap 
> trouble, mostly due to the long unit cell.  As a result, although we get 
> diffraction out to ~2.3 A with good signal (I/sigma of 3), the 
> completeness is only OK - dropping off from a high of 95% around 3.5 A 
> pretty much linearly to 75% at 2.3.  
> As a result, the maps look only OK for 2.3A data, but clearly show the 
> interactions between the inhibitor and the kinase that we are looking to 
> define.  In that sense, the experiment worked just fine.  That said, the 
> structure looks pretty good, although about 1/3 shows poor density for 
> sidechains, which I wouldn't expect at this resolution.  But as I put 
> the finishing touches on the structure a couple of things have worried 
> me - first, there is consistently a larger than average gap between R 
> and Rfree (Typically in the R~22 and Rfree~29).  This, according to 
> Polygon, is within the range of reported structures, so I think that's 
> OK. What is odd is that despite having an average B caluclated from the 
> reflection of ~50, the average B after refinement is ~80, which is very 
> much on the high end of what has been reported.  I have played around a 
> lot with the wxc/wxu balance, and although this can change the Rs and 
> tends to tighten the geometry perhaps inappropriately (rms bond ~0.003, 
> rms angle ~0.7), the B factors stay very high.  
> My question is, should I worry about this?  And if so, what can you 
> recommend trying.  As usual, I'm happy to provide data if it help 
> elucidate the problem.
> Thanks in advance -
> M
> Miles Pufall
> Postdoctoral Scholar
> Yamamoto Lab
> UC San Francisco
> Cellular and Molecular Pharmacology
> Mail Stop 2280
> 600 16th Street, Genentech Hall S-574
> San Francisco, California 94158-2517
> (415)476-4480
> ------------------------------------------------------------------------
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

More information about the phenixbb mailing list