[phenixbb] B(sol) too high

[Nathaniel Echols]

On Dec 15, 2009, at 11:53 PM, Matthew Merski wrote:
>  I am in the process of solving a series of protein/ligand complex structures and ran the Polygon tool on the series.  I noticed that, in general, even when all the other statistics were good, the B(sol) term was a bit too high according to Polygon. (The structures with poorer statistics also had high B(sol) values as well.) Since this is a series of multiple structures of the same protein, should this be a concern for me? And if so, what are typical causes of a high B(sol) value?


I erred in making B(sol) one of the default parameters displayed in Polygon - it is obtained from bulk solvent correction and is not terribly informative by itself, and it will not change very much with refinement.  There is some natural variation in this value and it's a property of the crystal rather than the model (mostly; Pavel would know for sure).  A higher-than-average B(sol) is nothing to worry about as long as it's not an extreme value and your other statistics are okay.  If all of your structures form similar or identical crystal lattices (very common, if no major conformational changes are involved), I would expect the B(sol) to be similar for each dataset.

(The same applies to K(sol) - unless it's close to zero, it isn't very interesting.)

--------------------
Nathaniel Echols
Lawrence Berkeley Lab
510-486-5136
NEchols at lbl.gov