Re: [ccp4bb]: B factors

[<david.borhani@abbott.com>]

Gina wrote:

> we (postdocs) are having a discussion in our lab about structures that
> are published with high B factors ( including in the wilson value) along
> the protein backbone and the legitimacy of these structures. We have
> seen several papers that base their arguments on specific amino acid
> interactions and yet the structures have Bfactors 70 + up to even
> 180......A recent one had an average B factor of 68 ( 60 for solvent)
> and wilson  value of 59 at 2.5 angstrom. Perhaps not the worst
> example.....but.....So I guess we would like some advice as to what is
> the current opinion about B factors, resolution and good structures?

At the risk of starting a protracted web discussion...

I was the originator of one structure that has very high B-factors (1av1, apolipoprotein A-1, determined at 4 Ang. resol., 72% solvent). Side chain density in many regions was quite clear (structure factors were "up-weighted" with negative temperature factor, ~-70 A^2, prior to map calculation, to correct for the extreme fall-off in SF amplitude w/ resolution). At no point were phases from any model used to generate electron density maps, so the refinement was biased only by common knowledge of chemical and protein structure (tried to get bond lengths/angles correct, tried to rebuild residues with obviously bad phi/psi angles, etc). (Full) Coords & SF's are deposited at RCSB (Gerard: perhaps a separate petition for SF's?), so you can have a look for yourself and tell me if you agree or not. I think Gerard's EDS server has it as well.

Group B-factors were refined, where each group consisted of several (11 or 22) amino acids (I believe it was each chunk of alpha-helix between the repeated proline residues), and the differing B-values clearly "fit" the higher and lower mobility evident in the experimental map. (Crystal contact regions had lower B-factors than the long stretches between contacts.)

Regarding what is believable, I think that must be decided on a case-by-case basis. Just to put things into Debye-Waller perspective, B = 8*pi^2*<U^2>, where <U^2> is the mean square amplitude of atomic vibration and pi is 3.14159... A B-factor of 150 Ang.^2 is equivalent to U = 1.38 Ang. (rms). Saying an atom is "here", +/- 1.4 Ang. (rms) is OK, if the atom is really present. It is very different than saying it is here +/- 0.1 Ang or here +/- 10 Ang, however. The trick is: "Is it really present?" You need to see the (unbiased) electron density map to know for sure!

If you look at 1av1, you'll see that some atoms have very high B's, yet the density clearly shows them to be present at roughly where they were modeled. Others have even higher B's, and are indeed much less reliable, though actually (I believe) roughly (within 2 A) where modeled...if not, they would be far outside the (exptl.) electron density. Of course, in this situation, you need to be very circumspect about claiming interactions. Nonetheless, we were able to describe some clear interactions regarding salt bridges that have been supported by further studies in other labs.

Cautions: If electron density maps are calculated using phases derived from the model, you can quickly get a map that supports your favorite interpretation, i.e., high B regions can have what appears to be very respectable electron density. Model bias is a real and present danger, always, no matter the resolution. If you see high B's for the background and low B's for side chains, that is usually a sign of something being wrong. Also, some programs will let the B "peg-out" at a certain low or high cutoff (2 A^2 or 100 A^2), so look carefully. Checking programs are available (e.g. WhatCheck), and also have a look at the all-atom model in O, etc., coloring by B-factor...core of molecule should be "cool", extremities "warm". 

Hope this helps!
Dave
~~~~~~~~~~~~~~~~~~~~~~~~
David Borhani
Group Leader, Biochemistry
Vox:        508-849-2944
Fax:        508-755-8361
Email:        david.borhani@abbott.com
Smail:         Abbott Bioresearch Center, Inc.
        100 Research Drive
        Worcester, MA 01605 U.S.A.
~~~~~~~~~~~~~~~~~~~~~~~~