ADL: Question

Mark Swingle mswingle at jaguar1.usouthal.edu
Tue Jan 13 12:01:28 PST 2009


ISTM that there's no easy answer to your question.

Prima facie, that would seem to be a reasonable approach. OTOH, B-factors can 
hide a multitude of sins, so be careful. I would be very suspicious of high 
B-factors in models fit to low resolution data. Even when high B-factors 
aren't just hiding errors in the model, the displacements could actually be 
highly anisotropic, which might be due in large part to rotations of the 
whole protein or subdomain rather than just purely local motions. You could 
also have more medium-scale motions (of loops, for example), not modeled 
particularly well by just flexible sidechains, causing high B-factors.  Also, 
a low B-factor is no guarantee that the sidechain won't have a different 
conformation when a different ligand is bound or crystallization conditions 
change. 

What you really want are multiple crystal structures or NMR ensembles to guide 
you ;-) In the absence of that, perhaps MD simulations of the protein would 
be useful (but in that case you might as well use the relaxed complex 
method.....well, maybe not for VLS;-) ).

If you're going to use B's as selection criteria then 
1) assess the veracity of the model (are the high B's real? reported stats are 
important but also see #3 below)
2) look closely at the context (is it really small-scale local motion, which 
can reasonably be modeled via flexible sidechains? where is the residue 
situated? are there few/weak interactions "locking it down"? etc.
3) if you can, look at the data (try the Electron Density Server at Uppsala 
University if you don't have the necessary software) - this can be VERY 
illuminating ;-)

Regards,

Mark

On Tuesday 13 January 2009 12:18:23 pm Andrew Ward wrote:
> Has anybody ever looked at the correlation between sidechain B-factor and
> flexible residue choice?  i.e. Is it best to leave sidechains with a low
> B-factor rigid and sidechains with a high B-factor flexible when screening
> against a virtual library.  Any insight would be appreciated.  Thanks,
> Andrew
> ________________________________________________
> --- ADL: AutoDock List  --- http://autodock.scripps.edu/mailing_list ---




More information about the autodock mailing list