parameter sets

breeze at breeze at
Fri Nov 24 08:59:31 EST 1995

Fellow X-PLOR users,

Looking at some crystal structures (other peoples') and NMR structures
(my own) of homologous proteins overlayed recently, I noticed some
quite wide variations in bond angles between the X-ray and NMR structures.
For instance, the CA-CB-CG angle for Phe in the crystal structures was
around 113-114deg, and ~110deg in the NMR structures.

The NMR structures were calculated in XPLOR using the
parameter set, so it's not surprising given the magnitude of the angle
force constant that the angles in the structures are pretty close to the
109.5deg of the parameter set. The crystal structures were also both 
refined in XPLOR, presumably using something like the
parameter set.

Looking at and comparing with reveals quite
a lot of differences in bond angles. These, I understand, are taken
largely from Cambridge Crystallographic Database small molecule crystal
structures in the case of (Engh & Huber forcefield). The
values in, on the other hand, seem to be mainly "ideal"
values for tetrahedral carbons etc.

My question is this: which is "right"? Do the differences matter -- is
the forcefield we use for NMR structure determination too simplistic?
(One might imagine that small "errors" in bond angles might propagate
over longer distances and contribute to experimental NOE violations,
for example.) Or are the effects of these small differences well within
the intrinsic error limits of our experimental data?

I just have in mind the thought that NMR structure determination is a bit
more heavily reliant than crystallography on "chemical knowledge" about,
for instance, bond angles and van der Waals radii, given its less
favourable observable-to-parameter ratio. If these forcefields are not
"right", then how "accurate" can our structures be?


Alex Breeze
Protein Structure Laboratory
Zeneca Pharmaceuticals
Alderley Park
Cheshire SK10 4TG
email: breeze at

More information about the X-plor mailing list