X-PLOR vs Hydrogens

[Yoram Puius]

Thomas J. Stout wrote:
[snip]
> The reason I am concerned is that the B-factors assigned
> to the protons appear to be the 'refined overall B-factor'
> from the beginning of the job, which isn't particularly
> appropriate.  Shouldn't the assigned value more closely
> approximate a 'riding model' under the assumption that
> the B-factor of a particular proton will be pretty closely
> related to the B-factor of the heavier atom to which it
> is attached?
> 
> -Tom

While that would make sense, there's really no way to judge the
B-factor of a hydrogen, so X-PLOR will refine them with the
"group" B-factor at the beginning and then leave them alone.

B-factors estimate how the atom in question is "smeared" over
space.  Since protein crystallographers almost *never* see
hydrogens, they never estimate the B-factors for them.  I suspect
the hydrogens won't affect the R-factor anyway.  (Anyway, lots of
hydrogens -- on the ends of lysines, or methyl groups -- are
completely disordered.)

The new X-PLOR allows you to run "generate.inp" without
building hydrogens.  It makes refinements run a bit faster.

If you are doing an NMR structure, or anything other than
protein X-ray crystallography above 1.5 A resolution, everything
I say is null and void and does not apply <grin>, and you might
want to ask the question again specifying exactly what sort of
structure you're doing.

By the way, you could have X-PLOR assign the B-factor from
the amide nitrogen to the amide hydrogen at various points in
the refinement (create your own 'riding model').
COntact me if you want help writing the
X-PLOR  code.

Later,
	Yoram

-- 
_______________________________________________________________________
Yoram A. Puius                Albert Einstein College of Medicine
mailto:puius at aecom.yu.edu     Department of Biochemistry
5th year M.D.-Ph.D.           1300 Morris Park Avenue, Bronx, NY  10461
_______________________________________________________________________