Re: [ccp4bb]: How to deal with sidechainatoms with low electrondensity?

[Bart Hazes <bhazes@ualberta.ca>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

On Fri, 22 Nov 2002, Florian Schmitzberger wrote:

> Dear all,
>
> I was wondering what the most acceptable way to deal with (disordered)
> sidechains/residues (e.g. sidechains of Arginines and Lysines) with very
> low or no electron density is? Especially with respect to submission of
> coordinates to the pdb-data bank.
>
> My initial approach was to set the occupancies of the atoms to 0 and then
> refine in Refmac. However I encountered that the geometry (bond lenghts
> and angles) gets distorted between the atoms with normal occupancy and the
> ones with zero occupancy. The other option to refine the residues as
> alanines does not seem a very good idea to me, since it does not account
> for something which is there, but strongly disordered. For me the most
> ideal way to deal with them is to fix the B-factor to a very high value
> and then refine with Refmac.
>
> What is the general opinion about this matter?

We have already heard a few "general opinions" which means that there is not a
clearly accepted code of best practise for this situation. I don't like either
the 0.01 or 0.00 occupancy or high B-value solutions. In both cases the naive
end user sees a sidechain in the model and will firmly believe that those
atoms are exactly there and nowhere else. Of course it takes only a quick
check of B-factor or occupancy to find out but I wonder how often that is
done. I often don't even do that myself unless I have reason to be suspicious
of a part of the structure.

The practise in my lab is to build what you see and to explicitly remove the
atoms you don't see. So a lysine may have its beta and gamma atoms modeled but
not the rest or a glutamate may lack its carboxylate. The residue name will of
course still be the real amino acid at that place. Some programs may complain
about incomplete residues but we don't have a problem with it. The best thing
is, I hope, that a user will actually notice that there are atoms missing and
therefore is alerted to the fact that these atoms are disordered (although we
may have educate people that missing carboxylates can be due to radiation
damage as well).

A different approach is to consider a structure as a model that accommodates
both experimental data and basic protein geometry rules. When there is strong
experimental evidence you build whatever the data supports even if that
involves an unusual dihedral angle or so. When there is no data you build
whatever molecular geometry dictate. In disordered regions where experimental
data is weak one could thus build all common rotamers of a sidechain, removing
those that are clearly stericaly impossible and set all occupancies to zero.
That should also really stand out to the user and, like the NMR world, we can
claim to "see" dynamics for those regions where we really see nothing. It may
even be possible to refine occupancies for the various rotamers if we restrain
or even constrain the atomic positions of the sidechain atoms (make them ride
on on the mainchain atoms).

Bart

===============================================================================

Assistant Professor
Dept. of Medical Microbiology & Immunology
University of Alberta
1-15 Medical Sciences Building
Edmonton, Alberta, T6G 2H7, Canada
phone:	1-780-492-0042
fax:	1-780-492-7521

===============================================================================