[phenixbb] phenix.refine change: support for residues onsymmetryelements

[Edward A. Berry]

The pdb file is expected to contain the contents of the asymmetric unit.

Suppose the end-user of the pdb file is a protein chemist who has just done a 
super-accurate atomic absorption analysis of her ultra-pure protein and determined the 
ratio of Cu to Fe is 0.4973.  If the pdb file shows one Cu and one Fe, she probably won't 
realize the Cu is on a 2-fold so only half is present in each asymmetric unit.

So let the crystallographers use whatever convention works for them, but in the PDB the 
file should contain the asymmetric unit (or some fraction of it, in the case of viruses)

Pavel Afonine wrote:
> Hi Maia,
>
>> when I submitted a pdb with some waters on a twofold sym. axis with
>> occupancies =1, the PDB wanted occupancies 0.5.
>> They do not want to take into account your "taken into account
>> internally".
>
> Hopefully, time will heal it -:)
>
> More seriously (I wrote it somewhere for someone, even not once):
>
> The maximal multiplicity of a special position in proteins is 24 (or so,
> correct me if I'm wrong, please), and the precision of the occupancy
> factor field in PDB file is only two digits. Let’s imagine a relatively
> small structure at high resolution that contains a handful amount of
> heavy atoms all located at special positions with the multiplicity 24.
> Now, the occupancies of these heavy atoms in a PDB file all have to be
> set to 1./24 ~ 0.04 and not to 0.042. Depending on the number of
> electrons in a heavy atom, the number of heavy atoms, multiplicity of
> special position and overall structure size, one can easily imagine a
> situation when this kind of rounding error may become very significant.
> Therefore, I think, it is a bad idea to rely on the file format in this
> case, and is much better to do this trivial operation internally without
> any unnecessary loss of accuracy. Plus, it all happens seamlessly for
> the end-user, which is another step towards the automation.
>
> Pavel.
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>