e16.html

SUMMARY: In principle, including them with zero occupancy makes sense, because such atoms do not contribute to the refinement, are clearly marked as modeled rather than observed, and do not cause problems such as changes in the sequence when a lysine, for example, is replaced by an alanine.

Recommended Reading: 1: Shneerson VL, Wild DL, Saldin DK.
An exponential modeling algorithm for protein structure completion by X-ray
crystallography. Acta Crystallogr A. 2001 Mar;57(Pt 2):163-75. [Abstract]

2: Roversi P, Blanc E, Vonrhein C, Evans G, Bricogne G.
Modelling prior distributions of atoms for macromolecular refinement and
completion. Acta Crystallogr D Biol Crystallogr. 2000 Oct;56 ( Pt 10):1316-23. [Abstract]

Alexander Wlodawer (NCI): It is difficult to decide whether "missing" atoms should be included in the final coordinate set or not. In principle, including them with zero occupancy should make sense, since such atoms do not contribute to the refinement, are clearly marked as modeled rather than observed, and do not cause problems such as changes in the sequence when a lysine, for example, is replaced by alanine. However, it is almost certain that such atoms will be interpreted by somebody somewhere as real. One solution would be to modify display programs so that they automatically change color for such atoms. Another would be to get PDB to put a much stronger marker for such atoms. In practice, I will probably continue to encourage including atoms with zero occupancy and B in cases where structure cannot be fitted properly, but I know that this will be found controversial, and I do not like this approach that much myself. However, I think that it is still better than nothing.

On the other hand, I would certainly not recommend using this approach for those parts of the chain (for example the termini) that are clearly not observed and for which it is not possible to create plausible models.