Tables for
Volume F
Crystallography of biological macromolecules
Edited by M. G. Rossmann and E. Arnold

International Tables for Crystallography (2006). Vol. F, ch. 21.1, p. 500   | 1 | 2 |

Section 21.1.6. Final model

G. J. Kleywegta*

aDepartment of Cell and Molecular Biology, Uppsala University, Biomedical Centre, Box 596, SE-751 24 Uppsala, Sweden
Correspondence e-mail:

21.1.6. Final model

| top | pdf |

Once the refinement is finished [i.e. once the (FoFc, αc) difference map is featureless (Cruickshank, 1950[link]) and parameter shifts in further refinement cycles are negligibly small], three tasks remain: validation of the final model, description and analysis of the structure, and deposition of the model coordinates and the crystallographic data with the Protein Data Bank (Bernstein et al., 1977[link]).

Until a few years ago, validation of the final model typically entailed calculating the conventional R value, r.m.s. deviations from ideal values of bond lengths and angles, average temperature factors, and a Luzzati-type estimate of coordinate error. Kleywegt & Jones (1995b[link]) showed that these statistics are not necessarily even remotely related to the actual quality of a model. Based on these criteria, a backwards-traced protein model was of higher apparent quality than a carefully refined correct model. After this, the realisation sunk in that the best validation criteria are those that assess aspects of the model that are `orthogonal' to the information used during model refinement and rebuilding. For instance, the main-chain ϕ and ψ torsion angles are usually not restrained during refinement; this makes the Ramachandran plot such a powerful validation tool (Kleywegt & Jones, 1996b[link], 1998[link]). Other examples of useful independent tests include the profile method of Eisenberg and co-workers (Lüthy et al., 1992[link]), the directional atomic contact analysis method of Vriend & Sander (1993[link]) and the threading-potential method of Sippl (1993[link]).

In general, all quality checks provide necessary, but in themselves insufficient, indications as to whether or not a model is essentially correct. A truly good model should make sense with respect to what is currently known about physics, chemistry, crystallography, protein structures, statistics and (last, but not least) biology and biochemistry (Kleywegt & Jones, 1995a[link]). A good model will typically score well on most if not all validation criteria, whereas a poor one will score poorly on many criteria. The same is true at the level of residues: a poor or erroneous region in a model will be characterized by violations of many residue-level quality criteria (Kleywegt & Jones, 1997[link]).


EU 3-D Validation Network (1998). Who checks the checkers? Four validation tools applied to eight atomic resolution structures. J. Mol. Biol. 276, 417–436.Google Scholar
Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer, E. F. Jr, Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, T. & Tasumi, M. (1977). The Protein Data Bank: a computer-based archival file for macromolecular structures. J. Mol. Biol. 112, 535–542.Google Scholar
Cruickshank, D. W. J. (1950). The convergence of the least-squares or Fourier refinement methods. Acta Cryst. 3, 10–13.Google Scholar
Kleywegt, G. J. & Jones, T. A. (1995a). Braille for pugilists. In Proceedings of the CCP4 study weekend. Making the most of your model, edited by W. N. Hunter, J. M. Thornton & S. Bailey, pp. 11–24. Warrington: Daresbury Laboratory.Google Scholar
Kleywegt, G. J. & Jones, T. A. (1995b). Where freedom is given, liberties are taken. Structure, 3, 535–540.Google Scholar
Kleywegt, G. J. & Jones, T. A. (1996b). Phi/Psi-chology: Ramachandran revisited. Structure, 4, 1395–1400.Google Scholar
Kleywegt, G. J. & Jones, T. A. (1997). Model-building and refinement practice. Methods Enzymol. 277, 208–230.Google Scholar
Lüthy, R., Bowie, J. U. & Eisenberg, D. (1992). Assessment of protein models with three-dimensional profiles. Nature (London), 356, 83–85.Google Scholar
Sippl, M. J. (1993). Recognition of errors in three-dimensional structures of proteins. Proteins Struct. Funct. Genet. 17, 355–362.Google Scholar
Vriend, G. & Sander, C. (1993). Quality control of protein models: directional atomic contact analysis. J. Appl. Cryst. 26, 47–60.Google Scholar

to end of page
to top of page