*Acta Cryst.* (1993).
D**49**, 355-356

**Pp. 427 and
399. Oxford Univ. Press (IRL Press), 1992
Price £55.00. ISBN
0-19-963302-9**

This atlas is a two
volume set describing all 400 possible pairwise
interactions between the 20 different types of amino-acid side chains.
It is an elaboration by the authors of the previously published paper:
`*SIRIUS*. An automated method for the
analysis of the preferred packing
arrangements between protein groups' [*J. Mol.
Biol.* (1990), **211**,
595-615].

The raw data consist of 62 highly refined structures
deposited in the
Brookhaven Protein Data Bank. An interacting pair is defined simply
based
on distance. A reference set of atoms is given for each amino acid. This
is used for the superposition of atoms in the pair and for calculations
of characteristic interaction angles. Their significance is assessed by
a chi^{2} test against a random distribution.

About 20 pages describe the making of the atlas. Then for each of the 400 pairs, two pages of information are presented. First, the atoms defining the geometry are given along with statistical data in the form of tables and histograms. Stereo line drawings giving three orthogonal views to highlight the clustering of the distributions are presented on the second page. To avoid clutter, at most 50 representative structures of the given pair are chosen randomly.

The selection of a reference set of atoms presents problems, as a subjective judgement must be made for most side chains. The choice made by the authors is to consider the functional groups or the extremities of the side chain, rather than a definition always involving the alpha-C atom. There is no consideration of rotamer conformations, nor any account of peptide-side-chain interactions here. There are only about 100 observations for each pair type with a median value of 56 observations. Approximately one fifth of the interactions have highly significant interaction angles (less than 5% likelihood that they are due to chance). Many of these significant patterns, however, are hard to discern by visual inspection of the stereo plots.

The overall presentation of the atlas is well done, but it would have been helpful had 20 × 20 tables been included to summarize the statistical data. The type of information provided should be useful to crystallographers and molecular modelers, and would be most useful when used in conjunction with an interactive display system. Thus, a visual check of a proposed model against the previously observed patterns for a particular pairwise interaction could be a useful modeling tool and an assessment of the statistical likelihood of proposed interaction angles by comparison with those in the atlas could provide an unbiased check of the reasonableness of a proposed side-chain packing mode.

**W.
Michael
Carson**

*
Macromolecular
Crystallography
University of
Alabama
Birmingham, AL
35294
USA
*

**Copyright © 1997 International Union of Crystallography**