Steven Bagley, Liping Wei,Carol Cheng, and Russ Altman
A protein site is a region of a three-dimensional protein structure with a distinguishing functional or structural role. Certain sites recur in different protein structures (for example catalytic sites, calcium binding sites, and some types of turns), but maintain critical shared features. To facilitate the analysis of such protein sites, we have developed a computer system for analyzing the spatial distributions of biochemical properties around a site. The system takes a set of similar sites and a set of control nonsites, and finds differences between them. Specifically, it compares distributions of the properties surrounding the sites with those surrounding the nonsites, and reports statistically significant differences. In this paper, we use our method to analyze the features in the active site of the serine protease enzymes. We compare the use of radial distributions (shells) with 3-D grids (blocks) in the analysis of the active site. We demonstrate three different strategies for focusing attention on significant findings, based on properties of interest, spatial volumes of interest, and on the level of statistical significance. Finally, we show that the program automatically identifies conserved sequential, secondary structural and biophysical features of the serine protease active site, using noncatalytic histidine residues as a control environment.