From Protein Structure to Function with Bioinformatics.pdf

176 N.J. Burgoyne and R.M. JacksonFig. 7.3 Schematic showing discrete flow theory. One Delaunay triangle acts as a sink for theflow. CASTp considers this a true pocketsurface spaces generates a triangulated surface. These triangular polyhedrons arepruned using the discrete flow theory (see Fig. 7.3). The polyhedrons that areenclosed by others are used to define the cavities.The general procedure adopted by all of these methods is to define a boundarybetween protein surface and pocket area. But an additional boundary, between thepocket area and open space, must also be defined. In some geometric definitionsthese boundaries are dependent on the structure of the protein and predicted sitevolumes show a tendency to increase with protein size. In energetically definedbinding sites (see below) sites have volumes roughly equivalent to ligand volumesirrespective of the overall size of the protein. This is more in keeping with the ideathat a ligand of a given size will occupy a similar sized binding pocket irrespectiveof protein size (Laurie and Jackson 2005).7.4.2.2 Energetically Defined Binding SitesRather than using a geometric approach, the energetic approach defines pockets asregions that are most likely to interact with other molecules (Laurie and Jackson2005). In Q-SiteFinder the proteins are first enclosed in grids where interactingprobes (modelled as a methyl (−CH 3) group) can be placed at all intersections thatdo not coincide with the protein atoms. The score of the probes are calculated asthe van der Waals interaction potential energy of that group in each location. Theprobes with a high enough affinity are retained and clustered. The defined clustersare ranked such that the pocket with the most favourable interaction energy (usuallythe largest) is assumed to be the most likely ligand-binding site. It was found that