Identifying Ligand Binding Sites of Proteins using Crystallographic Bfactors and Relative Pocket Sizes

One of the goals of proteomics is to predict the functions of all proteins. Many protein functions involve interactions with other biological molecules. Some of the most important interactions are those of proteins with small molecule ligands. Several experimental and computational approaches have been proposed previously to predict the locations of ligand binding regions of proteins. The aim of this thesis is to test a new approach to identify the ligand binding regions on the surface of X-ray crystal structures of proteins with unknown functions. The hypothesis that was tested was that X-ray crystallographic B-factors, also known as temperature factors, of atoms located on the surface of a protein X-ray crystal structure, in combination with information about the relative sizes of surface pockets, can help in identifying the functional regions of proteins. We used as our test set of protein structures a non-redundant dataset of 328 proteins for which X-ray crystal structures have been solved for both ligand-bound and unbound forms. The main steps involved identifying the surface pocket that, out of the ten largest surface pockets, contains the largest number of atoms with low B-factor values, and then checking to see if that pocket corresponds to the known ligand-binding pocket. If so, the crystallographic B-factor values could be used as a parameter to aid in predicting the ligand-binding regions of many uncharacterized proteins. For the majority of proteins in our study, that is, for 66% of the proteins in our working dataset, the proposed methodology was able to identify correctly the ligand-binding pockets in the protein crystal structures of the unliganded forms of the proteins.