Classification of likely functional state for ligand binding sites identified from fragment screening

Abstract

Fragment screening data from 37 experiments, and 1,309 protein structures binding to 1,601 ligands were analysed. A new method to group ligands by binding sites was developed and sites clustered according to profiles of relative solvent accessibility. This identified 293 unique ligand binding sites, which are grouped into four clusters (C1-4). C1 presents larger, more buried, conserved, and missense-depleted sites, and is enriched in known functional sites. C4 comprises smaller highly accessible, more divergent, missense-enriched sites, and is depleted in functional sites. A site in C1 is 28 times more likely to be functional than a site in C4. 17 novel sites in 13 proteins are identified as likely to be functionally important with examples from human tenascin and 5-aminolevulinate synthase discussed in more detail. An artificial neural network, and a K-nearest neighbours model are presented to predict cluster labels for new ligand binding sites with an accuracy of 96% and 100%, respectively so allowing functional classification of sites for proteins not in this set. Our findings will be of interest to those studying protein-ligand interactions and developing new drugs or function modulators.