Small Molecule Fragment Hotspot Analyses to Drive Semi-Automated Design of Selective Molecules across a Protein Family

This project falls within the Chemical biology/biological chemistry, biological informatics, computational chemistry research areas, as outlined on the EPSRC website.
Companies/collaborators: SGC (Structural Genomics Consortium, Oxford)/XChem, CCDC (Cambridge Crystallographic Data Centre), Exscientia
Summary of the proposed project:
The past decade has seen an explosion in the availability of genomic and structural data for a great number of biomolecular disease targets. Rational drug discovery aims to create potent and selective compounds against these targets, with the aim of developing not only drugs, but also highly selective probes to investigate protein function. This has proven to be a challenging and expensive endeavour, and considerable efforts have been made to automate this process. Currently, automated methods such as fragment screening by X-ray crystallography output a wealth of structural data on low molecular weight molecules in complex with protein targets. Interpreting this data presents a complex problem, so computational tools that can distill it into suggestions for compound elaboration are needed. While considerable effort has been put into developing tools to predict binding affinity, less is known about the ways in which selectivity can be characterised and predicted.
Fragment hotspot analysis is a new and promising method that can highlight the specific interactions a protein makes with a compound to drive its binding. Building on work from the rotation project, the DPhil project will initially focus on using fragment hotspot analysis to describe and look for selectivity when designing compounds against a family of related targets. During the rotation, I looked at ways to combine hotspot information across an ensemble of X-ray structures of the same protein into a hotspot "ensemble map", then subtracted the ensemble maps for two different proteins to highlight differences in the binding pockets of the two proteins. To extend this work, it needs to be applied to a wider variety of proteins and protein families, and further work will cluster and extract important features from the ensemble maps. In addition, differences between ensemble maps for apo- , fragment-bound and ligand-bound structures will be investigated.
In the longer term, the project will attempt to develop novel methods or approaches to ranking fragment hits from X-ray crystallography screening campaigns, focussing on combining information from both computational and experimental methods.