Fragment elaboration – i.e., the process of growing a weak fragment hit into a potent lead compound – is a complex task impacted by a multitude of structural, physicochemical and medicinal objectives and constraints. Machine learning and artificial intelligence are optimally suited for such complex optimization tasks by learning from historic data, but may benefit from the input of human experts in guiding their search through chemical space. Lucian will be developing AI-driven approaches for the structure-based design of potent and selective lead compounds, and explore the utility of these methods in an interactive design context. This is a challenging application of AI that is right in Astex’s field of expertise: developing fragment hits into potent leads, while making efficient use of Astex’s vast pool of X-ray structures of ligand-protein complexes.
Chan et al., “A multilevel generative framework with hierarchical self-contrasting for bias control and transparency in structure-based ligand design”; Nat. Mach. Intell. 2022 https://doi.org/10.1038/s42256-022-00564-7