CAFCW117 A Scalable, Validated Platform for Generative Lead Optimization of De Novo Molecules: Case Study in Discovery of Potent, Selective Aurora Kinase Inhibitors with Favorable Secondary Pharmacology
Licensed under
Category
Published on
Abstract
De Novo design of therapeutic agents is currently a slow, expensive process generally relying on a large high throughput screen and several follow up cycles of iterative design to enhance the potency, eliminate safety liabilities, and enable favorable pharmacokinetic behavior. Computer aided drug discovery (CADD) has significantly aided this though structure and ligand-based design techniques, however current application often focuses on optimizing a single molecular property at a time, leading to long design cycle times. Recent advances in generative models for chemistry, including variational autoencoders (VAEs) and generative adversarial networks (GANs) have enabled a continuous representation of chemical space, allowing for application of numerical optimization techniques to this multi-factorial problem. As a proof of concept exercise, optimization of generative network across a diverse range of pharmacological properties was performed utilizing a high-performance scalable framework. Specifically, the algorithm was challenged to design a potent inhibitor of Aurora Kinase B, an enzyme involved with cell division and drug target in at least 12 ongoing and completed oncology trials. For a more realistic lead optimization challenge, the algorithm was also tasked with selectivity against the related Aurora Kinase A while simultaneously maintaining favorable secondary pharmacology properties (i.e. safety and pharmacokinetics). Running on a multi-node cluster, the framework was able to generate, evaluate and rank 3 million compounds, including >200 de novo molecules with predicted properties meeting the acceptability criteria. Experimental validation was performed for approximately 100 of the de novo molecules through chemical synthesis and in vitro testing.
Cite this work
Researchers should cite this work as follows: