Beam Search for Automated Design and Scoring of Novel ROR Ligands with Machine Intelligence. Issue 35 (19th July 2021)

• Record Type: Journal Article
• Authors: Moret, Michael; Helmstädter, Moritz; Grisoni, Francesca; Schneider, Gisbert; Merk, Daniel
• Is Part Of: Angewandte Chemie international edition. Volume 60:Issue 35(2021)
• Page Start: 19477
• Page End: 19482
• Publication Date: 2021-07-19
• Subjects: de novo design -- deep learning -- drug discovery -- neural network -- nuclear receptor; Chemistry -- Periodicals; 540
• DOI: 10.1002/anie.202104405
• Languages: English
• ISSNs: 1433-7851

Abstract

Abstract: Chemical language models enable de novo drug design without the requirement for explicit molecular construction rules. While such models have been applied to generate novel compounds with desired bioactivity, the actual prioritization and selection of the most promising computational designs remains challenging. Herein, we leveraged the probabilities learnt by chemical language models with the beam search algorithm as a model‐intrinsic technique for automated molecule design and scoring. Prospective application of this method yielded novel inverse agonists of retinoic acid receptor‐related orphan receptors (RORs). Each design was synthesizable in three reaction steps and presented low‐micromolar to nanomolar potency towards RORγ. This model‐intrinsic sampling technique eliminates the strict need for external compound scoring functions, thereby further extending the applicability of generative artificial intelligence to data‐driven drug discovery. Abstract : The beam search algorithm was employed for automated molecule design and scoring from a chemical language model (CLM). Prospective application of this model‐intrinsic technique with a CLM trained on inverse RORγ agonists yielded novel ligands of this nuclear receptor with the intended bioactivity. Beam search sampling overcomes the need for external scoring methods and extends the applicability of machine learning‐driven molecular design.