Impacting Drug Discovery Projects with Large-Scale Enumerations, Machine Learning Strategies, and Free-Energy Predictions
Jennifer L. Knight, Karl Leswing, Pieter H. Bos, Lingle Wang
Abstract
Binding free energy predictions of small molecules are becoming increasingly impactful in drug discovery campaigns. By efficiently and reliably identifying potent and selective chemical matter, project teams can invest time and financial resources on the most promising design ideas and effectively explore the most relevant chemical space. A large pool of design ideas along with sufficient accuracy and throughput are all required for binding free energy predictions to accelerate drug discovery programs. In this chapter, we discuss how free energy calculations are being used at scale in our internal drug discovery teams and collaborations to positively impact project performance and we highlight our experience with and best practices that are emerging for active learning FEP (AL-FEP), an approach that combines large-scale library enumerations, machine learning strategies, and free energy calculations to efficiently explore diverse chemical space.