MO-MEMES: A method for accelerating virtual screening using multi-objective Bayesian optimization

Abstract

The pursuit of potential inhibitors for novel targets has become a very important problem especially over the last 2 years with the world in the midst of the COVID-19 pandemic. This entails performing high throughput screening exercises on drug libraries to identify potential “hits”. These hits are identified using analysis of their physical properties like binding affinity to the target receptor, octanol-water partition coefficient (LogP) and more. However, drug libraries can be extremely large and it is infeasible to calculate and analyze the physical properties for each of those molecules within acceptable time and moreover, each molecule must possess a multitude of properties apart from just the binding affinity. To address this problem, in this study, we propose an extension to the Machine learning framework for Enhanced MolEcular Screening (MEMES) framework for multi-objective Bayesian optimization. This approach is capable of identifying over 90% of the most desirable molecules with respect to all required properties while explicitly calculating the values of each of those properties on only 6% of the entire drug library. This framework would provide an immense boost in identifying potential hits that possess all properties required for a drug molecules.

Pareto optimization to accelerate multi-objective virtual screening

“…Only in case 1 does the best scalarization acquisition function outperform the best Pareto acquisition function. Performance metrics using the acquisition function described by Mehta et al 42 (MO-MEMES) are included as an additional baseline in Tables S1-S3. † Pareto optimization acquisition functions consistently outperform MO-MEMES.…”

“…Model-guided multi-objective optimization has the potential to reduce the computational cost of a multi-objective virtual screen without sacrificing performance. Mehta et al 42 have previously applied Bayesian optimization to identify molecules that simultaneously optimize the docking score to Tau Tubulin Kinase 1, calculated octanol–water partition coefficient (clogP), 43 and synthetic accessibility score (SA_Score). 44 Their implemented acquisition function is a product of acquisition scores for individual objectives, 42 leading to the recovery of over 90% of the most desirable molecules after scoring only 6% of the virtual library.…”

“…Mehta et al 42 have previously applied Bayesian optimization to identify molecules that simultaneously optimize the docking score to Tau Tubulin Kinase 1, calculated octanol–water partition coefficient (clogP), 43 and synthetic accessibility score (SA_Score). 44 Their implemented acquisition function is a product of acquisition scores for individual objectives, 42 leading to the recovery of over 90% of the most desirable molecules after scoring only 6% of the virtual library. Multi-objective virtual screens involve multiple design choices primarily related to the acquisition strategy, which have not yet been compared in the context of virtual screening.…”