Assessing multi-objective optimization of molecules with genetic algorithms against relevant baselines

Nathanael Kusanda, Gary Tom, Riley Hickman, AkshatKumar Nigam, Kjell Jorner, Alan Aspuru-Guzik

Keywords: machine learning, inverse design, genetic algorithms, multi-objective optimisation, de novo design, computational chemistry, lipophilicity, docking

TL;DR: Multi-objective optimization of molecules on drug-design-related tasks with the JANUS genetic algorithm showed better performance of the Chimera and Hypervolume algorithms compared to baselines.

Abstract: Chemical design is often complex, requiring the optimal trade-off between several competing objectives. Multi-objective optimization algorithms are designed to optimally balance multiple objectives, but many chemical design approaches use the naïve weighted sum method, which is not guaranteed to give desired solutions. Here, we rigorously assess the performance of genetic algorithms for inverse molecular design using more advanced multi-objective methods. Chimera and Hypervolume are assessed against relevant baselines for the optimization of molecules with high logP and high QED score. As a more realistic task, we also simulate a drug design campaign, optimizing for synthetically accessible molecules which bind to the 1OYT protein. Additionally, we include a three-objective task of optimizing logP, QED and SAS to investigate scalability to more than two objectives. We show that both methods achieve better formal optimality than the baselines and generate molecules closer to a user-specified Utopian point in property space, mimicking typical materials design objectives.

Paper Track: Papers

Submission Category: AI-Guided Design