Optimized Drug Design using Multi-Objective Evolutionary Algorithms with SELFIES
Tomoya Hömberg, Sanaz Mostaghim, Satoru Hiwa, Tomoyuki Hiroyasu
TL;DR
The paper tackles the challenge of navigating the enormous chemical space for drug design by integrating SELFIES with multi-objective evolutionary algorithms (NSGA-II, NSGA-III, MOEA/D) to optimize drug-likeness and synthesizability. It demonstrates that this combination can efficiently converge to diverse Pareto-optimal candidates while enabling discovery of novel compounds not present in public databases, evaluated via QED, SA, and GuacaMol MPO tasks. The study provides a comparative analysis of MOEAs in this setting, highlighting NSGA-II’s superior diversity and novelty, with NSGA-III and MOEA/D offering advantages under certain population sizes. The approach offers a practical, interpretable, and computationally efficient CADD workflow with potential for automated downstream evaluation and refinement of candidate molecules.
Abstract
Computer aided drug design is a promising approach to reduce the tremendous costs, i.e. time and resources, for developing new medicinal drugs. It finds application in aiding the traversal of the vast chemical space of potentially useful compounds. In this paper, we deploy multi-objective evolutionary algorithms, namely NSGA-II, NSGA-III, and MOEA/D, for this purpose. At the same time, we used the SELFIES string representation method. In addition to the QED and SA score, we optimize compounds using the GuacaMol benchmark multi-objective task sets. Our results indicate that all three algorithms show converging behavior and successfully optimize the defined criteria whilst differing mainly in the number of potential solutions found. We observe that novel and promising candidates for synthesis are discovered among obtained compounds in the Pareto-sets.