Sample Efficient Generative Optimization for Molecular Design

2026-07-14

AI Summary

Molecular optimization faces the dual challenge of an astronomically large chemical space and the high cost of high-fidelity property evaluations, necessitating improved sample efficiency. This work proposes SEGO, a novel framework that tightly integrates an adaptive conditional generative model with Bayesian optimization. By jointly updating the surrogate model and the generator within a closed-loop system—leveraging probabilistic surrogate modeling, acquisition function optimization, and online learning—SEGO substantially reduces reliance on expensive oracle evaluations. On the PMO benchmark, SEGO achieves state-of-the-art performance using only one-tenth the number of oracle calls required by competing methods. In multi-parameter docking tasks, it successfully identifies ten valid molecules with approximately half the evaluation budget, demonstrating exceptional sample efficiency and practical utility.

Abstract

Molecular optimization in drug discovery, materials design, and catalysis requires searching vast chemical spaces under tight evaluation budgets, since high-fidelity oracles and experimental measurements are costly. The practical impact of an optimization method therefore hinges on its sample efficiency: how few evaluations it needs to find strong candidates. We introduce Sample Efficient Generative Optimization (SEGO), a framework for Bayesian optimization on adaptively generated molecules. In SEGO, a probabilistic surrogate model forms a hypothesis about where hits lie in chemical space, a generative model is steered to propose candidates in that region, the most promising candidate is selected via an acquisition function, and the resulting oracle call is used both to sharpen the surrogate and to anchor the generator in real reward. SEGO attains state-of-the-art performance on the practical molecular optimization (PMO) benchmark using only one tenth of the oracle calls consumed by other methods, and on a multiparameter docking task it reaches ten hits in roughly half the oracle calls of existing approaches. These gains move molecular optimization closer to campaigns driven by direct experimental feedback.

Problem

• molecular optimization
• sample efficiency
• chemical space
• evaluation budget
• drug discovery

Innovation

• Sample Efficient Generative Optimization
• Bayesian optimization
• generative molecular design
• oracle-efficient optimization
• chemical space exploration