A Genetic Algorithm for Navigating Synthesizable Molecular Spaces

2025-09-24

AI Summary

To address the challenges of ensuring synthetic accessibility and low search efficiency in molecular design, this paper introduces SynGA—a genetic algorithm operating directly in the synthetic route space. Its key contributions are: (1) an explicit synthetic accessibility constraint mechanism, enforced via customized crossover and mutation operators that guarantee all offspring molecules are synthesizable; (2) a machine learning–driven building-block filtering strategy that substantially improves search efficiency; and (3) SynGBO, an extension integrating SynGA into a Bayesian optimization framework for sample-efficient property optimization. Experiments demonstrate that SynGA achieves state-of-the-art performance in synthesizable analog search and multi-objective property optimization. It is lightweight, highly sample-efficient, and seamlessly compatible with existing chemical workflows—enabling plug-and-play, synthesis-aware molecular discovery.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Constraint Satisfaction and Optimization: Search
• Machine Learning: Optimization

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Economic aspects and information design of Web, GenAI, and cloud computing

Abstract

Inspired by the effectiveness of genetic algorithms and the importance of synthesizability in molecular design, we present SynGA, a simple genetic algorithm that operates directly over synthesis routes. Our method features custom crossover and mutation operators that explicitly constrain it to synthesizable molecular space. By modifying the fitness function, we demonstrate the effectiveness of SynGA on a variety of design tasks, including synthesizable analog search and sample-efficient property optimization, for both 2D and 3D objectives. Furthermore, by coupling SynGA with a machine learning-based filter that focuses the building block set, we boost SynGA to state-of-the-art performance. For property optimization, this manifests as a model-based variant SynGBO, which employs SynGA and block filtering in the inner loop of Bayesian optimization. Since SynGA is lightweight and enforces synthesizability by construction, our hope is that SynGA can not only serve as a strong standalone baseline but also as a versatile module that can be incorporated into larger synthesis-aware workflows in the future.

Problem

• Navigating synthesizable molecular spaces using genetic algorithms
• Constraining molecular design to synthesizable compounds via custom operators
• Optimizing molecular properties while ensuring chemical synthesizability

Innovation

• Genetic algorithm operating directly on synthesis routes
• Custom crossover and mutation for synthesizable molecular space
• Coupling with ML-based filter for building block optimization