Scalable and Cost-Efficient de Novo Template-Based Molecular Generation

TL;DR

This paper introduces a scalable, cost-effective method for template-based molecular generation using recursive guidance and dynamic libraries, achieving state-of-the-art results in drug design applications.

Contribution

It presents Recursive Cost Guidance and Dynamic Library mechanisms to improve efficiency and scalability in template-based molecular generation.

Findings

Significantly reduces synthesis cost and improves molecular diversity.

Enhances performance with small building block libraries.

Achieves state-of-the-art results in molecular generation benchmarks.

Abstract

Template-based molecular generation offers a promising avenue for drug design by ensuring generated compounds are synthetically accessible through predefined reaction templates and building blocks. In this work, we tackle three core challenges in template-based GFlowNets: (1) minimizing synthesis cost, (2) scaling to large building block libraries, and (3) effectively utilizing small fragment sets. We propose Recursive Cost Guidance, a backward policy framework that employs auxiliary machine learning models to approximate synthesis cost and viability. This guidance steers generation toward low-cost synthesis pathways, significantly enhancing cost-efficiency, molecular diversity, and quality, especially when paired with an Exploitation Penalty that balances the trade-off between exploration and exploitation. To enhance performance in smaller building block libraries, we develop a Dynamic Library mechanism that reuses intermediate high-reward states to construct full synthesis trees. Our approach establishes state-of-the-art results in template-based molecular generation.