It Takes Two to Tango: Directly Optimizing for Constrained Synthesizability in Generative Molecular Design

Jeff Guo, Philippe Schwaller

TL;DR: generate molecules with predicted synthetic pathways that incorporate a user-specified set of building blocks

Abstract: Constrained synthesizability is an unaddressed challenge in generative molecular design. In particular, designing molecules satisfying multi-parameter optimization objectives, while simultaneously being synthesizable \textit{and} enforcing the presence of specific commercial building blocks in the synthesis. This is practically important for molecule re-purposing, sustainability, and efficiency. In this work, we propose a novel reward function called **TANimoto Group Overlap (TANGO)**, which uses chemistry principles to transform a sparse reward function into a dense and learnable reward function -- crucial for reinforcement learning. TANGO can augment general-purpose molecular generative models to directly optimize for constrained synthesizability while simultaneously optimizing for other properties relevant to drug discovery using reinforcement learning. Our framework is general and addresses starting-material, intermediate, and divergent synthesis constraints. Contrary to many existing works in the field, we show that *incentivizing* a general-purpose model with RL is a productive approach to navigating challenging synthesizability optimization scenarios. We demonstrate this by showing that the trained models explicitly learn a desirable distribution. Our framework is the first *generative* approach to successfully address constrained synthesizability.

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: generative molecular design, synthesizability, drug discovery, language models, reinforcement learning

Submission Number: 2111