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

TL;DR

This paper introduces TANGO, a novel reward function for reinforcement learning that enables generative molecular models to optimize for constrained synthesizability alongside other drug discovery properties, addressing a key challenge in molecular design.

Contribution

The work presents TANGO, the first reward function designed to directly optimize constrained synthesizability in molecular generative models, improving their practical applicability.

Findings

TANGO effectively transforms sparse rewards into dense, learnable signals.

Models trained with TANGO explicitly learn desirable molecular distributions.

The framework handles various synthesis constraints, enhancing molecule re-purposing and sustainability.

Abstract

Constrained synthesizability is an unaddressed challenge in generative molecular design. In particular, designing molecules satisfying multi-parameter optimization objectives, while simultaneously being synthesizable 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 most existing works in the field, we show that incentivizing a general-purpose (without any inductive biases) model is a productive approach to navigating challenging optimization scenarios. We demonstrate this by showing that the trained models explicitly learn a desirable distribution. Our framework is the first generative approach to tackle constrained synthesizability.