Graph method improves molecule design for multiple properties
Graph Forward Distribution Matching for Molecular Inverse Design
Machine Learning
Summary
Designing molecules with specific qualities is hard because changing some features can make the molecule invalid. The authors introduce GraphFDM, a new way to teach a computer model to build molecules by focusing on the growing process rather than retracing steps backwards. This approach helps the model learn better molecular designs while keeping them chemically valid and improves accuracy on multiple properties. It also works well when asked to create molecules with unusual combinations of traits.
What this means in practice
- •For chemical engineering teams: Design polymer and small-molecule candidates with precise multi-property control and high chemical validity for material development.
- •For pharmaceutical developers: Generate novel drug-like molecules optimized for multiple desired properties within valid chemical structures.
Authors
Yihan Zhu, Yuhan Liu, Brett Savoie, Tengfei Luo, Meng Jiang
Abstract
Achieving precise control over multiple properties without sacrificing chemical validity remains a central challenge in molecular inverse design. Existing reinforcement learning (RL) methods fine-tune graph diffusion models by treating **reverse** sampling as a sequential policy, using a single terminal reward to optimize hundreds of coupled decisions. They often suffer from instability, validity collapse, and limited property gains. We introduce GraphFDM (Graph Forward Distribution Matching), a new online RL paradigm for graph diffusion that performs optimization through the **forward** process. GraphFDM uses valid generations to define a reward-tilted target distribution jointly optimized over graph size and molecular structure for each property condition, incorporating reinforcement signals into supervised learning without storing reverse trajectories. We derive the unique optimal target, prove a condition-wise improvement guarantee, and show that the fixed graph-size prior of standard graph diffusion leaves an irreducible matching gap. In multi-conditional polymer and small-molecule generation, GraphFDM achieves the lowest MAE on every target property, with reductions of up to 53.0\% relative to the strongest baselines and chemical validity above 0.99. It further generalizes to out-of-distribution property combinations.