Don't Retrain, Just Reuse: Recovering Dual-Target Molecules from Single-Target Diffusion Models

2026-05-25

AI Summary

Designing dual-target molecules that simultaneously satisfy high binding affinity for both targets, drug-likeness, and synthetic feasibility presents substantially greater challenges than single-target design. This work proposes the REUSE framework, which formulates dual-target molecular generation as a constrained multi-objective input-space optimization problem. Without modifying the parameters or denoising mechanism of a pretrained single-target diffusion model, REUSE efficiently generates dual-target molecules through hierarchical evolutionary search, paired conditional exploration, and a multi-stage structured selection strategy. Compared to existing approaches that require altering the diffusion process, REUSE achieves a 20.9 percentage point improvement in the Dual High Affinity metric while maintaining excellent molecular quality and diversity.

Abstract

Designing a single molecule that modulates two targets is a promising strategy for polypharmacology, but it remains substantially harder than standard single-target generation because one candidate must satisfy two binding requirements while preserving drug-likeness and synthesizability. Existing dual-target generative methods typically introduce dual-target capability by either retraining the generator or intervening in the diffusion process during sampling. The former can be costly and difficult to stabilize when dual-target supervision is sparse, while the latter may be sensitive to denoising-time target balancing and competing update directions. These limitations motivate a generator-preserving alternative that keeps the pretrained prior intact: can dual-target candidates instead be recovered from the input space of a frozen single-target diffusion model, without modifying its parameters or denoising dynamics? We formulate this task as a constrained multi-objective optimization problem and propose REUSE, a hierarchical evolutionary input-space search framework that combines pair-conditioned exploration with structured multi-stage selection to enforce dual-target affinity, chemical quality, and diversity. Experiments show that, compared with methods that modify the diffusion process, REUSE consistently improves dual-target affinity and balance, achieving a 20.9-percentage-point gain in Dual High Affinity over the strongest prior baseline while maintaining competitive molecular quality.

Problem

• dual-target molecules
• polypharmacology
• diffusion models
• molecular generation
• multi-objective optimization

Innovation

• dual-target molecule generation
• diffusion model reuse
• input-space optimization
• multi-objective molecular design
• evolutionary search