Reimagining Target-Aware Molecular Generation through Retrieval-Enhanced Aligned Diffusion

TL;DR

This paper introduces READ, a retrieval-enhanced aligned diffusion model that improves target-aware molecular generation by combining retrieval-augmented generation with diffusion models, leading to more accurate and diverse drug-like molecules.

Contribution

The paper presents the first integration of retrieval-augmented generation with SE(3)-equivariant diffusion models for molecular design, enhancing geometric fit and chemical validity.

Findings

READ surpasses state-of-the-art models in CBGBench

It produces valid, diverse, and shape-complementary ligands

Retrieval and diffusion are effectively co-optimized for drug design

Abstract

Breakthroughs in high-accuracy protein structure prediction, such as AlphaFold, have established receptor-based molecule design as a critical driver for rapid early-phase drug discovery. However, most approaches still struggle to balance pocket-specific geometric fit with strict valence and synthetic constraints. To resolve this trade-off, a Retrieval-Enhanced Aligned Diffusion termed READ is introduced, which is the first to merge molecular Retrieval-Augmented Generation with an SE(3)-equivariant diffusion model. Specifically, a contrastively pre-trained encoder aligns atom-level representations during training, then retrieves graph embeddings of pocket-matched scaffolds to guide each reverse-diffusion step at inference. This single mechanism can inject real-world chemical priors exactly where needed, producing valid, diverse, and shape-complementary ligands. Experimental results demonstrate that READ can achieve very competitive performance in CBGBench, surpassing state-of-the-art generative models and even native ligands. That suggests retrieval and diffusion can be co-optimized for faster, more reliable structure-based drug design.