Perturbed Flow Matching for Structure-Based Drug Design

Yankai Yu, Guikun Xu, Zhuyang Xie, Yan Yang, Yongquan Jiang, Xiaobo Zhou, Kang Li

Keywords: Structure-Based Drug Design, Generative models

TL;DR: propose a novel method named Perturbed Flow Matching (PFM) for Structure-Based Drug Design

Abstract: Generating 3D molecules that bind to specific protein targets via generative models has shown great promise in structure-based drug design. Recent diffusion-based approaches are constrained by marginals known in closed form and lack a design of the conditional probability path, which may hinder improved molecular generation. To address this issue, we introduce a flow-matching–based method—Perturbed Flow Matching (**PFM**)—which introduces a unique *perturbed conditional probability path design* that incorporates pocket binding site information and atom type–coordinate coupled information to enhance molecular generation performance. Experiments on the CrossDocked2020 dataset show that our model generates molecules with competitive 3D structures and state-of-the-art (SOTA) binding affinities toward protein targets, achieving an average score of -7.83. Validation on broader molecular datasets further confirms the consistent effectiveness of the proposed method. The code is available at https://anonymous.4open.science/r/pfm_ev_rv.

Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)

Submission Number: 23341