Molecule Generation for Drug Design: a Graph Learning Perspective
Nianzu Yang, Huaijin Wu, Kaipeng Zeng, Yang Li, Junchi Yan
TL;DR
This paper addresses de novo molecule design using graph-based generative models by modeling molecules as graphs $G=(A,X)$ and learning to sample from the distribution $p(G)$. It categorizes methods into all-at-once, fragment-based, and node-by-node generation, surveys diffusion-based and RL/MCMC approaches, and compiles public datasets and evaluation metrics. It highlights diffusion-based graph generation (e.g., DiGress, Wave-GD) and reconstruction-based and search-based strategies, and discusses macro-molecule design, 3D drug discovery, and structure-based design as future directions. The work clarifies the landscape, guides reproducibility, and informs practical drug design with graph-based methods.
Abstract
Machine learning, particularly graph learning, is gaining increasing recognition for its transformative impact across various fields. One such promising application is in the realm of molecule design and discovery, notably within the pharmaceutical industry. Our survey offers a comprehensive overview of state-of-the-art methods in molecule design, particularly focusing on \emph{de novo} drug design, which incorporates (deep) graph learning techniques. We categorize these methods into three distinct groups: \emph{i)} \emph{all-at-once}, \emph{ii)} \emph{fragment-based}, and \emph{iii)} \emph{node-by-node}. Additionally, we introduce some key public datasets and outline the commonly used evaluation metrics for both the generation and optimization of molecules. In the end, we discuss the existing challenges in this field and suggest potential directions for future research.