Advances in Methods for Accurate Prediction of RNA–Small Molecule Binding Sites: From Isolated to AI-Integrated Strategies

Predicting RNA–small molecule binding sites is essential for developing RNA-targeted drugs. Identifying these sites experimentally is often costly, making computational methods essential for drug discovery. While traditional approaches rely on limited information, recent AI advancements allow the integration of diverse features, improving prediction accuracy. As methods for predicting RNA–small molecule binding sites continue to evolve, this review provides a timely overview of recent developments. It systematically traces the evolution from physics-based, isolated strategies to AI-integrated approaches, explains the fundamental principles behind different features, compares the tendencies of various features between binding and non-binding sites, evaluates the performance of approaches using different feature combinations on various test sets, and outlines remaining opportunities and challenges, offering guidance for researchers aiming for higher prediction accuracy.