Fitness aligned structural modeling enables scalable virtual screening with AuroBind
Zhongyue Zhang, Jiahua Rao, Jie Zhong, Weiqiang Bai, Dongxue Wang, Shaobo Ning, Lifeng Qiao, Sheng Xu, Runze Ma, Will Hua, Jack Xiaoyu Chen, Odin Zhang, Wei Lu, Hanyi Feng, He Yang, Xinchao Shi, Rui Li, Wanli Ouyang, Xinzhu Ma, Jiahao Wang, Jixian Zhang, Jia Duan
2025-08-05
Summary
This paper talks about AuroBind, a powerful virtual screening system that uses detailed atomic-level models to predict how small molecules bind to proteins, which helps in finding new drugs quickly and accurately.
What's the problem?
The problem is that most human proteins are not targeted by existing drugs, and current methods for predicting how molecules bind to proteins lack the precision needed to find the best drug candidates efficiently.
What's the solution?
AuroBind solves this by fine-tuning custom atomic-level structural models with a huge amount of chemical and genomic data, using advanced techniques like preference optimization and self-learning from high-confidence examples, allowing it to predict both the structure of molecule-protein binding and how well they fit together.
Why it matters?
This matters because AuroBind can screen massive libraries of compounds much faster and more accurately than older methods, helping scientists discover effective drugs for diseases more quickly, including for hard-to-target proteins involved in cancer and other illnesses.
Abstract
AuroBind is a scalable virtual screening framework that fine-tunes atomic-level structural models to predict ligand-bound structures and binding fitness, achieving high hit rates in prospective screens across disease-relevant targets.