ODesign: A World Model for Biomolecular Interaction Design
Odin Zhang, Xujun Zhang, Haitao Lin, Cheng Tan, Qinghan Wang, Yuanle Mo, Qiantai Feng, Gang Du, Yuntao Yu, Zichang Jin, Ziyi You, Peicong Lin, Yijie Zhang, Yuyang Tao, Shicheng Chen, Jack Xiaoyu Chen, Chenqing Hua, Weibo Zhao, Runze Ma, Yunpeng Xia, Kejun Ying, Jun Li
2025-10-30
Summary
This paper introduces ODesign, a new artificial intelligence tool that can design biomolecules – like proteins, DNA, and small molecules – to interact with each other in specific ways. It's like being able to program molecules to bind to each other for a desired function.
What's the problem?
Currently, designing biomolecules to interact is really hard. Existing AI tools are usually good at designing *one* type of molecule, but not at controlling exactly *how* different molecules will interact. Scientists need a way to precisely design interactions between various biomolecules, not just focus on individual pieces.
What's the solution?
The researchers created ODesign, which uses a 'generative world model'. Basically, it learns the rules of how biomolecules interact and then can create new molecules that follow those rules. You can tell ODesign what you want a molecule to bind to (the 'target') and it will generate different options for molecules that will bind to it, giving scientists a lot of control over the design process. It works not just for proteins, but also for DNA and small molecules.
Why it matters?
This is a big step towards being able to program biology. If we can reliably design biomolecules to interact in specific ways, we can create new drugs, build new materials, and even engineer biological systems to perform new functions. ODesign is a move towards a single, versatile AI tool that can handle all sorts of biomolecular interactions, making this kind of design much more accessible.
Abstract
Biomolecular interactions underpin almost all biological processes, and their rational design is central to programming new biological functions. Generative AI models have emerged as powerful tools for molecular design, yet most remain specialized for individual molecular types and lack fine-grained control over interaction details. Here we present ODesign, an all-atom generative world model for all-to-all biomolecular interaction design. ODesign allows scientists to specify epitopes on arbitrary targets and generate diverse classes of binding partners with fine-grained control. Across entity-, token-, and atom-level benchmarks in the protein modality, ODesign demonstrates superior controllability and performance to modality-specific baselines. Extending beyond proteins, it generalizes to nucleic acid and small-molecule design, enabling interaction types such as protein-binding RNA/DNA and RNA/DNA-binding ligands that were previously inaccessible. By unifying multimodal biomolecular interactions within a single generative framework, ODesign moves toward a general-purpose molecular world model capable of programmable design. ODesign is available at https://odesign.lglab.ac.cn ,