Article · AI translation
IT Home reported on November 8 that teams led by Chen Haibo and Wang Zhaoguo of the IPADS Laboratory at Shanghai Jiao Tong University's School of Computer Science collaborated with teams led by Fan Chunhai and Wang Fei of its School of Chemistry and Chemical Engineering and Frontiers Science Center for Transformative Molecules. The research was conducted through the DNA Storage Research Center at the Zhangjiang Institute for Advanced Study. The findings were published in Nature and prominently featured as the lead story on the journal's homepage.
The researchers built AutoDNA (Siyuan No. 1), an AI-native scientific laboratory system that performs molecular biology experiments autonomously. By jointly designing the model, experiments, and instruments, they addressed challenges that existing autonomous AI laboratories face when expanding into molecular biology experiments, including organizing complex procedures, coordinating multiple types of instruments, and ensuring experimental timing, environmental stability, and operational precision. This enables autonomous experiments to cover complex biomolecular procedures requiring micrometer-scale positioning and microliter-scale liquid handling.
According to the introduction, existing autonomous AI laboratories typically rely on experts to configure instruments, prescribe experimental procedures, and set optimization rules in advance, then incorporate AI into individual steps such as literature searches, protocol generation, or data analysis. Although this “add-on” architecture can improve the efficiency of individual steps, it struggles to accommodate the complex procedures and diverse instruments involved in biomolecular experiments, as well as their strict requirements for real-time execution, environmental stability, and operational precision. To address this, AutoDNA uses an AI-native architecture based on the joint design of the model, experiments, and instruments.
To address the complexity of experimental procedures, AutoDNA combines an understanding of processes suited to complex biomolecular operations with an experiment-focused multi-agent architecture. The system breaks complex experimental tasks into steps including literature searches, reagent selection, protocol design, code generation, hardware execution, results analysis, and iterative optimization. It coordinates multiple specialized agents according to experimental progress, allowing the model to manage the entire “design–experiment–optimization” process and continually adjust subsequent protocols based on experimental results.
To address the variety of instruments and the difficulty of coordinating across devices, AutoDNA distills the capabilities of different instruments into “atomic services” that the model can understand and invoke. Natural-language descriptions and a unified code interface tell the model what each device can do and how to call it. This abstraction masks underlying differences between instrument models, allowing the model to combine device capabilities as experiments require and to generate, check, and revise code that controls multiple devices, translating experimental designs in the digital world into instrument operations in the physical world.
To meet strict requirements for real-time operation, stability, and operational precision, AutoDNA combines high-precision automated hardware with a system for experimental execution and scheduling. The hardware provides micrometer-scale positioning and microliter-scale liquid handling, while the system checks programs, tracks instrument status, and schedules resources so that multiple experimental tasks run in an orderly manner while meeting timing and operational constraints. During the platform's development, the team also integrated a control terminal based on the OpenHarmony operating system and demonstrated validation of its applications in nucleic acid testing, enzymatic DNA synthesis, and cancer detection.
IT Home provides a link to the paper:
https://www.nature.com/articles/s41586-026-11228-6
Advertising notice: External links in this article (including but not limited to hyperlinks, QR codes, and access codes) are provided to convey more information and save time in selecting information. The results are for reference only. All IT Home articles containing external links include this notice.