Building an intelligent ecosystem,embarking ona new journey in scientific research

Access portal:https://ai4slab.pkusz.edu.cn

The School of AI for Science is committed to building a digital-intelligent scientific research platform featuring “inverse design” and “autonomous discovery capabilities.”It has developed the world’s first digital-intelligent life science research platform — AI4S LAB. Centered on the core concept of “AI-driven, dry–wet closed loop, full-chain digital intelligence,” AI4S LAB deeply integrates the four key elements of AI for Science (AI4S): “computing power, models, data, and experiments” to create a full-chain,closed-loop cloud-based research environment covering experimental design, intelligent execution, data analysis, and autonomous optimization.Supported byinterdisciplinary teams and self-developed technologies, the platform is dedicated to driving a paradigm shift in scientific research, providing users with “end-to-end” research services, and cultivating the next generation of interdisciplinary AI4S research talents.

Core Architecture: Integrated Coordination of Three Major Systems

Large Model Platform

Equipped with vertically specialized large models for life science experimental design, the platform supports intelligent generation of experimental schemes, autonomous optimization of experimental conditions, protein design, and structure prediction. Through continuous iterative training on multimodal data, it enables intelligent research assistance characterized by “description as design.”

Agent Platform

Afull-process multi-agent collaborative system is constructedfor experimental design, execution, and analysis, enabling dynamic task decomposition, automated resource scheduling, autonomous monitoring of experimental processes, and exception handling.This forms a full-chain digital-intelligent closed-loop workflow of “perception – decision-making – execution – learning.”

Experimental Platform

The platformintegratesahigh-throughput automated experimental systems. Through intelligent scheduling of robotic arms, liquid handling workstations,microbialclonescreeningsystems, biochemical analysis instruments, incubators, and centrifuges, and other automated equipment,it enables fully unmanned operation from sample preprocessing to detection and analysis. By submitting experimental schemes via the cloud, the system can automatically allocatesresources and executes complex experimental workflows,including DNA component assembly, cell factory construction, enzyme directed evolution, high-throughput expression screening, automated protein purification, high-throughput chemical screening, automated mass spectrometry preprocessing, and NGS library construction. The platform supports parallel processing at the scale of thousands of samples per week, and provides visual access to experimental progress, equipment status, and real-time data. Upon completion, experimental results are automatically returned to the cloud for timely analysis and archiving. The platform provides stable,efficient,large-scale experimental capabilities for fields such as synthetic biology, enzyme engineering, and drug screening.

Empowered Research Directions

•Synthetic biology: gene circuit design, cell factory construction, enzyme molecular evolution

•Protein engineering: structure prediction, directed evolution, high-throughput expression and screening

•Drug development: compound screening, target validation, pharmacodynamic analysis

•Omics and molecular detection: NGS library construction, high-throughput qPCR system construction, genotyping and targeted capture

•Life science education: interdisciplinary courses, virtual simulation experiments, scientific research training platforms

Through the deep coupling of “intelligent hardware,AI central system,anddata engine,”AI4S LABestablishes an evolvable and scalable cloud-based research infrastructure, aiming to become a globally leading digital-intelligent life science innovation engine. We sincerely invite faculty and students to actively participate in using the platform, jointly promote its iterative improvement, and accelerate the transformation from scientific discovery to industrial application.

Note:The platform is currentlyunder continuous development. The multi-agent collaborative system is undergoing ongoing iteration and enhancement, and some advanced features (e.g., multi-task concurrency, 7×24-hour operation) are under development. Feedback and suggestions from faculty and students are warmly welcomed as we work together to jointly build a new digital-intelligent scientific research ecosystem.