The 4th Bio-OS AI Open Source Competition of Guangzhou National Laboratory

With the deep evolution of global science and technology, life science research has entered a new stage of intelligence-driven, autonomous discovery. Big data and artificial intelligence(AI), as core engines, are upgrading from "supportive assistance" to "innovation leadership" in key fields such as drug and vaccine development,digital organ construction, multi-omics population cohort studies,and precise prevention and control of epidemics.The Biomedical Big Data Operating System(BioOS), independently developed by the Greater Bay Area Bioinformatics Center of Guangzhou National Laboratory, has evolved from a basic data foundation into an intelligent hub supporting the entire-chain innovation of Al for Science.It efficiently integrates cutting-edge IT and Al technologies to enable intelligent processing of multi-omics data and closed-loop scientific research.

Building on the success of the first three national open-source competitions—which gathered top global intellectual resources and promoted the implementation of multiple key technologiesthe 4th BioOS AI OpenSource Competition is specially held to further serve the laboratory's major strategic directions and fully activate the innovative potential of the opensource ecosystem.Focusing on cuttingedge innovation and industrial application at the intersection of Al and biomedicine, the competition is committed to building a globally leading intelligent innovation platform,gathering global wisdom to tackle key scientific challenges,and promoting theefficient transformation of scientific research results from the laboratory to clinical application and industrialization.

The "Bio-OS AI Open Source Competition" has been officially included in the competition system of South China University of Technology (SCUT) and recognized as a discipline and technology competition. Its award results serve as a bonus point basis for funding or scholarships.

Emerging infectious diseases pose a severe threat to people's lives and health, and constitute a major challenge to social and economic stability and national security. Neutralizing antibodies, due to their high efficiency and specificity, are one of the key intervention measures for the prevention and control of major epidemics. However, traditional antibody research and development technologies have shortcomings such as low throughput, high cost, and low efficiency, making it difficult for the development speed to keep up with the rapid mutation and evolution of viruses.

In recent years, breakthroughs in artificial intelligence (AI) technology have been reshaping the paradigm of antibody drug research and development. For example, MAGE (Cell, 2025) based on protein language models and RFdiffusion (Nature, 2025) based on diffusion models have been successively developed, providing new pathways for antibody design and optimization. Most recently, GB-0895, the world's first fully AI-designed antibody drug, has officially entered Phase III clinical trials. These achievements mark that using computational methods to design functional antibodies has moved from theory to reality, opening a new era in antibody drug research and development.