报告人：Jiangning Song 教授 澳大利亚蒙纳士大学
主持人：耿新
报告时间：2026年3月26日（周四）上午10:00-11:00
报告地点：h片 九龙湖校区计算机楼513报告厅
报告摘要：The integration of machine learning into biomedicine has revolutionized data analysis, yet the "black box" nature of these models remains a barrier to clinical validation. This seminar demonstrates a paradigm shift—transforming AI from a predictive tool into a partner for biological discovery—by addressing the challenge of T-cell receptor (TCR) binding to antigens. Historically, the immense diversity of the immune repertoire has made modeling TCR–antigen recognition notoriously difficult. To solve this, we developed EPACT (Epitope-anchored Contrastive Transfer Learning), a deep-learning framework that establishes a shared semantic space for immune recognition. Crucially, by integrating structural data, we moved beyond binary prediction to mechanistic characterization, identifying critical contact residues at the TCR–pMHC interface. This explainable AI approach provides a transformative roadmap for deciphering TCR cross-reactivity, significantly accelerating the design of safer, more potent precision immunotherapies.
报告人简介：Jiangning Song is a Professor and Director of the AI-driven Bioinformatics and Biomedicine Unit in the Monash Biomedicine Discovery Institute (BDI), Faculty of Medicine, Nursing and Health Science, Monash University, Australia. He is an Associate Investigator of the ARC Centre of Excellence in Advanced Molecular Imaging, and an affiliated member of the Monash AI Institute and also an Honorary Professor of the University of Melbourne. He has published extensively in top-tier journals, e.g. Cell, Nature Biotechnology, Nature Methods, Lancet Oncology, Nature Machine Intelligence, Nature Computational Science, Nature Sustainability, Science Immunology, Science Advances, Cell Genomics, Nucleic Acids Research, Briefings in Bioinformatics, and Bioinformatics. He is an Associate Editor of several international journals including IEEE Journal of Biomedical and Health Informatics, BMC Bioinformatics, Genomics, Proteomics & Bioinformatics, and BMC Genomic Data. He is motivated to design, develop, and deploy cutting-edge data-driven methodologies and techniques to better understand and address a range of open and challenging scientific questions in biomedicine.
