12月18日 宋江宁教授学术报告

发布时间:2025-12-18

报告题目：From Machine Learning to Biological Insight:Biologically-informed Al for Decoding Immune Recognition

主讲人：宋江宁教授， Monash University 

报告时间：2025年12月18日（周四），10:00-12:00

报告地点：中山大学东校园超算中心一楼会议室

主持人：杨跃东教授

摘要：

Machine leamning has revolutionized pattern extraction from complex high-dimensional data in characterizationsciences, yet its "black box" nature hinders reliable validation and prediction. This talk focuses on tackling the coreimmunology challenge of predicting T cell receptor (TCR)-antigen binding-essential for adaptive immunity buttough to model due to TCR diversity and cross-reactivity-to tum machine leaming into a tool for interpretablebiological insights. We developed EPACT, a novel deep-learning framework for paired human CD8t TCR data. Its"divide-and-conquer" strategy generates TCR-pMHC co-embeddings, building a shared immune recognitionsemantic space. Fine-tuned on structural data, EPACT moves beyond prediction to mechanistic characterization.quantifying interchain distances and identifying key TCR-pMHC interface residues. This enabled decoding thestructural basis of TCR cross-reactivity with tumour-associated antigens, demonstrating how explainable AIadvances biological discovery and accelerates safer, more effective precision immunotherapies.

主讲人简介：

Jiangning Song is a Professor at Monash University, Australia, and leads the AI-driven Bioinformatics andBiomedicine Laboratory in its Biomedicine Discovery Institute (BDI). He also serves as an Associate Investigator atthe ARC Centre of Excellence in Advanced Molecular Imaging and an affiliate of the Monash AI Institute. He has acareer total of 355 publications in intemnational joumals with ~15000 citations, H-index of 65 and i10-index of 248 inGoogle Scholar, publishing extensively in top joumals like Cell, Nature series, Lancet series, and Science series.Currently an Associate Editor for multiple international joumals, his research focuses on developing cutting-edgedata-driven methodologies to address key challenges in biomedicine.