Biomolecular Interface: Variational Implicit-Solvent Model, Numerical Methods, and Deep-Learning Prediction

报告题目：Biomolecular Interface: Variational Implicit-Solvent Model, Numerical Methods, and Deep-Learning Prediction

报告人： 周圣高  教授  上海交通大学

报告时间：2026年4月21日10:00-11:00

报告地点：红瓦楼726

报告内容简介：This talk will present a variational implicit-solvent model (VISM) for biomolecular solvation. Equipped with numerical methods, such as the level-set method and string method, the VISM is able to faithfully capture the interface between biomolecules and aqueous solvent, as well as dry-wet transition pathways of hydrophobic pockets. Based on such computed results, a multistate Brownian dynamics simulation algorithm and the related generalized Fokker–Planck theory are proposed to study the binding/unbinding kinetics of a ligand–receptor system. In addition, a deep learning algorithm based on local geometric and physio-chemical feature of the pocket interface is proposed to predict possible binding pockets on a protein surface. Our study could provide a useful tool for binding site identification and virtual screening of drugs.

报告人简介：周圣高，2012年6月于浙江大学取得博士学位，现为上海交通大学数学科学学院教授，国家级青年人才，已在PNAS、PRL、Math. Comput.、SIAM系列期刊发表研究工作，主持完成多项国家自然科学基金项目。

报告邀请人：黄雪玲