主讲人：王涛 上海交通大学教授
时间：2026年10月29日15：00
地点：徐汇校区三号楼332室
举办单位：数理学院
主讲人介绍：王涛，上海交通大学教授。2007年获东南大学学士学位，2010年获华东师范大学硕士学位，2013年获香港浸会大学博士学位；2014-2015年在美国耶鲁大学生物统计系从事博士后研究。现任PLOS Computational Biology等国际学术期刊编委，并担任Statistical Learning and Data Science客座编辑。主要研究方向为数据科学、计算组学与生物统计学。研究成果发表于PNAS、JASA、Biometrika、JRSSB、Microbiome、Genome Biology、Nucleic Acids Research、Bioinformatics等期刊。
内容介绍：High-dimensional omics data pose significant challenges for statistical modeling and inference due to their complex dependence structures, heterogeneous distributions, and technical variation. In this talk, I will present three applications of generalized factor models to the analysis of microbiome and single-cell data. First, I will introduce a zero-inflated probabilistic PCA model for microbiome abundance data denoising, accounting for excess zeros, overdispersion, and dependence among microbial taxa. Second, I will discuss a factor-based approach to single-cell data integration that separates batch effects from biological variation while preserving phenotype-associated signals. Third, I will present a differential abundance analysis method that uses projection-based decomposition to jointly account for technical biases and unmeasured confounding. Together, these applications demonstrate the flexibility and utility of generalized factor models for analyzing complex omics data.
