2015年3月23日，美国密西根大学李洪东博士前来中心作学术报告，作题为“Towards splice isoform-level functional genomics through multi-source genomic data inte-gration.”的讲座，介绍其博士后期间的工作与研究成果。
讲座简介：Functional annotation is mainly conducted at the gene level without differentiating the potentially different or even opposing functions of isoforms generated by the same gene through alternative splicing. Gene-level functional networks have been shown to be promising in identifying disease-associated genes but have limited accuracy since isoforms are not considered. We developed multiple-instance learning (MIL), a computational approach to predict isoform-level functional network that is able to reveal functional diversity of isoforms of the same gene. Using MIL, we built the genome-wide isoform network for the mouse. Based on isoform network, we defined Highest Connected Isoforms (HCIs) and examined their expression signals at both transcript and protein level. Finally, advantages, limitations and potential ways for improvement of isoform networks are discussed.