Biography:  Dr. Ruibin Xi (Chinese: 席瑞斌) is an associate professor at School of Mathematical Sciences, Peking University. He obtained his PhD from Washington University in St. Louis and received post doc training at Harvard Medical School. His main research interests include statistical analysis of big biological data, cancer genomics, network and graphical models, Bayesian analysis and high-dimensional statistics.

Title:  Single cell gene regulatory network analysis for mixed cell populations

Abstract: Gene regulatory network (GRN), representing the regulatory relationships between genes, is important for understanding the complex biological system. Single-cell RNA sequencing (scRNA-seq) technologies can now allow unveiling GRNs at the single cell level. However, most scRNA-seq data contain cells of more than one cell type and each cell’s cell type is unknown. The GRNs of different cell types are different. A common practice to infer GRNs in different cell types is to first cluster the cells and infer GRNs for every cluster separately. However, the clustering can be inaccurate especially for clusters that are close to each other, thus leading to inferior performance of GRN inference. Here, we propose to model scRNA-seq by the mixture multivariate Poisson Log-Normal (MPLN) distribution. The precision matrices of the MPLN are the cell-type-specific GRNs and maximizing its lasso-penalized likelihood can provide the estimates of the cell-type-specific GRNs without first clustering the cells. To avoid the intractable evaluation and optimization of the MPLN’s log-likelihood, we develop an uncentered variational inferential approach (called VMPLN) to estimate the precision matrices. Under an irrepresentability condition and a few other mild conditions, we establish the convergence rates and sign consistency of the estimator of the precision matrices. Comprehensive simulation analyses and scRNA-seq data analyses further reveal that VMPLN gives better performance for estimating the cell-type-specific GRNs than other methods.

Joint work with Junjie Tang, Changhu Wang, Feiyi Xiao.