地 点: 理科大楼A1514室
题 目：Statistical Methodologies for Multi-condition Genomic and Spatial Transcriptomic Analyses
报告人：Qunhua Li 宾夕法尼亚州立大学副教授
In this talk, I will present two innovative statistical methodologies that address important challenges in genomic and spatial transcriptomic analyses.
Joint analyses of genomic datasets obtained under various conditions are crucial for understanding tissue-specificity and cell differentiation mechanisms, but they pose computational challenges. To tackle this, I will introduce CLIMB (Composite LIkelihood eMpirical Bayes), a statistical methodology that identifies condition-specific patterns in genomic data. CLIMB enables clustering of genomic features with similar condition-specific patterns and identification of features involved in cell fate commitment. Through the analysis of ChIP-seq, RNA-seq, and DNase-seq datasets, we demonstrate that CLIMB improves statistical precision compared to existing methods while capturing biologically relevant and interpretable clusters.
Spatial transcriptomics (ST) enables the profiling of gene expression in intact tissues. However, ST data obtained at each spatial location often represents the expression of multiple cell types, hindering the identification of cell-type-specific transcriptional variations across spatial contexts. Existing cell-type deconvolution methods in ST rely on single-cell transcriptomic references, which can be limited in availability and completeness. To overcome these limitations, we present RETROFIT, a reference-free Bayesian hierarchical method that accurately deconvolves cell types underlying each spatial location, independent of single-cell references. Through synthetic and real ST datasets, we demonstrate that RETROFIT outperforms existing methods in estimating cell-type composition and reconstructing gene expression. Application of RETROFIT to human intestinal development ST data reveals spatiotemporal patterns of cellular composition and transcriptional specificity.
Qunhua Li is an Associate Professor of Statistics at Penn State and a key member of the Bioinformatics and Genomics program. She earned her Ph.D. in Statistics from the University of Washington in 2008 and joined Penn State as an Assistant Professor in 2011. Dr. Li's research focuses on developing statistical methods for analyzing complex biological data. She combines statistics and biology to uncover patterns in large datasets. Specifically, she works on creating models and using machine learning techniques to identify meaningful structures in high-throughput genomic and proteomic data.