【数学与统计及交叉学科前沿论坛------高端学术讲座第166场】

报告题目： A cluster-based cell-type deconvolution of spatial transcriptomic data

报 告 人：王清悦 爱尔兰科克大学

报告时间：2025年10月15日周三16：00--17:00

报告地点：良乡校区数统楼401

报告摘要：Spatial transcriptomics is a method to locate and measure gene expression on tissue slices. However, most current spatial transcriptomics technologies perform it on individual spots containing a mixture of cells, thus lacking cell composition information. In this study, we developed DECLUST, a cluster-based methodology for cell-type deconvolution in spatial transcriptomics. DECLUST clusters the spots to preserve the spatial structure of tissues and identifies the cell-type-specific markers from single-cell RNA sequencing for accurate deconvolution. We applied this method to multiple simulated breast cancer datasets, as well as to two real-world datasets: ovarian cancer and mouse brain. The results showed that DECLUST performs well against other existing methods in both robustness and accuracy. In summary, DECLUST provides an effective and reliable model for efficiently identifying cell-type composition in spatial transcriptomics data.

报告人简介：Ms. Qingyue Wang is a fourth-year Ph.D. candidate at University College Cork (UCC), mainly supervised by Dr. Jian Huang (UCC), co-supervised by Prof. Yudi Pawitan, and Dr. Nghia Vu(Karolinska Institutet). She earned her undergraduate degree in Applied Statistics from Heilongjiang University of Science and Technology and her Master’s degree in Actuarial Science from UCC. Her doctoral research focuses on developing advanced statistical and computational models for spatial omics data. Specifically, her work aims to analyze and interpret the complex spatial organization of molecular features, such as RNA, proteins, and other biomolecules within biological tissues. By integrating biostatistical and bioinformatics methodologies, she designs robust and scalable frameworks for spatial omics analysis, contributing to a deeper understanding of tissue architecture and disease mechanisms.