Statistics & Data Science Colloquium
Title: Spatially variable gene detection in integrated single-cell and spatial transcriptomics data
Abstract: The cellular architecture of tissues, i.e., the relationship between cells and their relative locations within the tissue, is crucial to understand cell-cell communications and disease pathology. Although it remains a challenge to comprehensively map single-cells in tissues, the emerging spatial transcriptomics technologies have allowed researchers to resolve spatial gene expression profiles of diverse tissues. To detect genes with spatial patterns, we propose to utilize a leverage-based variable screening method in integrated single-cell and spatial transcriptomics data, which provides higher sensitivity and resolution. The proposed method is designed to borrow strength across locations within a tissue, thus identifying marker genes with remarkable spatial patterns. We will highlight in this talk the applications of the proposed method in a Visium dataset of the human lymph node.