时间：14:00 - 15:00

地点：数学高等研究院报告厅

组织者：阮勇斌，黄德力，印焘， 宋海，周家耀

报告人：杨灿 Can Yang（香港科技大学）

摘要：Cellular functions within the tissue ecosystem are profoundly shaped by their immediate local environments, known as functional niches. Deciphering the functional niches within complex tissues is essential to understanding how localized cellular interactions drive overall tissue physiology and pathology. While spatial transcriptomics enables high-resolution mapping of gene expression in situ, systematically deciphering cell–cell signaling dialogues remains a significant challenge. Specifically, characterizing a functional niche requires not only defining the spatial communities formed by co-localized cell neighborhoods, but also deciphering the active molecular dialogues and ligand-receptor signaling driving these interactions. Here, we introduce S2-MAP, a statistical model designed to infer putative interacting neighbor cells and the resulting alterations in gene expression programs at the same time. Tailored for multi-condition and multi-slice ST data, S2-MAP enables the detection of both shared and condition-specific interaction niches. In a pancreatic cancer data, our framework traced niche remodeling during tumor progression and captured condition-specific cellular interaction shifts driven by p53 deficiency. Furthermore, across mouse kidney datasets spanning six stages of injury and repair, we identified injury-specific fibro-inflammatory niches and reconstructed the complete trajectory of cellular crosstalk.