报告题目：Data Integration in Spatial Transcriptomics

报告时间：2022-07-11  10:00 - 11:00

报告人：刘瑾  新加坡国立大学

ZOOMID：561 420 9883   密码：tmcc2022

Abstract：Spatially resolved transcriptomicsare a set of emerging technologies that enable transcriptomic profiling on tissues with their physical locations. Although a variety of methods have been developed to perform data integration, most of them are for single-cell RNA-seqdatasets without consideration of spatial information. Thus, methods that are capable of integrating spatial transcriptomicsdata from multiple tissue slides, possibly in multiple individuals, are highly needed. Here, we present PRECAST, an efficient data integration method for multiple spatial transcriptomicsdata with non-cluster-relevant effects such as the complex batch effects. PRECAST unifies spatial factor analysis simultaneously with spatial clustering and embedding alignment, requiring only partially shared cell/domain clusters across datasets. We show that PRECAST effectively integrates multiple tissue slides with spots mixed across datasets and cell/domain clusters separated in both simulated and four spatial transcriptomicsdatasets from either low or near single-cell resolution, demonstrating the improved cell/domain detection with outstanding visualization, and the estimated embeddingsand cell/domain labels facilitate many downstream analyses. Particularly, with a hepatocellular carcinoma Visiumdataset, we detected two cell lineages in tumor/normal epithelium via spatial RNA velocity analysis using the estimated embeddingsand domain labels by PRECAST. With a mouse olfactory bulb Slide-seqV2 dataset containing 16 slides of ~700,000 spots, we further show the scalability of PRECAST.