A lack of distinct cellular identities in scRNA-seq data: revisiting Waddington’s landscape




11:00 a.m.


MRB 202 Conference Room and Zoom (email matthew@ku.edu for the link)


Eric J. Deeds (Associate Professor, Department of Integrative Biology and Physiology, Institute for Quantitative and Computational Biosciences, UCLA)


Single-cell RNA sequencing is revolutionizing our understanding of development, differentiation and disease. Analysis of this data is often challenging, however, and tasks like clustering cells to uncover distinct cellular identities sometimes yields results that fail to align with existing biological knowledge. We analyzed publicly available data where the cell identity for each cell is known a priori, and found that cells of very different types and lineages do not occupy distinct regions of gene expression space. Rather, cells from different lineages overlap extensively with one another, significantly complicating attempts to recover distinct identities within the data. Indeed, our analysis of available epigenetic data for a wide variety of tissues and organisms revealed these data are not consistent with the predictions of Waddington’s landscape, suggesting a need to revisit our picture of gene expression changes during differentiation and development.