Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2020.11.16.373274v1?rss=1 Authors: Wang, Z., Yang, S., Koga, Y., Corbett, S. E., Johnson, W. E., Yajima, M., Campbell, J. D. Abstract: Complex biological systems can be understood by dividing them into hierarchies. Each level of such a hierarchy is composed of different subunits which cooperate to perform distinct biological functions. Single-cell RNA-seq (scRNA-seq) has emerged as a powerful technique to quantify gene expression in individual cells and is being used to elucidate the molecular and cellular building blocks of complex tissues. We developed a novel Bayesian hierarchical model called Cellular Latent Dirichlet Allocation (Celda) to perform bi-clustering of co-expressed genes into modules and cells into subpopulations. This model can also quantify the relationship between different levels in a biological hierarchy by determining the contribution of each gene in each module, each module in each cell population, and each cell population in each sample. We used Celda to identify transcriptional modules and cell subpopulations in publicly-available peripheral blood mononuclear cell (PBMC) dataset. In addition to the major classes of cell types, Celda also identified a population of proliferating T-cells and a single plasma cell that was missed by other clustering methods in this dataset. Transcriptional modules captured consistency in expression patterns among genes linked to same biological functions. Furthermore, transcriptional modules provided direct insights on cell type specific marker genes, and helped understanding of subtypes of B- and T-cells. Overall, Celda presents a novel principled approach towards characterizing transcriptional programs and cellular and heterogeneity in single-cell data. Copy rights belong to original authors. Visit the link for more info