ABSTRACT

Summary: Single-cell RNA sequencing technology is a powerful method for dissecting intercellular heterogeneity during development and reprogramming. However, it can rarely detect signaling genes due to detection limit, and cannot distinguish biologically relevant versus irrelevant variance and therefore often has to discard cell cycle related events in the analysis. Here, we show that one way to address these limitations is to simultaneously generated and analyzed cell population RNA-seq (cpRNA-seq) and single-cell RNA-sequencing (scRNA-seq) data. We developed the iCpSc package to integratively analyze cpRNA-seq and scRNA-seq data, and applied our methods to in vitro neural differentiation of mouse embryonic stem cells mESCs. By generating a computational model for the reference ¡°biological differentiation time¡± using the cell population data and apply it to the single cell data, we unbiasedly associate cell-cycle check points to the internal molecular timer at single cell level, which is obscured by the single cell RNA-seq analysis alone. By inferring a network flow cpRNA-seq to scRNA-seq data, we unbiasedly inferred and validated a role of M phase in controlling the speed of neural differentiation. Our novel approach to linking the temporally matched cpRNA-seq and scRNA-seq data provide an invaluable resource for understanding timing of differentiation.