A Novel Method to Identify the Differences Between Two Single Cell Groups at Single Gene, Gene Pair, and Gene Module Levels.

Single-cell sequencing technology can not only view the heterogeneity of cells from a molecular perspective, but also discover new cell types. Although there are many effective methods on dropout imputation, cell clustering and lineage reconstruction based on single cell RNA sequencing (RNA-seq) data, there is no systemic pipeline on how to compare two single cell clusters at the molecular level. In the study, we present a novel pipeline on comparing two single cell clusters, including calling differential gene expression, coexpression network modules, and so on. The pipeline could reveal mechanisms behind the biological difference between cell clusters and cell types, and identify cell type specific molecular mechanisms. We applied the pipeline into a famous single cell database named the Usoskin data, consisting of 622 single cell from 4 cell clusters: Neurofilament containing, non-peptidergic nociceptors, Peptidergic nociceptors, and tyrosine hydroxylase containing. As a result, we identified many significant differential genes, differential gene coexpression and network modules among the cell clusters, which confirmed that different cell clusters might perform different functions.