scMEB: A fast and clustering-independentmethod for detecting differentially expressedgenes in single-cell RNA-seq data

Background: Cell clustering is a prerequisite for identifying differentiallyexpressed genes (DEGs) in single-cell RNA sequencing (scRNA-seq) data.Obtaining a perfect clustering result is of central importance for subsequentanalyses, but doing so is not easy. Additionally, the increase in cell throughputdue to the advancement of scRNA-seq protocols exacerbates many computationalissues, especially regarding method runtime. To address these difficulties, a new,accurate, and fast method for detecting DEGs in scRNA-seq data is needed. Results: Here, we propose single-cell minimum enclosing ball (scMEB), a noveland fast method for detecting single-cell DEGs without prior cell clusteringresults. The proposed method utilizes a small part of known non-DEGs (stablyexpressed genes) to build a minimum enclosing ball and defines the DEGs basedon the distance of a mapped gene to the center of the sphere in a feature space. Conclusions: We benchmarked scMEB against two other methods that haveimplications for DEG identification without clustering the cells. The analysisresults of 11 real datasets demonstrated that scMEB had a better or at leastcomparable performance with the competing methods in terms of identifyingmarker genes, cell clustering, and predicting genes with biological functions.Moreover, scMEB ran much faster than the other methods, which is especiallyuseful for detecting DEGs in high-throughput scRNA-seq data. Our method“scMEB” has been incorporated into the R package (MEB) and could beaccessed at https://github.com/FocusPaka/MEB.