Fig. 7. Comparison of runtimes, similarity of differential expression (DE) workflows and other measures.
a CPU times (log-scale) of DE workflows taken for analyzing lung adenocarcinoma (LUAD) epithelial cell (n = 7728) and COVID-19 monocyte (n = 100361) scRNA-seq data, each containing 10278 and 7242 genes, respectively. b Clustering heatmap of 46 DE workflows for LUAD epithelial cell data and four DE methods for TCGA LUAD data. The Spearman rank correlation of DE genes were used as the similarity measure. c Comparison of the capability of prioritizing disease-related genes between 46 DE workflows and their performance classified in terms of false positive/discovery controls, sign preservation of DE genes, speed, and scalability.