Acceleration of Large-Scale Single Cell Differential Gene Expression Analysis with FastDE

The major bottleneck in single cell transcriptomic processing is identification of marker genes for individual cell clusters. We introduce FastDE, an implementation of the standard Wilcoxon Rank Sum test in C++ that results in 50X speed-up of the standard Seurat pipeline on a dataset of 1.2 million cells. Proof of methodology is demonstrated by replicating age and gender effects on immune cell differentiation in two large PBMC atlases.