Bias, robustness and scalability in single-cell differential expression analysis

Many methods have been used to determine differential gene expression from single-cell RN A (scRNA)-seq data. We evaluated 36 approaches using experimental and synthetic data and found considerable differences in the number and characteristics of the genes that are called differentially expressed. Prefiltering of lowly expressed genes has important effects, particularly for some of the methods developed for bulk RN A-seq data analysis. However, we found that bulk RN-Aseq analysis methods do not generally perform worse than those developed specifically for scRN A-seq. We also present conquer, a repository of consistently processed, analysis-ready public scRN A-seq data sets that is aimed at simplifying method evaluation and reanalysis of published results. Each data set provides abundance estimates for both genes and transcripts, as well as quality control and exploratory analysis reports.