Skip to main content

Table 2 Differential expression methods tested in this paper. The “Data sets” column lists the data sets that each method is tested on in this work. The top block contains the methods that are presented in this work; implementations of these methods can be found in the repository linked in the data availability disclosure. The second block of methods consists of general statistical tests. The third block consists of methods that were designed specifically for scRNA-seq data. The fourth block consists of standard machine learning methods; Log. Reg. stands for logistic regression. We also consider selecting markers randomly without replacement. See the marker selection methods description in the Methods for more information

From: A rank-based marker selection method for high throughput scRNA-seq data

Method Data sets Package Version Ref
RankCorr All custom   
Spa Zeisel, Paul implementation   [14]
t-test All scanpy 1.3.7; see text  
Wilcoxon All   1.3.7; see text  
edgeR Zeisel, Paul, ZhengFilt edgeR, rpy2 v2.9.4 3.24.1 [27]
MAST Zeisel, Paul, ZhengFilt MAST, rpy2 v2.9.4 1.8.1 [28]
scVI Zeisel, Paul Source from GitHub 0.2.4 [29]
Elastic Nets Zeisel, Paul scikit-learn [30] 0.20.0 [12]
Log. Reg. All scanpy 1.3.7; see text [31]
Random selection Zeisel, Paul, ZhengFilt,    
  ZhengFull