Skip to main content
. 2020 Feb 13;36(10):3115–3123. doi: 10.1093/bioinformatics/btaa097

Fig. 2.

Fig. 2.

Boxplots of metrics evaluating batch-effect correction methods in the 50 simulations with gene-specific batch-effects, under different configurations of sample size. PR-AUC, AUC and NDE(0.2) were calculated based on adjusted P-values from Seurat DE tests