Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: CeDAR: incorporating cell type hierarchy improves cell type-specific differential analyses in bulk omics data

Fig. 3

Simulation results for comparing different methods in cell type-specific differential expression. The simulation is based on a two-group comparison, with 100 samples in each group. Data were generated as a mixture of six common blood immune cell types (1: neutrophils, 2: monocytes, 3: CD4, 4: CD8, 5: B, 6: NK cells). a Cell type hierarchy used in simulation. b Mean proportion of each cell type. c ROC curves for csDE detection in six cell types for six methods (TOAST, TCA, csSAM, CellDMC, CeDAR-S, and CeDAR-M). Reported ROC curves are averaged from 50 simulations. d Observed FDR for csDE detection from different methods. DE genes are defined with rules: estimated FDR < 0.05 (TOAST, TCA, csSAM, and CellDMC); posterior probability of DE > 0.95 (CeDAR-S, CeDAR-M). Observed FDR from 50 simulations are summarized by box plot

Back to article page