Fig. 3

Pairwise differential expression of GSEA-driving genes. 33 genes were selected based on belonging to at least one comparison-wise enrichment core and on showing significant F-statistics at the same time (Benjamini–Hochberg correction, 5% FDR). Each box-and-whisker representation summarizes the batch-corrected, logarithmic counts-per-million (log₂ CPM) of one gene in different sample groups (Subtype). Individual data points are also shown. The genes are sorted row-wise by average expression. IrMyositis, Mi2 subtype, TIF1γ subtype, and NDC are denoted by different colors. Horizontal brackets indicate significant pairwise DGE, after applying Benjamini–Hochberg correction (5% FDR, hierarchical post hoc correction – see Methods for further details).