Fig. 1

Analysis of the TCGA transcriptomic database. Dotplots showing functional enrichment for co-expression modules found in various cancer types and predominant sarcoma subtypes. Htseq raw counts were retrieved from TCGA using GDCquery [22] and VST-normalized [23]. For each dataset, the unsigned co-expression network was produced using WGCNA with automatic pick for soft-thresholding powers. Genes in each module were queried for functional enrichment against Reactome Pathway Database [24] using clusterProfiler [25]. p values were adjusted using Benjamini–Hochberg procedure. For each dataset-pathway pair, the p value corresponds to the lowest one from all the co-expression modules. A subset of the significant (q-value < 0.05) pathways was manually annotated into functional groups for display in the figure. Dots highlight significant pathway-dataset pairs