Repeating GSEA with the 50 hallmarks finds 12 that are significantly enriched (FDR < 0.25, Table S2). The three top hallmarks correspond to the biological themes in the top 100 full MSigDB analyses, but the hallmark collection is more sensitive and also highlights the other eight biological processes. Notably, the hallmarks not only produce more parsimonious but equivalent results representing the main biological themes found in the prior analysis, but they also avoid the problem of gene set redundancy and over-representation altogether.
