We determined significant enrichment for gene sets using the GSEA87,88, which is less susceptible to gene length bias because it uses permutation enrichment within gene sets. In this study, we performed GSEA with the GO gene set database from the clusterProfiler package (v.4.6.2)89 and DisGeNET gene set database90 from the DOSE package (v.3.24.2)91. We defined the gene set ‘universe’ as all unique genes tested for differential expression. When examining isoform-level enrichment (transcript, exon or junction), we selected, for each unique gene, the feature with the largest absolute effect size. For the GO gene set database, the minimal gene set size (minGSSize) was set to ten, the maximum gene set size (maxGSSize) was set to 500, and the P cutoff was set to 0.05. For theDisGeNET gene set database, minGSSize was set to five and the P cutoff to 0.05. We used the default settings for all other parameters.
