Gene Set Enrichment Analysis (GSEA) and Single Sample Gene Set Enrichment Analysis (ssGSEA)

Gene set enrichment analysis was performed in python with GSEAPy [12,32]. For consistency with eVIP Pathways, we used the same 50 Hallmark pathways (h.all.v6.0.symbols.gmt) and required at least 10 genes for each gene set. We performed two GSEA comparisons: (1) GFP vs RNF43 G659fs (2) RNF43 WT vs RNF43 G659fs. To account for different eVIP2’s processing steps, we used four types of input data: (1) gene TPM counts, (2) filtered and log2 transformed gene TPM counts, (3) filtered, log2 and z-transformed gene TPM counts, (4) mutation-specific genes from filtered and log2 transformed gene TPM counts. We evaluated both permutation types (1)“gene_set” and (2) “phenotype” and all five GSEA methods for used to rank samples (1) “signal_to_noise”, (2) “t_test”, (3) “ratio_of_classes”, (4) “diff_of_classes”, (5) “log2_ratio_of_classes”. Pathways with a FDR under 0.25 were considered significant. The results for the 80 GSEA runs are available in S11 File.

We used the GSEAPy ssGSEA function to run ssGSEA, with the same four inputs and each of the three sample normalization method parameters (1) "rank”, (2) "log", (3) "log_rank". The resulting normalized enrichment scores of the 48 ssGSEA runs (12 for each of the 4 RNF43 G659fs replicates) are available in S12 File.