These tables contain feature importance rankings for the coefficients of the recommended model (L2 regression). Specifically, there is one table for each gene set. Each table has 21 columns corresponding to the 21 different binary predictive tasks. The rows of each table are the ensembl IDs of the genes in the gene set. The values are the mean (across five CV folds) normalized ranking of that gene's coefficient in the recommended model on the corresponding task.