Features selected by SVMRFE for each omic type after 5-fold cross-validation.

Cumulative score is the classification metric (f1-score) of an SVM used to predict the cluster labels using only the features at and above that feature (e.g., score for Fibroblast growth factor 20 is the result of an SVM trained on feature set [Interleukin-23, Fibroblast growth factor 20]). These scores are not used to select the size of the feature sets because the size is selected before features are ranked (see Methods).