Hi everyone. I used the machine learning /cross-validation function of PERSEUS (maxquant) to determine a set of proteins that can discriminate a cancer subtype from the others. The result is a list of ranked proteins, and the software "said" that the 13 top proteins discriminate the subtypes with only a 10% error. Now, I want to confirm the minimum number of this proteins that discriminate the subtype using a ROC(AUC) curve... I saw various R packages that make ROC curves, but only curves to individual proteins, and not mesure the potential of combined proteins... Someome have any tips on how do this?