Can we use peptide activity in a more useful way before trying to predict AUC? Use networks, correlation, etc.