Think about implementing bi-cross-validation to the data

[aarmey]

This could help to explore how much the kinase enrichment helps with predicting the data across phosphosites, internally.