Functional annotation prediction: all for one and one for all