Identifying novel protein phenotype annotations by hybridizing protein–protein interactions and protein sequence similarities