Protease data mining can take advantage both of the many specialist, Web-available databases that cover the genetic, protein and nucleic acid sequence information that is specific to a variety of organisms, and of a flexible, but defined, classification system. However, precomputed data, such as gene predictions, should be used with care. Unless there is definitive supporting information, ideally sequencing of a cDNA to show that the predictions are accurate, followed by expression and biochemical characterization of the predicted protein, the predicted gene and its product remains a possibility, rather than a certainty.