Classically solved protein structures are chosen, and prediction tools are only given the amino acid sequences to work with.