Abstract: Classifying proteins into evolutionary families is important for identifying conserved sequence and structure features that are key to the functional mechanisms of these proteins. Our in-house CATH classification currently classifies ~450,000 protein structures and nearly 150 million protein domain sequences into ~5500 evolutionary families. The recent success in protein structure prediction by DeepMind’s AlphaFold2 (AF2) method and the expected release of hundreds of thousands of AF2 models, will change the scientific landscape by massively extending the structural data available for these protein evolutionary families. We have developed a strategy to bring this extensive new 3D data into CATH families and are examining how this data will expand our understanding of structure – function relationships and our ability to detect functional sites. Functional site predictions can be enhanced by combining structural features and evolutionary conservation patterns and some examples will be given of the application of CATH functional site data to understand protein splice events, the risk of Covid infection and the development of more aggressive lung cancer following whole genome duplication.