Abstract
There are far more carbohydrate structures than protein folds and, not unexpectedly, sequence-based families of carbohydrate-active enzymes often group together enzymes of different specificity. While this fact illustrates the remarkable plasticity of proteins to evolve novel specificities from pre-existing folds, this is also the source of potential errors during the functional prediction of genes derived from whole genome sequencing efforts. This chapter reviews the problems and the advances associated with the sequence-based classification of plant carbohydrate-active enzymes as used in the CAZy database (www.cazy.org).