A clean data set of annotated, experimentally verified, glycosylated and nonglycosylated sequons derived from 617 well-defined nonredundant N- and N-,O-glycoproteins listed in SWISS-PROT (June 2002) was used.