Computational structural genomics: large-scale structural annotation of the proteins encoded by genomes by identifying homologies between the genomic proteins and proteins of known structure. - Methods and databases for computational structural genomics: complete proteome assignments to CATH domains in the Gene3-D database, SCOP domains in the SUPERFAMILY database. - Insights into evolution of proteomes and multidomain proteins through structural annotation: a few large families1 dominate proteomes, and there are many small families. At least two-thirds of the proteins in a proteome are multidomain proteins. In multidomain proteins, the large families have many types of domain neighbors, while the small families only occur next to one or two other types of domains. - Insights into enzyme and metabolic pathway evolution through structural annotation: enzymes are assembled into metabolic pathways in a mosaic manner, with domain families tending to be distributed across different pathways.