Thanks to various genome projects, genome scale protein sequence data have become available. Predicting gene locations and gene functions by computational methods is a very important stage in the genome project. We are developing a systematic method of predicting enzyme genes for an organism with its entire genomic sequence known, utilizing the knowledge organized in KEGG (Kyoto Encyclopedia of Genes and Genomes). The system can also help reconsider the functional assignments of the genomes previously determined and the gene catalog can be fixed accordingly.