Classification and Analysis of KEGG/REACTION Database

For recent decades, the EC (Enzyme Commission) number has been used to research various biological reactions and metabolic pathways. However, the EC classification is unsuitable for bioinformatics analyses because of the, at least, following two problems. One is inconsistent criteria for classification, utilizing reaction patterns, substrates, transfered groups, and acceptor groups in different ways in different hierarchies. Another is that when one EC number is given to an enzyme which catalyzes an overall reaction consisting of multiple elementary reactions steps, the same number is often assigned to all elementary reactions. For example, there are several reactions which oxidize substrate and then dissociate inorganic compounds (CO2, NH3, etc.), but once these reactions are classified into an overall oxidoreductase reaction (EC.1), it is not easy to reveal inherent property of dissociation only from the EC number. In order to solve these problems, we are constructing a new classification scheme for enzymatic reactions as accumulated in the LIGAND/REACTION database [3]. Here, we report the decomposition of reactions into the combinations of reactant-product pairs and its application to the classification of reactions.