Automatic generation of KEGG OC (Ortholog Cluster) and its assignment to draft genomes

As the number of sequenced genomes are rapidly growing, a method for automatic generation of orthologous gene clusters is needed. However, it is computationally hard to cluster a large number of genes at once. To address this problem, we have developed a heuristic method to assign gene groups from closely related organisms to an ortholog cluster in a bottom-up approach. In this method, we consider each gene subgroup as a representative gene and find their correspondence using bi-directional best hit (BBH) relations obtained from the KEGG SSDB database which stores all-vs-all Smith-Waterman similarity scores [1]. We have clustered all the genes in the KEGG GENES database [1] to generate KEGG Ortholog Clusters (OCs) which represent various aspects of the protein universe. As an application of KEGG OC, we have performed automatic gene assignment of the draft genomes which are not yet included in KEGG. Our method provides an efficient way for rapid annotation of the genes of newly sequenced organisms.