Self-Organizing Clustering: A Novel Non-Hierarchical Method for Clustering Large Amount of DNA Sequences

To cluster and characterize DNA sequences focusing on the oligonucleotide frequency, we developed a novel method and program package designed designated as Self-Organizing Clustering (SOC) [4]. Being based on Self-Organizing Map (SOM) [1, 2, 3], the algorithm of SOC made use of K-means with modification. In the SOC, the oligonucleotide frequency was regarded as a series of oligonucleotide patterns, and calculated by counting the frequency of an individual oligonucleotide. At present, the SOC programs are tested with DNA sequence from several bacterial genomes for its future improvement.