Naoyuki Kamatani
Director, Research Institute for Artificial Intelligence in Medicine, StaGen Co.


Big genomic data have become available from various fields of healthcare and medicine. Although all genomes are coded in the same material DNA and by the same four letters of nucleotides, the associations of the genomic data with phenotypes are quite different between different situations. For locating responsible genes for Mendelian traits, linkage analysis by the maximum-likelihood method is performed using EM algorithm or Hidden Markov Model. To find genes associated with non-Mendelian traits, GWAS (genome-wide association study) is used based on > 500,000 SNPs (single-nucleotide polymorphisms). We performed GWAS for 47 diseases using genome data from 250,000 subjects in which 600,000 SNP data from a person were used. We are now able to sequence all genomic DNA from an individual with 3 x 109 nucleotides. Sequencing individuals and cancer genomes elucidated responsible germline mutations for genetic diseases and somatic mutations for cancers.

The diversity of the situations associated with genomes makes an integrated understanding of the entire field of genomics quite difficult. I proposed a powerful conceptual framework to understand, do research, make new tests and develop new therapeutic strategies in the field of genomics. Thus, I proposed a six-layer structure that describes the entire scientific field for “genomics”. The proposed layers are “life” as the uppermost layer, followed by “species”, “population”, “family”, “individual”, and finally “cell” as the bottommost layer. In each pair of adjacent layers, each member of the upper layer comprises a set of members of the lower layer. In each layer, we can define two types of causalities, i.e. parent to offspring and genotype to phenotype. I will show that mathematical genetics studies can be understood as attempts to bridge gaps between layers of the proposed six-layer structure. Furthermore, the six-layer structure is useful to develop therapeutic strategies. I will show my experiences in biological and medical researches that suggest the importance of six-layer structure; the discovery of methylthioadenosine phosphorylase (MTAP) deficiency in human cancers, the first proposal of personalized cancer therapy based on MTAP deficiency, and the development of cladribine and febusostat. In all the new drug development, information from genotype-phenotype associations played important roles. I will present some data about our new drug under development, ATP enhancer, and will show how important the information from genotype-phenotype association is in a new drug development.


Currently, I am the Director, Research Institute for Artificial Intelligence in Medicine, StaGen Co. LTD and Director of StaGen Co. LTD. Until Dec. 2011, I served as the director of Center for Genomic Medicine (CGM), RIKEN, Japan. Before that, I had been the director and professor of Institute of Rheumatology, Tokyo Women's Medical University where I served as a clinical rheumatologist. I have published more than 600 peer reviewed papers including 34 Nature or Nature Genetics papers. My publications cover a quite wide area of medicine and biology spanning from cancer genetics, nucleic acid metabolism, gout, rheumatic diseases, monogenic diseases, statistical genetics, statistics, bioinformatics and pharmacogenomics. Recently, I especially focus on genome- wide association studies (GWASs), new drug development based on genomics data and artificial intelligence.

I started to work for StaGen Co. Ltd, a private venture company that is dedicated to statistical and genetic data analysis, new drug development and artificial intelligence. Among a few drugs being developed by our company, ATP enhancer is the most important new drug. This drug is targeted at mitochondrial diseases and neurodegenerative disorders, and the efficacy and safety has been predicted by the data from many genetic diseases and GWASs.