Individualized preemptive medicine utilizing artificial intelligence and genomic information

Abstract

Targeted goal of our project is to realize precision medicine of genetic diseases caused by aberrant splicing, induced by deep-intronic VUS. In eukaryotic gene expression system, a precursor mRNA transcribed from genome consists of exonic sequences and intervening intronic sequences, and exonic sequences are connected by RNA splicing reaction, resulting in production of protein-coding sequences. Splicing regulation is regulated by multiple factors, such as chromosomal structure, RNA cis-elements, and trans-acting splicing factors, and due to its complexity, precise prediction of alternative splicing profiles still remains to be achieved today. Recent accumulation of whole genome sequence data facilitated access to deep-intronic sequences, which provides essential information to understand splicing codes. Compared to conventional exome sequencing data, covering approximately 1% of genomic information, whole genome sequencing revealed presence of numerous VUS within the deep-intronic region, which are not accessed by exome studies. Our preliminary observations indicate there are many deep-intronic VUS that affect splicing code to create pseudoexonization of a given intronic region, resulting in pathogenesis by causing frameshifting or insertion, as we recently characterized for NEMO deficiency syndrome (J. Clin. Invest. 2019) and cystic fibrosis (Cell Chem. Biol. 2020).　To achieve our goal, we are challenging elucidation of splicing code using artificial intelligence (AI)-driven novel strategy. We will conduct a genome-wide functional annotation for deep-intronic VUS, for more than 10,000 whole genome sequence data available from Tohoku Medical Megabank (ToMMo), The 1,000 Genome Project, and Genome Asia 100k Project. We constructed an original AI and conduct a clinical trial to realize precision medicine of genetic diseases.