https://bio.nikkeibp.co.jp/atcl/news/p1/24/07/29/12195/
AMED in 2019 had initiated a project where artificial intelligence (AI) has been trained to learn not only public databases but also assay data, including structural formulas, held by 17 domestic pharmaceutical companies, in order to develop an integrated drug discovery AI platform. The application and verification of this AI platform has begun for 5 to 10 academic drug discovery targets since the beginning of FY2024, and 17 pharmaceutical companies will also begin using the platform by the end of FY2024.
The assay data held by companies, including the structural formulas of compounds, was considered to be very useful in advancing the development of AI. Thus aa method called federated learning was introduced, which allows multiple organizations to cooperate in training an AI model by having the AI learn the data and taking outside only the AI model parameters. It is considered “a method of building AI in cooperation while competing.”
The result is an integrated drug discovery AI platform that integrates AI that predicts compound profiles, AI that proposes new compounds, and predictive AI based on omics information. It is hoped that this AI that proposes promising compound structures as drug candidates will dramatically streamline drug discovery. Among the features is AI that predicts target molecules based on omics information such as gene expression when a compound is added to cells, suggests the mechanism of action of a compound, and generates the structure of a new compound with a gene expression pattern that cancels the gene expression pattern during disease.
The AI is already being ported to the following 17 participating companies (see box below), and it is expected that each company will fully utilize it for drug discovery.
Eisai, Ono Pharmaceutical, Kaken Pharmaceutical, Kissei Pharmaceutical, Kyorin Pharmaceutical, Kyowa Kirin, Sanwa Kagaku Kenkyusho, Taiho Pharmaceutical, Takeda Pharmaceutical, Mitsubishi Tanabe Pharma, Teijin Pharma, Torii Pharmaceutical, Toray, Nippon Chemipha, Nippon Shinyaku, Meiji Seika Pharma, and 1 other company
In order to maintain and further develop the project results, the Life Intelligence Consortium (LINC), a general incorporated association represented by Professor Okuno will be entrusted with the operation. The 17 companies participating in DAIIA will be able to continue using the AI by paying maintenance and renewal fees, and it is also expected that they will pay license fees to the academia involved in the development.