Jul 10, 2023 12:46:00

It is reported that AI has derived a candidate molecule for an anti-aging drug that takes several weeks or more to investigate in a few minutes, the possibility of accelerating the drug discovery process

In the process of '

drug discovery', which involves discovering and designing drugs, it is necessary to select compounds and chemical substances that are candidates for creating the desired drug, and to screen their actions. The process of drug discovery requires a large amount of money and a long time, but by performing machine learning, AI was able to derive candidate anti-aging drug molecules in just five minutes. Reported by the research team of Barrett et al.

Discovery of senolytics using machine learning | Nature Communications
https://doi.org/10.1038/s41467-023-39120-1

AI finds drugs that could fight aging and age-related diseases

https://theconversation.com/ai-finds-drugs-that-could-fight-ageing-and-age-related-diseases-208390

Cells called

senescent cells are cells that have irreversibly stopped cell division due to damage to chromosomes caused by various stresses. An increase in the number of senescent cells in the body is thought to increase the risk of developing diseases such as type 2 diabetes , novel coronavirus, interstitial pneumonia osteoarthritis , and cancer.

Previous studies have shown that administering anti-aging drugs to mice to eliminate senescent cells can reduce the risk of these diseases. Also, these drugs can effectively kill only senescent cells without adversely affecting healthy cells.

About 80 anti-aging drugs have been discovered so far, but only two, a combination of dasatinib and quercetin, have actually been tested in humans. It is hoped that anti-aging drugs that can be used for various diseases will be discovered, but it takes decades and billions of dollars to bring drugs to the market through drug discovery and manufacturing. It is considered a problem that it costs (hundreds of billions of yen).

Barrett's research team trained a machine learning model using AI to identify new anti-aging drugs. The research team gave AI examples of known anti-aging drugs and non-anti-aging drugs and learned to distinguish between the two.

As a result of learning to predict whether molecules that have never been seen before can be anti-aging agents, AI was given 4340 types of molecules, and after only 5 minutes, AI displayed a list of results. The list presented by AI contained 21 likely anti-aging molecules.

Mr. Barrett said, ``If we were to identify molecules that are candidates for anti-aging drugs from 4,340 kinds of molecules, it would take at least several weeks and cost more than 50,000 pounds (about 9 million yen). I would.”

The research team then tested the anti-aging drug candidates discovered by AI to see what effects they had on normal and senescent cells. As a result of the test, out of 21 kinds of compounds, 3 kinds of periplocin , oleandrin and ginkgetin succeeded in removing only senescent cells without affecting normal cells.

Further tests were carried out by the research team to investigate in detail how these three anti-aging drug candidate molecules act in animals. Of these candidates, oleandrin proved to be more effective than the known anti-aging agents.

``If we can confirm that AI-powered machine learning provides data of sufficient quality, we can accelerate the process of developing treatments and drugs for chemists and biologists,'' Barrett said. It is.'

The research team is testing the three types of molecules discovered in the experiment in human lung tissue, and the results will be reported in 2025.