Finding new drugs – called “drug discovery” – is an expensive and time-consuming task. But a form of artificial intelligence called machine learning can vastly speed up the process and get the job done for a fraction of the cost.
My colleagues and I recently used this technology to find three promising candidates for senolytics – drugs that slow down aging and prevent age-related diseases.
Senolytics work by killing senescent cells. These are cells that are “alive” (metabolically active), but can no longer multiply, hence their nickname: zombie cells.
The inability to replicate is not necessarily a bad thing. These cells have had damage to their DNA, for example skin cells damaged by the sun’s rays, so stopping the replication prevents the damage from spreading.
But senescent cells aren’t always a good thing. They separate one cocktail of inflammatory proteins that can spread to neighboring cells. Throughout our lives, our cells go through a barrage of attacks, from UV rays to chemical exposure, and that’s how these cells pile up.
Increased numbers of senescent cells are involved in a range of diseasesincluding type 2 diabetes, COVID, pulmonary fibrosis, osteoarthritis and cancer.
Studies in laboratory mice have shown to eliminate senescent cells, using senolytics, can improve these diseases. These drugs can kill zombie cells while keeping healthy cells alive.
All around 80 senolytics are known, but only two have been tested in humans: a combination of dasatinib and quercetin. It would be great to find more senolytics that can be used in different diseases, but it takes ten to twenty years and billions of dollars to get a drug on the market.
Results in five minutes
My colleagues and I – including researchers from the University of Edinburgh and the Spanish National Research Council IBBTEC-CSIC in Santander, Spain – wanted to know if we could train machine learning models to identify new senolytic drug candidates.
To do this, we fed AI models with examples of known ones senolytics and non-senolytics. The models learned to distinguish between the two and could be used to predict whether molecules they had never seen before could also be senolytics.
When solving a machine learning problem, we usually first test the data against a range of different models, as some perform better than others.
To determine the best performing model, we segregate a small portion of the available training data at the beginning of the process and keep it hidden from the model until after the training process is complete.
We then use this test data to quantify how many errors the model makes. The one who makes the fewest mistakes wins.
We determined our best model and set it up to make predictions. We gave it 4,340 molecules and five minutes later it returned a list of results.
The AI model identified 21 top-scoring molecules that it thought were most likely senolytics. If we had tested the original 4,340 molecules in the lab, it would have taken at least a few weeks of intensive work and £50,000 to buy the compounds, not counting the cost of the experimental machinery and setup.
We then tested these drug candidates on two types of cells: healthy and senescent cells. The results showed that of the 21 compounds, three (periplocin, oleandrin and ginkgetin) were able to eliminate senescent cells, while leaving most normal cells alive. These new senolytics were then further tested to learn more about how they work in the body.
More detailed biological experiments showed that of the three drugs, oleandrin was more effective than the top performing known senolytic drug of its kind.
The potential implications of this interdisciplinary approach – involving data scientists, chemists and biologists – are enormous. Given enough high-quality data, AI models can accelerate the amazing work that chemists and biologists do to find treatments and cures for diseases, especially those with unmet needs.
Having validated them in senescent cells, we are now testing the three candidate senolytics in human lung tissue. We hope to be able to report our next results in two years’ time.
Vanessa Smer-BarretoResearch Fellow, Institute of Genetics and Molecular Medicine, The University of Edinburgh
This article has been republished from The conversation under a Creative Commons license. Read the original article.