The researchers devised an AI-based method to find chemicals that stop alpha-synuclein, the protein that causes Parkinson's, from clumping together.

The team quickly reviewed a chemical library with millions of entries using machine learning methods and found five powerful compounds that require further research.

Parkinson's disease

More than six million people around the world have Parkinson's disease, and that number is expected to triple by 2040. At this time, there are no medicines that can change the course of the disease. Screening huge collections of chemicals for possible drugs takes a very long time, costs a lot of money, and doesn't always work. It must be done long before potential treatments can be tried on patients.

Researchers used machine learning to speed up the initial screening process ten times and cut the cost by a thousand times. It could mean that patients could get possible treatments for Parkinson's much faster. The findings are published in the journal Nature Chemical Biology.

Brain disease

Parkinson's is a brain disease that is spreading the fastest around the world. One out of every 37 people living in the UK will be diagnosed with Parkinson's at some point in their lives. Parkinson's can affect more than just the muscles. It can also affect the digestive and nervous systems, sleeping habits, mood, and thinking. These things can make life less enjoyable and severely limit a person's abilities.

When someone has Parkinson's, these proteins go rogue and kill nerve cells. Proteins usually do important things inside cells. When proteins don't fold correctly, they can form Lewy bodies, which are weird groups of proteins that build up inside brain cells and stop them from working correctly.

No medicine

Although there are ongoing clinical trials for Parkinson's disease, no medicine that can influence the progression of the condition has been authorized yet. It reflects the inability to specifically target the molecular species responsible for causing the disease.

The absence of practical techniques to identify and interact with the appropriate molecular targets has been a significant challenge in Parkinson's research. This technology divide has significantly impeded progress in creating efficient treatments.

Machine learning technique

The research team devised a machine-learning technique to analyze chemical libraries comprising millions of compounds. The goal was to identify tiny molecules that can bind to the amyloid aggregates and impede their growth.

Subsequently, a limited quantity of highly ranked compounds underwent experimental testing to identify the most effective aggregation inhibitors. The data obtained from these experimental assays was incorporated into the machine learning model iteratively, identifying potent compounds after several iterations. 

Conclusion

The research team used this method to create chemicals that target pockets on the surfaces of the aggregates. These pockets cause the aggregates to grow exponentially. These chemicals are hundreds of times more potent than those described before and much cheaper to make. The study took place in Cambridge's Chemistry of Health Laboratory, which was set up with help from the UK Research Partnership Investment Fund (UKRPIF) to encourage academic research in clinical programs.

Source: https://www.nature.com/articles/s41589-024-01580-x

Image source: Unsplash