AI Method Accelerates Drug Discovery for Parkinson’s 10-Fold, Study Finds

University of Cambridge researchers have harnessed the power of artificial intelligence (AI) to speed up the screening of new drugs to treat Parkinson’s disease.

The technology identified 5 highly potent potential drug candidates to take forward for further analysis, suggesting the method can speed up the drug discovery process 10-fold.

The research is published in Nature Chemical Biology.

Developing disease-modifying treatments

Parkinson’s disease affects over 6 million people across the globe. It causes a wide range of symptoms, varying from its characteristic motor symptoms to problems affecting the gut, sleep, mood and cognition.

The number of people with Parkinson’s is expected to triple by 2040 and it is the fastest-growing neurological condition worldwide. Despite its growing burden, no disease-modifying treatments – which aim to directly target the mechanisms causing the disease to improve its symptoms – have yet been approved for Parkinson’s.

Parkinson’s is thought to be caused by rogue proteins that misfold and clump together to form Lewy bodies, eventually building up within nerve cells leading to impaired function or even cell death.

Trials for potential disease-modifying Parkinson’s treatments are in progress, but experimental methods to identify the correct molecular targets are lacking – creating a technological gap that has hampered their development.

“One route to search for potential treatments for Parkinson’s requires the identification of small molecules that can inhibit the aggregation of alpha-synuclein, which is a protein closely associated with the disease,” said the study’s lead author Prof. Michele Vendruscolo, a professor of biophysics in Cambridge’s Yusuf Hamied Department of Chemistry. “But this is an extremely time-consuming process – just identifying a lead candidate for further testing can take months or even years.”

Vendruscolo and colleagues from the University of Cambridge harnessed the power of AI in their new study to accelerate and decrease the costs associated with Parkinson’s drug development.

Iterative screening with AI

The researchers developed a machine learning-based approach to screen libraries containing millions of chemical compounds to identify candidates that bind to and prevent the growth of the protein aggregates.

The top-ranking compounds were then tested experimentally to find the ones that inhibited protein aggregation with the most potency. This information was then fed back into the machine learning model in iterations, to eventually identify the best candidate compounds.

“Instead of screening experimentally, we screen computationally,” explained Vendruscolo. “By using the knowledge we gained from the initial screening with our machine learning model, we were able to train the model to identify the specific regions on these small molecules responsible for binding, then we can re-screen and find more potent molecules.”

This method allowed the researchers to develop compounds targeting pockets on the aggregate surface that enable their proliferation. These compounds are significantly more potent – and less expensive to develop – than previous examples.

“Machine learning is having a real impact on the drug discovery process – it’s speeding up the whole process of identifying the most promising candidates,” Vendruscolo said. “For us, this means we can start work on multiple drug discovery programs – instead of just one. So much is possible due to the massive reduction in both time and cost – it’s an exciting time.”

Reference: Horne RI, Andrzejewska EA, Alam P, et al. Discovery of potent inhibitors of α-synuclein aggregation using structure-based iterative learning. Nat Chem Biol. 2024. doi: 10.1038/s41589-024-01580-x 

This article is a rework of a press release issued by the University of Cambridge. Material has been edited for length and content.