Repurposing FDA-approved drugs for treatment of amyotrophic lateral sclerosis using machine learning.

Abstract

Introduction: Amyotrophic Lateral Sclerosis (ALS) is a neurodegenerative disease characterized by loss of motor neurons. Current medications are largely ineffective, associated with side effects, and hindered by a lack of agreement over treatment pathways. The time-intensive process and high costs further limit the development of therapeutics. Therefore, this research aimed to identify FDA-approved drugs that inhibit three proteins (Casein kinase 1, Protein tyrosine kinase 2, Ephrin type-A receptor 4) associated with ALS.

Methods: A machine learning (ML) model was trained for each protein to identify an inputted compound as an active inhibitor of that protein. The FDA-approved drugs were then screened through these models, and 18 drugs were identified as likely inhibitors for all three proteins. The results were validated through protein-ligand docking of each drug to its respective protein(s).

Results: Risperidone was the most active drug, with an average ML score of 1 and binding affinity of -8.9. The ML scores and binding affinities had a strong correlation, indicating reliability.

Conclusion: This research predicted multiple drugs that can simultaneously target many proteins involved in ALS, creating more effective treatment options at a lower cost. This procedure can be applied to efficiently discover drugs for other diseases in the future.