Photo credit: phys.org
AI Enhances Drug Development for Mental Health Disorders
Recent advancements in artificial intelligence (AI) are paving the way for the identification of new drug candidates targeting mental health disorders. A study from Uppsala University, published in the journal Science Advances, highlights how AI can effectively predict the three-dimensional structures of critical receptors, thereby expediting the drug discovery process.
In the traditional drug development pipeline, experimental techniques are often employed to elucidate the three-dimensional configurations of target proteins. Understanding how specific molecules bind to these proteins is essential for the efficient design of new pharmaceuticals. However, determining these structures can be resource-intensive, which restricts the feasibility of this approach in many cases.
The study demonstrates that AI methodologies have significantly increased the accuracy of protein structure predictions. Researchers at Uppsala University focused on the TAAR1 receptor, known for its potential in mental health drug development. Agonists that activate TAAR1 have previously indicated encouraging outcomes for conditions such as schizophrenia and depression.
Utilizing supercomputers, the research team probed extensive chemical libraries containing millions of compounds to identify those that would optimally interact with the AI-generated model of the TAAR1 receptor. Subsequent experimental testing at Karolinska Institutet revealed that a substantial number of the identified molecules could activate TAAR1. Notably, one of the most effective candidates exhibited promising results in preclinical animal trials.
During the concluding phase of the research, experimental structures of the TAAR1 receptor became available, enabling the team to validate their AI-derived models against these actual structures.
“The precision of the structures produced by AI was remarkable—far exceeding our expectations,” remarked Jens Carlsson, who led the research efforts at Uppsala University. “These findings underline that AI modeling is not only advantageous but is also a transformative approach compared to conventional methods. We can now apply these strategies to receptor targets that were previously considered out of reach,” he added.
More information: Alejandro DÃaz-HolguÃn et al, AlphaFold accelerated discovery of psychotropic agonists targeting the trace amine–associated receptor 1, Science Advances (2024). DOI: 10.1126/sciadv.adn1524
Citation: AI predicts 3D structures of receptors for drug development (2024, August 8) retrieved 8 August 2024 from https://phys.org/news/2024-08-ai-3d-receptors-drug.html
Source
phys.org