Artificial intelligence (AI) has the potential to discover new molecules that may be developed into drugs for mental health issues. By utilizing AI to predict the three-dimensional frameworks of critical receptors, the drug development process can be accelerated. This breakthrough stems from a recent study conducted at Uppsala University, as published in Science Advances.
In the realm of drug development, scientists often rely on experimental techniques to ascertain the three-dimensional configurations of target proteins and comprehend how various molecules interact with them. This knowledge is essential for efficiently designing drug candidates. However, determining these structures can be quite labor-intensive, which limits the applicability of this approach.
Fortunately, advancements in AI technologies now allow for the prediction of protein structures with greater precision than ever before.
In this study, researchers from Uppsala University employed AI to develop a model representing the unknown three-dimensional structure of the TAAR1 receptor, a significant target protein for drug development aimed at treating mental health disorders. Molecules that stimulate TAAR1 have demonstrated encouraging results for conditions like schizophrenia and depression.
Using powerful supercomputers, the team examined vast chemical libraries containing millions of molecules to find optimal matches for the model. The shortlisted molecules predicted to bind to the receptor were subsequently tested in experiments conducted by colleagues at Karolinska Institutet. Remarkably, a substantial number of these molecules activated TAAR1, with one of the strongest candidates showing promising results in animal trials.
As the study reached its concluding phase, experimental structures for TAAR1 unexpectedly became accessible, allowing researchers to compare them with the AI-generated models.
“The precision of the AI-generated structures was astonishing—almost hard to believe. The findings indicate that AI modeling outperforms traditional methods significantly. We can now adopt this approach for receptors we previously thought were out of reach,” stated Jens Carlsson, who led Uppsala University’s involvement in the research.
