Artificial Intelligence: A New Tool for Structure-Based G Protein-Coupled Receptor Drug Discovery

Area of Science:

• Computational biology
• Artificial intelligence
• Drug discovery

Background:

• Traditional experimental methods (X-ray crystallography, NMR, cryo-EM) for protein structure determination are resource-intensive and time-consuming.
• Recent AI advancements like AlphaFold and RoseTTAFold offer rapid and accurate protein structure predictions from amino acid sequences.

Purpose of the Study:

• To review the latest AI developments in protein structure prediction.
• To assess the potential of AI approaches in structure-based drug discovery, focusing on GPCRs.
• To identify limitations of current AI methods for drug discovery applications.

Main Methods:

• Review of recent literature on AI-driven protein structure prediction.
• Analysis of AI model performance in predicting protein structures.
• Evaluation of AI's accuracy for specific drug discovery tasks, such as ligand docking.

Main Results:

• AI models demonstrate high accuracy in predicting overall protein structures.
• Essential details for ligand docking, like side-chain positioning in binding pockets, are not yet accurately predicted by AI.
• Current docking methodologies generate numerous false positives, limiting their precision.

Conclusions:

• AI significantly accelerates protein structure prediction but requires further refinement for precise drug discovery applications.
• The accuracy of AI predictions for critical drug target details needs improvement.
• AI's role in structure-based drug discovery, particularly for GPCRs, is promising but not yet fully realized.