AI-driven transformation of GPCR research: from structural biology to precision medicine

G protein-coupled receptors (GPCRs) regulate complex pathophysiological networks and represent a critical class of drug targets, but their structural determination and ligand screening are constrained by dependence on homology templates, static conformations, and a bias toward known ligands. Artificial intelligence is fundamentally transforming GPCR structural biology and pharmacology by overcoming these bottlenecks. The methodologies encompass high-precision static structure prediction, dynamic conformational ensemble modeling, ultra-large-scale virtual screening, de novo ligand design, allosteric mechanism decoding, biased signaling quantification, and systems pharmacology. Deep learning architectures enable end-to-end structure prediction exemplified by AlphaFold series, billion-compound virtual screening and unbiased de novo ligand design via diffusion models. Three-dimensional structure-aware Graph neural networks and molecular dynamics simulations are shifting research focus from single static conformations to transient conformations, cryptic allosteric sites, and ligand-induced biased signaling. Systems-level investigation through multi-omics-integrated GPCR digital twins advances personalized precision medicine. This review summarizes the innovative AI-driven paradigm, breaking traditional restrictions and greatly improving structural resolution of orphan receptors and GPCR complexes, thereby offering a comprehensive reference for dynamic GPCR pharmacological research. Core challenges include the black-box nature of deep learning, structural distribution biases, and the translational gap, which can be addressed by physics-informed models and high-throughput validation.