BOLD-GPCRs: A Transformer-Powered App for Predicting Ligand Bioactivity and Mutational Effects across Class A GPCRs

Area of Science:

• Pharmacology
• Computational Biology
• Drug Discovery

Background:

• G Protein-Coupled Receptors (GPCRs) are crucial drug targets involved in numerous diseases.
• Traditional drug discovery methods struggle with GPCRs lacking extensive ligand or structural data.
• Advanced strategies are needed for accurate ligand bioactivity prediction across the GPCR family, particularly for understudied subtypes.

Purpose of the Study:

• To introduce BOLD-GPCRs, a deep learning framework for enhanced ligand bioactivity prediction in class A GPCRs.
• To provide a user-friendly web interface for accessing the BOLD-GPCRs framework.
• To leverage transfer learning and curated datasets for improved GPCR drug discovery.

Main Methods:

• Developed BOLD-GPCRs (BERT-Optimized Ligand Discovery for GPCRs) using deep learning.
• Integrated dense neural network classifiers with transformer-based protein language models.
• Utilized curated datasets of class A GPCR ligands, receptor sequences, and mutations.

Main Results:

• BOLD-GPCRs demonstrates robust predictive performance for ligand bioactivity across class A GPCRs.
• The framework accurately predicts the effects of receptor mutations on ligand activity.
• Achieved strong results, especially for poorly characterized GPCR subtypes.

Conclusions:

• BOLD-GPCRs is a valuable tool for enhancing GPCR ligand discovery.
• The framework shows significant potential for identifying therapeutics targeting understudied GPCRs.
• Deep learning approaches offer advanced solutions for GPCR-targeted drug development.