GLASS2: An Updated Database for Comprehensive Experimentally Validated GPCR-ligand Associations

Area of Science:

• Pharmacology
• Bioinformatics
• Computational Biology

Background:

• G protein-coupled receptors (GPCRs) are crucial drug targets, but interaction data is fragmented.
• Existing resources lack comprehensive and standardized GPCR-ligand association information.

Purpose of the Study:

• To present GLASS2, an expanded and enhanced database of GPCR-ligand interactions.
• To provide task-ready datasets for AI-driven drug discovery.
• To consolidate fragmented GPCR-ligand data into a unified resource.

Main Methods:

• Integrated data from major pharmacological databases.
• Employed a large language model (LLM)-powered text-mining pipeline on PubMed literature.
• Developed task-ready classification and regression datasets with defined positive and negative samples.

Main Results:

• GLASS2 contains 1,715 receptors and 450,736 ligands, representing over one million unique interactions.
• Achieved a 72% increase in total associations compared to its predecessor.
• Provides structured datasets suitable for machine learning applications in drug discovery.

Conclusions:

• GLASS2 offers a comprehensive and standardized resource for GPCR-ligand interactions.
• The database facilitates AI-driven drug discovery by providing ready-to-use datasets.
• GLASS2 is accessible via a web portal with advanced search and integration capabilities.