Predicting associations between drugs and G protein-coupled receptors using a multi-graph convolutional network

Area of Science:

• Pharmacology and Cheminformatics
• Computational Biology
• Drug Discovery

Background:

• Drug development is costly and time-consuming, with safety concerns.
• Drug repurposing offers a faster, more economical alternative by finding new uses for existing drugs.
• G protein-coupled receptors (GPCRs) are a major drug target class, making them crucial for drug repurposing strategies.

Purpose of the Study:

• To develop an advanced computational model for predicting novel drug-GPCR interactions.
• To overcome limitations of existing methods that fail to integrate multiple data types.
• To accelerate the drug repurposing process through precise interaction prediction.

Main Methods:

• Development of an end-to-end deep learning model utilizing a multi-graph convolutional network (MGCN).
• Integration of multi-source data, including drug structure, drug-drug interactions, GPCR sequences, and subfamily information.
• Comparative analysis against existing deep learning and non-deep learning models.

Main Results:

• The proposed MGCN model demonstrated superior performance in inferring drug-GPCR relationships compared to existing methods.
• The model successfully integrated diverse data sources for enhanced prediction accuracy.
• Multi-source data integration proved crucial for advancing drug-GPCR relationship detection.

Conclusions:

• The developed multi-graph convolutional network model offers an efficient and precise approach for drug repurposing.
• Integrating multi-source data significantly improves the prediction of novel drug-GPCR associations.
• This computational strategy holds promise for accelerating the identification of new therapeutic applications for existing drugs.