Graph neural networks-based prediction of drug gene association of P2X receptors in periodontal pain

Abstract

The P2X7 receptor, a member of the P2X receptor family, plays a crucial role in various physiological processes, particularly pain perception. Its expression across immune, neuronal, and glial cells facilitates the release of pro-inflammatory molecules, thereby influencing pain development and maintenance, as evidenced by its association with pulpitis in rats. Notably, P2X receptors such as P2X3 and P2X7 are pivotal in dental pain pathways, making them promising targets for novel analgesic interventions. Leveraging graph neural networks (GNNs) presents an innovative approach to model graph data, aiding in the identification of drug targets and prediction of their efficacy, complementing advancements in genomics and proteomics for therapeutic development. In this study, 921 drug-gene interactions involving P2X receptors were accessed through https://www.probes-drugs.org/. These interactions underwent meticulous annotation, preprocessing, and subsequent utilization to train and assess GNNs. Furthermore, leveraging Cytoscape, the CytoHubba plugin, and other bioinformatics tools, gene expression networks were constructed to pinpoint hub genes within these interactions. Through analysis, SLC6A3, SLC6A2, FGF1, GRK2, and PLA2G2A were identified as central hub genes within the context of P2X receptor-mediated drug-gene interactions. Despite achieving a 65 percent accuracy rate, the GNN model demonstrated suboptimal predictive power for gene-drug interactions associated with oral pain. Hence, further refinements and enhancements are imperative to unlock its full potential in elucidating and targeting pathways underlying oral pain mechanisms.