Variational Approaches for Drug-Disease-Gene Links in Periodontal Inflammation.

Introduction And Objectives: Oral diseases, including gingivitis and periodontitis, are linked to the Wnt signaling pathway, vital for bone metabolism, cementum homeostasis, and mesenchymal stem cell differentiation. Advances in generative AI techniques, such as variational autoencoders (VAEs) and quantum variational classifiers (QVCs), offer promising tools for predicting gene associations between drugs and diseases. This study aims to compare the predictive performance of VAEs and QVCs in modeling drug-disease gene networks within the Wnt signaling pathway in periodontal inflammation.

Methods: Genes associated with Wnt-related periodontal inflammation were identified through comprehensive literature reviews and genomic databases. Their roles in various biological processes were evaluated using gene set enrichment analysis, employing tools like Enrichr, which integrates diverse gene sets from sources such as DSigDB, DisGeNET, and Lincs_l1000.drug. The study then applied VAEs and QVCs to predict gene-disease associations related to the Wnt signaling pathway.

Results: The analysis revealed an extensive network comprising 1738 nodes and 1498 edges, averaging 1.992 neighbors per node. The network exhibited a diameter of 2, a radius of 1, and a characteristic path length of 1.992, indicating limited interconnectivity. The VQA model demonstrated a high accuracy rate of 97.5%, although it only detected 50% of anomalies. The VQC model achieved a precision of 78%, with Class 1 samples showing improved recall and a balanced F1 score.

Conclusion: VQC and VAE models exhibit strong potential for discovering FDA-approved drugs by predicting gene-drug associations in periodontitis based on the Wnt signaling pathway.

Clinical Relevance: This study highlights the potential of VAEs and QVCs in predicting gene-drug associations for periodontal inflammation. This could lead to more targeted therapies for oral diseases like periodontitis, improving patient outcomes and advancing personalized treatment strategies in clinical practice.