GRNIX: A Graph Neural Network Framework for Explainable Gene Regulatory Network Inference in Autoimmune Diseases Using XAI

GRNIX: A Graph Neural Network Framework for Explainable Gene Regulatory Network Inference in Autoimmune Diseases Using XAI

Autoimmune diseases arise from dysregulated immune mechanisms influenced by intricate gene regulatory networks (GRNs). Understanding these networks is crucial for uncovering disease mechanisms, predicting progression, and identifying therapeutic targets. However, traditional GRN inference methods, relying on statistical correlations or deterministic models, struggle to capture nonlinear interactions and often lack interpretability. Similarly, machine learning (ML)-based approaches, though powerful, are frequently black-box systems, limiting their utility in clinical applications.

To address these challenges, we present GRNIX, an innovative framework for GRN inference that integrates predictive accuracy with explainability. GRNIX leverages graph neural networks (GNNs) and incorporates multi-omics data while embedding biological and structural priors to enhance relevance. Using explainable artificial intelligence (XAI) techniques, GRNIX provides interpretable outputs, making it suitable for both research and clinical settings.

Our framework offers a novel approach to deciphering GRNs in autoimmune diseases, paving the way for more informed therapeutic strategies and improved disease management.