M3NetFlow: a novel multi-scale multi-hop multi-omics graph AI model for omics data integration and interpretation

Abstract

Abstract The integration and interpretation of multi-omics data play a crucial role in systems biology for prioritizing vital molecular targets and deciphering core signaling pathways of complex diseases, such as cancer, COVID-19 and Alzheimer’s disease. However, it remains a challenge that has not been adequately addressed. Graph neural networks (GNN) have emerged as powerful artificial intelligence models for analyzing data with a graphical structure. Nevertheless, GNN models have not been sufficiently designed for integrative and interpretable multi-omics data analysis. In this study, we propose a novel multi-scale multi-hop multi-omics GNN model, M3NetFlow , to integrate and interpret multi-omics data to rank key targets and infer core signaling pathways. Specifically, we applied the M3NetFlow model to infer cell-line-specific core signaling networks explaining drug combination response. The evaluation and comparison results on drug combination prediction showed that the M3NetFlow model achieved significantly higher prediction accuracy than existing GNN models. Furthermore, M3NetFlow can predict key targets and infer essential signaling networks regulating drug combination response. It is critical for guiding the development of personalized precision medicine for patients with drug resistance. This model can be applied to general multi-omics data-driven research. Aside from that, we developed the visualization tool, NetFlowVis , the better analysis of targets and signaling pathways of drugs and drug combinations.