Abstract
INTRODUCTION: Type 2 diabetes (T2D) is a complex metabolic disorder with significant global health implications. Understanding the molecular mechanisms underlying T2D is crucial for developing effective therapeutic strategies. This study employs single-cell RNA sequencing (scRNA-seq) and machine learning to explore the the pathogenesis of T2D, with a particular focus on immune cell infiltration. METHODS: We analyzed scRNA-seq data from islet cells of T2D and nondiabetic (ND) patients, identifying differentially expressed genes (DEGs), especially those related to metal ion transport (RMITRGs). We employed 12 machine learning algorithms to develop predictive models and assessed immune cell infiltration using single-sample gene set enrichment analysis (ssGSEA). Correlations between immune cells and key RMITRGs were investigated, and the interactions among these genes were explored through protein-protein interaction (PPI) network analysis. Additionally, we performed a detailed cell-cell communication analysis to identify significant signaling pathways in T2D. RESULTS: Our analysis identified 1953 DEGs between T2D and ND patients, with the Stepglm[backward] plus GBM model demonstrating high predictive accuracy and identifying 13 hub RMITRGs. Twelve protein structures were predicted using AlphaFold 3, revealing potential functional conformations. We observed a strong correlation between hub RMITRGs and immune cells, and PPI network analysis revealed key interactions. Cell-cell communication analysis highlighted 16 active signaling pathways, with CXCL, MIF, and COMPLEMENT linked to immune and inflammatory responses, and WNT, KIT, LIFR, and HGF pathways uniquely activated in T2D. CONCLUSION: Our analysis identified genes crucial for T2D, emphasizing ion transport, signaling, and immune cell interactions. These findings suggest therapeutic potential to enhance T2D management. The identified pathways and genes provide valuable insights into the disease mechanisms and potential targets for intervention.