Integrated Analysis of Single-Cell RNA-Seq and Bulk RNA-Seq Combined with Multiple Machine Learning Identified a Novel Immune Signature in Diabetic Nephropathy

Abstract

Yue-Ling Peng,1 Yan Zhang,1 Lin Pang,2 Ya-Fang Dong,3 Mu-Ye Li,4 Hui Liao,5 Rong-Shan Li1 1Department of Nephrology, Shanxi Provincial People’s Hospital (Fifth Hospital of Shanxi Medical University), Taiyuan, People’s Republic of China; 2Department of Health Statistics, School of Public Health, Shanxi Medical University, Taiyuan, People’s Republic of China; 3Department of Pathology and Pathophysiology, School of Basic Medicine, Shanxi Medical University, Taiyuan, People’s Republic of China; 4Department of Ocular Fundus Diseases, Shanxi Eye Hospital, Shanxi Medical University, Taiyuan, People’s Republic of China; 5Drug Clinical Trial Institution, Shanxi Provincial People’s Hospital (Fifth Hospital of Shanxi Medical University), Taiyuan, People’s Republic of ChinaCorrespondence: Rong-Shan Li, Department of Nephrology, Shanxi Provincial People’s Hospital (Fifth Hospital of Shanxi Medical University), Taiyuan, People’s Republic of China, Email [email protected]: Increasing evidence suggests that immune modulation contributes to the pathogenesis and progression of diabetic nephropathy (DN). However, the role of immune modulation in DN has not been elucidated. The purpose of this study was to search for potential immune-related therapeutic targets and molecular mechanisms of DN.Methods: Gene expression datasets were obtained from the Gene Expression Omnibus (GEO) database. A total of 1793 immune-related genes were acquired from the Immunology Database and Analysis Portal (ImmPort). Weighted gene co-expression network analysis (WGCNA) was performed for GSE142025, and the red and turquoise co-expression modules were found to be key for DN progression. We utilized four machine learning algorithms, namely, random forest (RF), support vector machine (SVM), adaptive boosting (AdaBoost), and k-nearest neighbor (KNN), to evaluate the diagnostic value of hub genes. Immune infiltration patterns were analyzed using the CIBERSORT algorithm, and the correlation between immune cell type abundance and hub gene expression was also investigated.Results: A total of 77 immune-related genes of advanced DN were selected for subsequent analyzes. Functional enrichment analysis showed that the regulation of cytokine–cytokine receptor interactions and immune cell function play a corresponding role in the progression of DN. The final 10 hub genes were identified through multiple datasets. In addition, the expression levels of the identified hub genes were corroborated through a rat model. The RF model exhibited the highest AUC. CIBERSORT analysis and single-cell sequencing analysis revealed changes in immune infiltration patterns between control subjects and DN patients. Several potential drugs to reverse the altered hub genes were identified through the Drug-Gene Interaction database (DGIdb).Conclusion: This pioneering work provided a novel immunological perspective on the progression of DN, identifying key immune-related genes and potential drug targets, thus stimulating future mechanistic research and therapeutic target identification for DN.Keywords: diabetic nephropathy, machine learning, single cell, cell-to-cell communication, immune therapy

Keywords

• diabetic nephropathy
• machine learning
• single cell
• cell-to-cell communication
• immune therapy