Machine learning-based metabolism-related genes signature and immune infiltration landscape in diabetic nephropathy

Abstract

BackgroundTo identify the diagnostic biomarkers of metabolism-related genes (MRGs), and investigate the association of the MRGs and immune infiltration landscape in diabetic nephropathy (DN).MethodsThe transcriptome matrix was downloaded from the GEO database. R package “limma” was utilized to identify the differential expressed MRGs (DE-MRGs) of HC and DN samples. Genetic Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of DE-MRGs were performed using “clusterProfiler” R package. WGCNA, LASSO, SVM-RFE, and RFE algorithms were employed to select the diagnostic feature biomarkers for DN. The ROC curve was used to evaluate discriminatory ability for diagnostic feature biomarkers. CIBERSORT algorithm was performed to investigate the fraction of the 22-types immune cells in HC and DN group. The correlation of diagnostic feature biomarkers and immune cells were performed via Spearman-rank correlation algorithm.ResultsA total of 449 DE-MRGs were identified in this study. GO and KEGG pathway enrichment analysis indicated that the DE-MRGs were mainly enriched in small molecules catabolic process, purine metabolism, and carbon metabolism. ADI1, PTGS2, DGKH, and POLR2B were identified as diagnostic feature biomarkers for DN via WGCNA, LASSO, SVM-RFE, and RFE algorithms. The result of CIBERSORT algorithm illustrated a remarkable difference of immune cells in HC and DN group, and the diagnostic feature biomarkers were closely associated with immune cells.ConclusionADI1, PTGS2, DGKH, and POLR2B were identified as diagnostic feature biomarkers for DN, and associated with the immune infiltration landscape, providing a novel perspective for the future research and clinical management for DN.