Identification and validation of two hub genes involved in membranous nephropathy based on machine learning

Abstract

Abstract Background: Membranous nephropathy (MN) is an autoimmune disease. It is an important cause of end-stage renal disease in primary glomerulonephritis. Significant breakthroughs in its diagnosis have been made in previous studies, however, the pathogenesis of MN has still remained elusive. In recent years, bioinformatics has provided new research strategies to investigate the mechanisms of kidney disease. This study aimed to explore potential biomarkers of MN through bioinformatics analysis. Methods: Differentially expressed genes (DEGs) were identified by performing a differential expression analysis with the "limma" R package, and then, the weighted gene co-expression network analysis (WGCNA) was applied to obtain the most MN-related genes. After intersecting these genes, the least absolute shrinkage and selection operator (LASSO) and the support vector machine-recursive feature elimination (SVM-RFE) algorithms were utilized to identify hub genes. To assess the diagnostic value of hub genes, the receiver operating characteristic (ROC) curve analysis was performed. Finally, the relationship between hub genes and the immune microenvironment (IME) was analyzed. Results: The differential expression analysis yielded 1,466 DEGs, and using the WGCNA, 442 genes, which were the most MN-related genes, were obtained. From the intersection of these genes, 130 genes were identified. Subsequently, two hub genes (ECM1 and ATP8B1) were detected by the LASSO and SVM-REF algorithms. It was found that they were associated with components of the IME (natural killer T cells, gamma delta T cells, macrophages, etc.). Conclusion: Two hub genes (ECM1 and ATP8B1) were identified by machine learning, and their diagnostic value was evaluated. It was revealed that these two genes were associated with the components of the IME. Our findings may provide new ideas for developing new biomarkers for MN.