Identifying potential biomarkers and molecular mechanisms of postmenopausal osteoporosis using weighted coexpression analysis and multiple machine learning modeling

Abstract

Abstract Objective: The goal of this study was to discover key genes associated with postmenopausal osteoporosis (PMOP), and evaluate their roles in disease development and prognosis. Methods: Microarray analysis of GSE56815 was conducted to identify differentially expressed genes (DEGs) between PMOP patients and normal individuals. A combined analysis was performed involving upregulated DEGs and genes within the weighted coexpression analysis (WCGNA) modules. This was done to determine key genes with a high association with PMOP and showing increased expression at the transcriptional level. Functional enrichment analysis and protein‒protein interaction (PPI) network analysis were implemented on these key genes. LASSO regression was utilized to minimize the dimensionality of these genes. Multiple machine learning models were established on the training set, and the random forest model with the best performance was chosen to detect potential biomarkers for PMOP. Results: Three key genes, MX2, IFI35, and SERPINB2, were identified through our analysis. These genes occupy a central position in the regulatory network and showed excellent performance in the machine learning models. The area under the curve (AUC) of the validation set implied that these identified key genes possess strong predictive power and applicability. The genes were found to be mainly enriched in the B-cell receptor signaling pathway and osteoclast differentiation. Conclusion: This study provides significant insights into the mechanisms underlying the development and progression of osteoporosis. The identification of potential biomarkers may aid in early diagnosis, prevention, and treatment of osteoporosis, and serves as a foundation for the development of more effective therapeutic strategies and preventive measures.