Identification of potential ferroptosis-related biomarkers and a pharmacological compound in diabetic retinopathy based on machine learning and molecular docking.

Background: Diabetic retinopathy (DR), a neurovascular disease, is a leading cause of visual loss worldwide and severely affects quality of life. Several studies have shown that ferroptosis plays an important role in the pathogenesis of DR; however, its molecule mechanism remains incompletely elucidated. Hence, this study aimed to investigate the pathogenesis of ferroptosis and explore potential ferroptosis-related gene biomarkers and a pharmacological compound for treating DR. Methods: Ferroptosis-related differentially expressed genes (DEGs) were identified in the GSE102485 dataset. Functional enrichment analyses were then performed and a protein-protein interaction (PPI) network was constructed to screen candidates of ferroptosis-related hub genes (FRHGs). FRHGs were further screened based on least absolute shrinkage and selection operator (LASSO) regression and random forest algorithms, and were then validated with the GSE60436 dataset and previous studies. A receiver operating characteristic (ROC) curve monofactor analysis was conducted to evaluate the diagnostic performance of the FRHGs, and immune infiltration analysis was performed. Moreover, the pharmacological compound targeting the FRHGs were verified by molecular docking. Finally, the FRHGs were validated using quantitative real-time polymerase chain reaction (qRT-PCR) analysis. Results: The 40 ferroptosis-related DEGs were extracted, and functional enrichment analyses mainly implicated apoptotic signaling, response to oxidative stress, ferroptosis, and lipid and atherosclerosis pathways. By integrating the PPI, LASSO regression, and random forest analyses to screen the FRHGs, and through validation, we identified five FRHGs that performed well in the diagnosis ( Conclusions: