Bioinformatics analysis and machine learning identification of ferroptosis-related biomarkers in epilepsy

Background: Epilepsy is one of the common nervous system diseases, and early diagnosis has an important impact on prognosis. Ferroptosis plays an important role in the pathogenesis of epilepsy. Using bioinformatics analysis and machine learning, we aimed to identify hub genes associated with ferroptosis and provide therapeutic targets for epilepsy. Methods: Through bioinformatics analysis of datasets GSE49030 and GSE88992, differentially expressed genes(DEGs) related to ferroptosis in epilepsy were identified. Functional enrichment of ferroptosis-related DEGs was performed, and a protein interaction network was used to identify the top 30 node genes. LASSO, SVM-RFE, and random forest machine learning algorithms were employed to screen for hub genes. We analyzed the infiltration of immune cells in epilepsy and the control group and analyzed the correlation between hub genes and immune cells. Finally, the value of hub genes in the diagnosis of epilepsy was evaluated in the training set (GSE49030 and GSE88992) and validation set (GSE16969). Results: GSE49030 and GSE88992 obtained a total of 1550 DEGs, including 55 ferroptosis-related DEGs. Enrichment analysis showed that ferroptosis, oxidative stress, and inflammatory signaling pathways were enriched. Jun and Wwtr1 were screened as hub genes. Immunological analysis showed that 9 kinds of cells were differentially expressed between the epilepsy group and control group. Two hub genes were mainly associated with NK cell activation. Jun and Wwtr1 were up-regulated in epilepsy, with AUROC of 0.978 and 0.973 for the diagnosis of epilepsy, respectively, and were verified in dataset DSE16969. Conclusion: Our results suggest that Jun and Wwtr1 are potential biomarkers of epilepsy and NK cells ferroptosis can influence both the onset and progression of epilepsy. These results provide more evidence for the role of ferroptosis in the pathogenesis of epilepsy.