Identification and verification of anoikis-related genes immune infiltration characterization based on machine learning for epilepsy

Background: Epilepsy is a significant neurological disorder characterized by a complex etiology. Understanding the molecular mechanisms, notably those implicating the immune system and anoikis, is crucial for developing targeted therapies against epilepsy. Methods: Firstly, two epilepsy-related transcriptomic datasets (GSE143272 and GSE4290) were selected from the GEO database; then, thedifferentially expressed genes (DEGs) were identified. Focusing on anoikis in epilepsy, we collected anoikis-related genes (ARGs). Bioinformatics analysis and machine learning were employed for comprehensive analysis, and epilepticus mice were used to in vivo experimental validation of analytical results. Results: A total of 3,525 DEGs from GSE143272 and GSE4290 datasets were identified, and a total of 49 ARGs were obtained. The five key differentially expressed ARGs (DE-ARGs) were screened through machine learning analysis, including ANKRD13C, PIK3R1, BSG, CEACAM6, and BRMS1; These DE-ARGs emerged as potential biomarkers for epilepsy and were involved in various signaling pathways and immune cell activities, and results were further experimentally validated. Besides, the risk score model based on the DE-ARGs demonstrated high diagnostic efficiency; moreover, connectivity map database analysis suggested MPEP, LY-341495, and MDL-28170 as potential therapeutic agents. Conclusion: This study identified the five ARGs as potential therapeutic targets, highlighting the role of anoikis in epilepsy pathogenesis. Our result provides a novel insight into the molecular landscape of epilepsy and paves the way for further exploration and the development of more effective treatment strategies.