Using Bioinformatics and Machine Learning to Predict the Genetic Characteristics of Ferroptosis-Cuproptosis-Related Genes Associated with Sleep Deprivation

Liang Wang,1,2,* Shuo Wang,3,* Chujiao Tian,3 Tao Zou,3 Yunshan Zhao,2 Shaodan Li,3 Minghui Yang,3 Ningli Chai1 1Department of Gastroenterology, First Medical Center, Chinese PLA General Hospital, Beijing, 100853, People’s Republic of China; 2Health Medicine Department, the 955th Hospital of the Army, Changdu, Tibet, 854000, People’s Republic of China; 3Department of TCM, Sixth Medical Center, Chinese PLA General Hospital, Beijing, 100048, People’s Republic of China*These authors contributed equally to this workCorrespondence: Shaodan Li, Department of TCM, Sixth Medical Center, Chinese PLA General Hospital, Beijing, 100048, People’s Republic of China, Email [email protected] Ningli Chai, Department of Gastroenterology, First Medical Center, Chinese PLA General Hospital, Beijing, 100853, People’s Republic of China, Email [email protected]: Sleep deprivation (SD), a common sleep disease in clinic, has certain risks, and its pathogenesis is still unclear. This study aimed to identify ferroptosis-cuproptosis-related genes (FCRGs) associated with SD through bioinformatics and machine learning, thus elucidating their biological significance and clinical value.Methods: SD-DEGs were obtained from GEO. We intersected key WGCNA module genes of DE-FCRGs with SD-DEGs to obtain SD-DE-FCRGs. GO and KEGG analyses were performed. Machine learning was used to screen SD-DE-FCRGs, and filtered genes were intersected to obtain SD characteristic genes. ROC curves were used to evaluate the accuracy of SD characteristic genes. CIBERSORT was used to analyze the correlation between SD-DE-FCRGs and immune cells. We constructed a ceRNA network of SD-DE-FCRGs and used DGIbd to predict gene drug targets.Results: The 156 DEGs were identified from GSE98566. Five SD-DE-FCRGs from DE- FCRGs and SD-DEGs were analyzed via WGCNA, and enrichment analysis involved mainly ribosome regulation, mitochondrial pathways, and neurodegenerative diseases. Machine learning was used to obtain Four SD-DE-FCRGs (IKZF1, JCHAIN, MGST3, and UQCR11), and these gene analyses accurately evaluated the distribution model (AUC=0.793). Immune infiltration revealed that SD hub genes were correlated with most immune cells. Unsupervised cluster analysis revealed significant differential expression of immune-related genes between two subtypes. GSVA and GSEA revealed that enriched biological functions included oxidative phosphorylation, ribonucleic acid, metabolic diseases, activation of oxidative phosphorylation, and other pathways. Four SD-DE-FCRGs associated with 29 miRNAs were identified via the construction of a ceRNA network. The important target lenalidomide of IKZF1 was predicted.Conclusion: We first used bioinformatics and machine learning to screen four SD-DE-FCRGs. These genes may affect the involvement of infiltrating immune cells in pathogenesis of SD by regulating FCRGs. We predicted that lenalidomide may target IKZF1 from SD-DE-FCRGs. Keywords: sleep deprivation, ferroptosis, cuproptosis, bioinformatics, machine learning, immune infiltration