Bioinformatics analysis to screen the key genes in pediatric Chronic Active Epstein-Barr Virus Infection

Chronic active EBV infection (CAEBV) is associated with poor prognosis and high mortality. We performed bioinformatics analysis to screen out key genes associated with CAEBV. Weighted gene co-expression network analysis (WGCNA) was used to identify the gene module which was most correlated with pediatric CAEBV. Furthermore, the differentially expressed genes (DEGs) between pediatric acute infectious mononucleosis (AIM) and pediatric CAEBV were investigated. Least absolute shrinkage and selection operator (LASSO) and random forest then were performed to identify the key variables associated with pediatric CAEBV. We also explored the correlation between these hub genes with EBV infection related pathway and immune cell abundance. Compared with pediatric AIM, 1561 DEGs were up-regulated in pediatric CAEBV, and these genes were mainly enriched in inflammatory response and inflammation-related pathways. WGCNA analysis showed that genes in blue module were mostly related to pediatric CAEBV. Genes in the blue module and DEGs are intersected to get 174 genes and these genes are also enriched in inflammatory response-related pathways. The key CAEBV-related genes were selected from these 174 genes by applying the random Forest and LASSO algorithm, resulting in TPST1, TNFSF8 and RAB3GAP1. These three genes showed good diagnostic performance in distinguishing pediatric CAEBV from pediatric AIM. Furthermore, Cibersort and GSEA analysis indicated that these three genes were positively correlated with myeloid cell enrichment and persistent EBV infection pathway, respectively. Our finding systematically analyzed the difference between AIM and CAEBV and identified TPST1, TNFSF8 and RAB3GAP1 were the key genes in the development of CAEBV.