Identifying effective diagnostic biomarkers for childhood cerebral malaria in Africa integrating coexpression analysis with machine learning algorithm

Abstract

Abstract Background Cerebral malaria (CM) is a manifestation of malaria caused by plasmodium infection. It has a high mortality rate and severe neurological sequelae, existing a significant research gap and requiring further study at the molecular level. Methods We downloaded the GSE117613 dataset from the Gene Expression Omnibus (GEO) database to determine the differentially expressed genes (DEGs) between the CM group and the control group. Weighted gene coexpression network analysis (WGCNA) was applied to select the module and hub genes most relevant to CM. The common genes of the key module and DEGs were selected to perform further analysis. The least absolute shrinkage and selection operator (LASSO) logistic regression and support vector machine recursive feature elimination (SVM-RFE) were applied to screen and verify the diagnostic markers of CM. Eventually, the hub genes were validated in the external dataset. Gene set enrichment analysis (GSEA) was applied to investigate the possible roles of the hub genes. Results The GO and KEGG results showed that DEGs were enriched in some neutrophil-mediated pathways and associated with some lumen structures. Combining LASSO and the SVM-RFE algorithms, LEF1 and IRAK3 were identified as potential hub genes in CM. Through the GSEA enrichment results, we found that LEF1 and IRAK3 participated in maintaining the integrity of the blood–brain barrier (BBB), which contributed to improving the prognosis of CM. Conclusions This study may help illustrate the pathophysiology of CM at the molecular level. LEF1 and IRAK3 can be used as diagnostic biomarkers, providing new insight into the diagnosis and prognosis prediction in pediatric CM.

Keywords

• Cerebral malaria
• WGCNA
• Machine learning
• Neutrophil
• Blood–brain barrier (BBB)