Using bioinformatics analysis to screen abnormal methylated differentially expressed hub genes of Kawasaki disease and construct diagnostic model

Objective: By using bioinformatics analysis, abnormal methylated differentially expressed genes (MDEGs) in Ka-wasaki disease (KD) were identified and a random forest diagnostic model for KD was established. Methods: The expression (GSE18606, GSE68004, GSE73461) and methylation (GSE109430) profiles was retrieved and download from Gene Expression Omnibus (GEO). We conducted enrichment analyses by using R software. In addition, we constructed a protein interaction network, and obtained 6 hub genes. We used expression profiles GSE100154 from GEO to verify the hub genes. Finally, we constructed a diagnostic model based on random forest. Results: We got a total of 55 MDEGs (43 hyper-methylated, low-expressing genes and 12 hypo-methylated, high -expressed genes). Six hub genes (CD2, IL2RB, IL7R, CD177, IL1RN, and MYL9) were identified by Cytoscape software. The area under curve (AUC) of the six hub genes was from 0.745 to 0.898, and the combined AUC was 0.967. The random forest diagnostic model showed that AUC was 0.901. Conclusion: The identification of 6 new hub genes improves our understanding of the molecular mechanism of KD, and the established model can be employed for accurate diagnosis and provide evidence for clinical diagnosis.

Automated summary

By using bioinformatics analysis, we identified abnormal methylated differentially expressed genes in Ka-wasaki disease and established a random forest diagnostic model. We identified 6 hub genes which improved our understanding of the molecular mechanism of KD. The constructed model can be used for accurate diagnosis and provide evidence for clinical diagnosis.