RESUMO
Juvenile idiopathic arthritis (JIA) is common childhood rheumatic disease harming children health. However, there is still lack of effective biomarkers for diagnosis JIA at early onset. We aim to construct a classification model to predict JIA disease. The peripheral blood gene expression profile data of JIA were downloaded from GEO database. We compared and analyzed differentially expressed genes (DEGs) between different JIA samples through Pearson's correlation coefficient method and unsupervised clustering analysis. Diagnostic model were constructed based on the deviation pathway through bioinformatics method. Eighteen specific correlated DEGs were obtained, but the correlations altered in different disease states. Although most JIA and control samples were clustered by unsupervised clustering analysis, respectively, a few JIA samples could not be clustered well. Four co-expression networks were next constructed with gene connections dynamically altered under variable conditions. Eight signaling pathways were significantly enriched including B/T cell receptor, ErbB and MAPK signaling pathways. The deviation scores of pathways were calculated. Applying these eight signaling pathways as feature to construct a classification model could predict JIA disease with high accuracies. Our data provide some light into pathogenic mechanism of JIA, the specific gene sets and the related signaling pathways may be potential biomarkers for diagnosis or therapeutic targets of JIA.
