Analyzing the Differential Expression of Vitiligo Genes by Bioinformatics Methods.

Abstract

Background: Vitiligo is a hypopigmentation skin disease that is easy to diagnose but difficult to treat. The etiology of vitiligo is unknown, which may be related to genetic and immune factors. Objective: To provide potential targets for the treatment of vitiligo through identifying signature genes based on an artificial neural network (ANN) model. Methods: We downloaded two publicly available datasets from GEO database and identified DEGs. We trained the random forest and ANN algorithm using training set GSE75819 to further identify new gene features and predicted the possibility of vitiligo. In addition, we further validated the performance of our model through the test set GSE53148 and verified the diagnostic value of our model with the validation set GSE53148. Finally, we used RT-qPCR to compare the expression of two genes randomly selected in this study in patients with vitiligo and healthy people. Results: Two genes were randomly selected from the 30 key genes identified by ANN and validated through RT-qPCR in 6 vitiligo patients. The results showed that compared with the control group, the mRNA expression of FLJ21901 in the disease group was significantly upregulated, and the mRNA expression of MAST1 was significantly downregulated, with statistical significance. Conclusions: Through the identification of characteristic genes and the construction of a neural network model, it was found that the differentially expressed genes can provide a new potential target for the treatment of vitiligo.