Multi-dataset identification of innovative feature genes and molecular mechanisms in keratoconus.

This study aimed to identify feature genes and explore the molecular mechanisms of keratoconus (KC). We downloaded data files from NCBI GEO public database. The Limma package was used for differential expression analysis of gene profiles. Lasso regression was used to identify the feature genes. The CIBERSORT algorithm was used to infer the proportion of immune-infiltrating cells and analyse the correlation between gene expression levels and immune cells. Related transcription factors and miRNAs of key genes were predicted using the Cistrome DB and Mircode databases. Analysis of expression differences in disease genes was based on the GeneCards database. The CMap was used to analyse targeted therapeutic drugs. IHC was performed to verify the expression levels of ATOH7 and MYRF in corneas. Exactly 593 upregulated and 473 downregulated genes were identified. Lasso regression analysis identified ATOH7, DBNDD1, RNF217-AS1, ARL11, MYRF and SNORA74B as feature genes for KC. All key genes were correlated with immune infiltration and the levels of activated memory CD4+ T cells and plasma cells were significantly increased. miRNA, IRF and STAT families were correlated to feature genes. The expression levels of key genes were significantly correlated to KC-related genes. Entinostat, ochratoxin-a, diphencyprone and GSK-3-inhibitor-II were predicted as potential KC medications. The expression of MYRF was significantly higher in the KC samples, contrary to the expression of ATOH7. KC is related to both immune infiltration and genetic factors. MYRF and ATOH7 were newly identified and verified feature genes of KC.