Background: Atrial fibrillation (AF), one of the most prevalent heart rhythm disorders, may lead to thromboembolism, heart failure, and sudden death. However, the mechanism of AF has not yet been fully explained. Objective: This study aims to identify novel gene signatures and to investigate the potential therapeutic targets of AF with an integrated bioinformatic approach. Methods: The gene expression and methylation datasets of AF were obtained through the Gene Expression Omnibus (GEO) database. Subsequently, a set of differentially expressed genes and differential methylation sites were identified. Gene functional annotation analysis was conducted to explore the potential function of differentially-methylated/expressed genes. Then, we constructed a PPI network and TF–miRNA–mRNA network. Finally, weighted gene co-expression network analysis (WGCNA) was presented to study critical modules of AF. Results: Seven hypomethylated-high expression genes and nine hypermethylated-low expression genes were acquired from AF patients. Functional enrichment results indicated that the differentially-methylated/expressed genes were mainly concentrated in decidualization, maternal placenta development, regulation of nitric-oxide synthase activity, and osteoclast differentiation. Based on the results of the PPI, we defined 4 key genes namely FHL2, STC2, ALPK3, and RAP1GAP2 as the core genes, playing essential roles in the TF-miRNA-mRNA network. In the end, we constructed two co-expression modules that highly correlated with AF-related phenotype. Conclusion: In our study, we found critical genes for AF that might help understand the molecular changes in AF.