A Network and Pathway Analysis of Genes Associated With Atrial Fibrillation

Abstract

Background: Atrial fibrillation (AF) is affected by both environmental and genetic factors. Previous genetic association studies, especially genome-wide association studies, revealed a large group of AF-associated genes. However, little is known about the functions and interactions of these genes. Moreover, established genetic variants of AF contribute modestly to AF variance, implying that numerous additional AF-associated genetic variations need to be identified. Hence, a systematic network and pathway analysis is needed. Methods: We retrieved all AF-associated genes from genetic association studies in various databases and performed integrative analyses including pathway enrichment analysis, pathway crosstalk analysis, network analysis, and microarray meta-analysis. Results: We collected 254 AF-associated genes from genetic association studies in various databases. Pathway enrichment analysis revealed the top biological pathways that were enriched in the AF-associated genes related to cardiac electromechanical activity. Pathway crosstalk analysis showed that numerous neuro-endocrine-immune pathways connected AF with various diseases including cancers, inflammatory diseases, and cardiovascular diseases. Furthermore, an AF-specific subnetwork was constructed with the prize-collecting Steiner forest algorithm based on the AF-associated genes, and 24 novel genes that were potentially associated with AF were inferred by the subnetwork. In the microarray meta-analysis, six of the 24 novel genes (APLP1, CREB1, CREBBP, PRMT1, IRAK1, and PLXND1) were expressed differentially in patients with AF and sinus rhythm. Conclusions: AF is not only an isolated disease with abnormal electrophysiological activity but might also share a common genetic basis and biological process with tumors and inflammatory diseases as well as cardiovascular diseases. Moreover, the six novel genes inferred from network analysis might help detect the missing AF risk loci.

Keywords: atrial fibrillation; genetic association study; microarray meta-analysis; network and pathway analysis.

Figure 1

Flow diagram of the integrative analysis. We retrieved all genes associated with atrial fibrillation (AF) from various databases. These genes were reviewed by the authors to exclude insignificant genes. The filtered genes were used for pathway analysis (pathway enrichment analysis and pathway crosstalk analysis) and to construct the AF-specific subnetwork under the human interactome for the inference of potential AF-associated genes. These potential genes were evaluated by means of microarray meta-analysis (prize nodes, the known AF-associated genes. Hidden nodes, the potential AF-associated genes inferred by the known AF-associated genes in the AF-specific subnetwork. PCSF, prize-collecting Steiner forest).

Figure 2

Pathway crosstalk of AFgenes. In this figure, each node represents a significant pathway, and each edge represents pathway crosstalk. The size of each node is approximately proportional to the number of AF-associated genes in the corresponding pathway. The width of each edge is approximately proportional to the kappa score.

Figure 3

An analysis of various parameter sets when running Forest on the AFgenes. A good choice of parameter set is indicated by the red circle.

Figure 4

AF-specific protein–protein interaction subnetwork constructed by the prize-collecting Steiner forest algorithm, containing 272 nodes and 271 edges. Blue circular vertices, genes of AFgenes (the known AF-associated genes); red circular vertices, expanding genes (the potential AF-associated genes).

Figure 5

Main biological pathways associated with the AFgenes. AF might share a common genetic basis and biological process with tumors and inflammatory diseases in addition to cardiovascular diseases. Numerous neuro-endocrine-immune pathways connect these diseases with AF.