View Abstract

A2228 - Prioritizing Idiopathic Pulmonary Fibrosis Candidate Genes Based on “Guilt by Association” Analysis
Author Block: Y. Wang1, J. Yella2, S. K. Madala3, A. Jegga4; 1Biomedical Informatics, Cincinnati Children's Hospital, Cincinnati, OH, United States, 2Biomedical Informatics, Cincinnati Children's Hospital, CINCINNATI, OH, United States, 3Pulm Med, Cincinnati Children's Hosp, Cincinnati, OH, United States, 4Cincinnati Childrens Hospital Medical Center, Cincinnati, OH, United States.
Introduction: Both rare and common genetic variants are associated with idiopathic pulmonary fibrosis (IPF), a rare lung disease characterized by irreversible scarring of the lung parenchyma leading to dyspnea, progressive decline in lung function, and respiratory failure. Additionally, epigenetic changes and transcriptional changes are also associated with IPF risk and clinical phenotype. We aimed to identify and rank novel genes not previously considered as IPF candidate genes. Methods: Publicly available IPF gene expression signatures (6 independent cohorts - lung tissue samples from patients with IPF and controls) were used to identify differentially expressed genes (DEGs) in at least 3 out of 6 (50%) studies. The resulting conserved DEGs were further prioritized using “guilt by association” analyses. These include functional similarity-based approach and protein interactions network topology-based approach (ToppGene and ToppNet applications of the ToppGene Suite). Results: We identified a total of 197 genes that are upregulated and 84 genes downregulated in at least 3 out of six studies (1.5 Fold; p-value FDR 0.05). We then intersected these genes with a compiled list of all known genes (1338) associated with pulmonary fibrosis. This resulted in a total of 221 novel candidate genes that are differentially expressed in IPF and have not been previously reported to be related to IPF. Finally, we prioritized these 221 novel candidate genes by computational approaches based on guilt by association principle. Briefly, we analyzed if the 221 novel candidate genes interact with known IPF-related genes either functionally (using ToppGene application) or topologically (using ToppNet application). The top 5% of ToppGene-ranked genes included CBS, CCL13, COL7A1, COLEC11, EPAS1, FAP, FREM1, HLA-DOB, ITM2C, STEAP3, and TOP2A. Among the top 5% ToppNet-ranked genes were CLU, DIRAS3, EFNB3, ENC1, EPAS1, FHL2, KCNMA1, KRT5, LRRC32, TUBB3, and VIPR1. There was one gene common to both - EPAS1 (Endothelial PAS Domain Protein 1 or HIF-2-alpha; downregulated in IPF). Interestingly, a variant of EPAS1 is known to confer increased athletic performance in certain individuals and EPAS1 is referred to as “super athlete gene”. Mutations in this gene are associated with pulmonary hypotension and chronic mountain sickness. Further, EPAS1 protein levels are reported to be lower in human COPD lung tissue compared to non-disease control. Conclusion: Computational suite of tools based on the guilt by association principle when coupled with integrated heterogeneous genomic data sources are capable of identifying and ranking novel candidate genes for IPF.