Background Genome-wide association studies (GWAS) and subsequent follow up studies validated a significant number of ankylosing spondylitis (AS) loci. It is often speculated that risk loci detected from autoimmune diseases converges into shared pathways through molecular level interactions. To identify such convergence a systematic approach is required to further investigate and understand the underlying disease biology.
Objectives The primary goal of this study is to show systematic molecular level convergence of AS risk genes with nine other autoimmune diseases.
Methods Here we investigate 418 risk genes from ten complex autoimmune diseases (ankylosing spondylitis (AS), crohn's disease (CD), psoriasis (PS), ulcerative colitis (UC), celiac disease (CeD), multiple sclerosis (MS), primary biliary cirrhosis (PBC), rheumatoid arthritis (RA), type 1 diabetes (T1D), systemic lupus erythematous (SLE)) and their complex interactions with AS. We have used a network based method to infer connectivity between AS genes found to be associated in GWAS studies with another autoimmune risk gene set. The network is consisting of 6,998,947 pairwise interactions (symmetric gene pair interaction excluded) that is a set derived from physical protein-protein interactions (PPI), co-expression or co-localization analysis. For a given pair of disease gene set, the constructed interaction network consists of one degree connectivity for any pair of gene. Exhaustive permutation test was conducted by randomly replacing equal number of genes (50000 times, from a background of 22,695 protein coding genes) for a disease risk gene set to quantify the connectivity significance of the original risk gene sets.
Results The constructed PPI network showed significantly (P<1.8x10–4) dense modular connectivity for AS genes with six autoimmune disease genes (CD, PS, UC, CeD, MS, PBC). The most significant connectivity observed (after correcting for gene number) with Crohn's disease and ulcerative colitis that is consistent with the known biology of AS where CD and UC often manifest as comorbid conditions. To better understand the shared pathway and to identify etiological genes, each gene was ranked according to their contributing connectivity of the inferred network. Apart from the shared genes (associated in multiple diseases, i.e. IL23R, TNFAIP3) we have also identified AS specific genes (i.e. NOS2, GPR65) that are highly connected with multiple non-AS autoimmune risk genes.
Conclusions We have developed a systematic approach to infer causal genes that converge at the molecular level to a common pathway involve in disease pathogenesis. The highly connected non-AS genes identified in our study should be prioritized for further investigation to identify their contribution to AS pathogenesis and to better understand the complex relationship of autoimmune diseases.
Disclosure of Interest None declared