The datasets analyzed in this study.

Rheumatoid arthritis (RA) is a systemic autoimmune disease with persistent synovitis and joint destruction, leading to a huge economic and physical burden on patients. The detection of RA is important for the individual’s guiding therapeutic. However, current signatures lacked enough effects for the diagnosis of RA. Here, a pariwise signature, including genes ICAM2 and OSTF1, was derived based on a rank-based method, which was called ICAM2-OSTF1 signature (IOS). The sensitivity and specificity of IOS in the training dataset were 87.39% and 86.79%, respectively. The accuracy of IOS was 91.07% in the validation dataset that contained a total of 280 samples from two independent datasets. Besides, when using eight methods, such as ssGSEA, xCell and TIMER, to quantitate the immune infiltration characteristics in RA. We found that RA presented elevated pro-inflammation immune infiltration and immune score. In addition, transcriptome analysis demonstrated that the consistent transcriptional differences between RA and healthy control were significantly enriched in some pathways typically related to the immune microenvironments, such as T cell activation. Finally, network analysis demonstrated that ICAM2, CXCL16, CKLF and SLPI may be related to the occurrence of RA. In brief, IOS can individually distinguish RA from healthy controls measured by different laboratories, and be an auxiliary test for diagnosing RA.