Abstract:N6-methyladenosine (m(exp 6)A) is the most abundant RNA modification on mRNA and lncRNA in human. Recent studies have shown that it is implicated in various critical biological processes, such as translation and alternative splicing, and involves in multiple human diseases, including cancer and obesity. However, only a small number of RNA methylation sites have been explicitly associated to disease conditions with experimental approaches, since RNA m(exp 6)A methylation sites may potentially play a pivotal regulatory role in a wide range of human pathogenesis and should not be ignored, an efficient predictor for disease-associated m(exp 6)A RNA methylation sites becomes a major challenge. In order to obtain a comprehensive understanding of disease-associated m(exp 6)A RNA methylation site, we purpose here a computational framework to integrate the three different layers of a network structure, including the expression profiles of genes, the methylation profiles of m(exp 6)A RNA methylation sites and gene-disease associations, and utilize subsequently the Hypergenomic test to predict integrally the potential disease-associated m(exp 6)A sites. We show with a rigorous cross-validation that the AUROC of the proposed approach is 0.73; and a number of predictions are supported by existing literatures, suggesting our prediction is helpful for identifying the novel m(exp 6)A sites associated to human disease. Ultimately, the predicted results are freely available online at: http://180.208.58.19/DRRMSDB/, which supports the queries of diseaserelated m(exp 6)A RNA methylation sites. We presented a very first attempt for computational prediction of diseaseassociated RNA methylation sites, helping researchers of the field to understand the roles of m(exp 6)A RNA methylation in human diseases and facilitating the development of the treatments.

Predict disease-related RNA methylation sites from the methylation-expression association by using hypergeometric test