Identification of mA-Associated RNA Binding Proteins Using an Integrative Computational Framework.

N6-methyladenosine (mA) is an abundant modification on mRNA that plays an important role in regulating essential RNA activities. Several wet lab studies have identified some RNA binding proteins (RBPs) that are related to mA's regulation. The objective of this study was to identify potential mA-associated RBPs using an integrative computational framework. The framework was composed of an enrichment analysis and a classification model. Utilizing RBPs' binding data, we analyzed reproducible mA regions from independent studies using this framework. The enrichment analysis identified known mA-associated RBPs including YTH domain-containing proteins; it also identified RBM3 as a potential mA-associated RBP for mouse. Furthermore, a significant correlation for the identified mA-associated RBPs is observed at the protein expression level rather than the gene expression level. On the other hand, a Random Forest classification model was built for the reproducible mA regions using RBPs' binding data. The RBP-based predictor demonstrated not only competitive performance when compared with sequence-based predictions but also reflected mA's action of repelling against RBPs, which suggested that our framework can infer interaction between mA and mA-associated RBPs beyond sequence level when utilizing RBPs' binding data. In conclusion, we designed an integrative computational framework for the identification of known and potential mA-associated RBPs. We hope the analysis will provide more insights on the studies of mA and RNA modifications.