Circular RNAs, a family of covalently circularized RNAs with tissue-specific expression, were recently demonstrated to play important roles in mammalian biology. Regardless of extensive research to predict, quantify, and annotate circRNAs, our understanding of their functions is still in its infancy. In this study, researchers from the Baylor Scott & White Research Institute developed a novel computational tool: Competing Endogenous RNA for INtegrative Annotations (Cerina), to predict biological functions of circRNAs based on the competing endogenous RNA model. Pareto Frontier Analysis was employed to integrate ENCODE mRNA/miRNA data with predicted microRNA response elements to prioritize tissue-specific ceRNA interactions. Using data from several circRNA-disease databases, the researchers demonstrated that Cerina significantly improved the functional relevance of the prioritized ceRNA interactions by several folds, in terms of precision and recall. Proof-of-concept studies on human cancers and cardiovascular diseases further showcased the efficacy of Cerina on predicting potential circRNA functions in human diseases.
Flow chart of Cerina
(a) Analysis workflow for ENCODE tissue RNA-Seq/miRNA-Seq data. (b) Prediction of circRNA-miRNA bindings using TargetScan. (c) Pan-tissue analysis of ceRNA interactions. Incorporation of ENCODE gene expression data allows construct of tissue-specific circRNA-miRNA interaction networks. (d) Integrative analysis of ceRNA interactions. Pareto frontiers were calculated by integrating co-expression data with TargetScan-predict MRE data. (e) Based on the circRNA-miRNA-mRNA (gene)-function axis, circRNA functional prediction was performed by permuting the connections between a given circRNA and its interacting miRNAs/mRNAs.
Availability – A web service of Cerina can be accessed through: https://www.bswhealth.med/research/Pages/biostat-software.aspx. Source code for Cerina is available through GitHub at https://github.com/jcardenas14/CERINA.