CircRNAFisher: a systematic computational approach for de novo circular RNA identification

Circular RNAs (circRNAs) are emerging species of mRNA splicing products with largely unknown functions. Although several computational pipelines for circRNA identification have been developed, these methods strictly rely on uniquely mapped reads overlapping back-splice junctions (BSJs) and lack approaches to model the statistical significatnce of the identified circRNAs by simultaneously utilizing BSJ overlapping reads and discordant BSJ spanning reads to identify circRNAs. Moreover, we developed a novel procedure to estimate the P-values of the identified circRNAs. A computational cross-validation and experimental validations demonstrated that our ethod performed favorably compared to existing circRNA detection tools. We created a standalone tool, CircRNAFisher to implement the method which might be valuable to computational and experimental scientists studying circRNAs.