CirComPara: A multi-method comparative bioinformatics pipeline to detect and study circRNAs from RNA-seq data

Abstract

Circular RNAs (circRNAs) are generated by back-splicing of immature RNA forming covalently closed loops of intron/exon RNA molecules. Pervasiveness, evolutionary conservation, massive and regulated expression, and post-transcriptional regulatory roles of circRNAs in eukaryotes have been appreciated and described only recently. Moreover, being easily detectable disease markers, circRNAs undoubtedly represent a molecular class with high bearing on molecular pathobiology. CircRNAs can be detected from RNA-seq data using appropriate computational methods to identify the sequence reads spanning back-splice junctions that do not co-linearly map to the reference genome. To this end, several programs were developed and critical assessment of various strategies and tools suggested the combination of at least two methods as good practice to guarantee robust circRNA detection. Here, we present CirComPara (http://github.com/egaffo/CirComPara), an automated bioinformatics pipeline, to detect, quantify and annotate circRNAs from RNA-seq data using in parallel four different methods for back-splice identification. CirComPara also provides quantification of linear RNAs and gene expression, ultimately comparing and correlating circRNA and gene/transcript expression levels. We applied our method to RNA-seq data of monocyte and macrophage samples in relation to haploinsufficiency of the RNA-binding splicing factor Quaking (QKI). The biological relevance of the results, in terms of number, types and variations of circRNAs expressed, illustrates CirComPara potential to enlarge the knowledge of the transcriptome, adding details on the circRNAome, and facilitating further computational and experimental studies.