- Circular RNA (circRNA) detection is an important area of bioinformatics focused on identifying and characterizing circular RNA molecules from genomic and transcriptomic sequencing data. Circular RNAs are a class of noncoding RNA molecules characterized by covalently closed loop structures formed through a process known as back-splicing. Unlike conventional linear RNAs, circRNAs lack free 5′ and 3′ ends, making them relatively resistant to exonuclease-mediated degradation and often more stable than their linear counterparts. They are found in animals, plants, fungi, and other organisms, where they may participate in gene regulation, cellular development, stress responses, and disease-associated processes. Bioinformatics approaches are essential for discovering circRNAs, identifying their genomic origins, quantifying their expression, and investigating their potential biological functions.
- Circular RNAs are primarily produced through back-splicing, in which a downstream splice donor site is joined to an upstream splice acceptor site. This process creates a circular RNA molecule rather than the conventional linear arrangement of exons. Depending on their composition and genomic origin, circRNAs are classified into several major types. Exonic circRNAs (ecircRNAs) consist mainly of exons and are commonly located in the cytoplasm. Circular intronic RNAs (ciRNAs) originate from intronic sequences, while exon–intron circRNAs (EIciRNAs) contain both exonic and intronic regions. Some circRNAs are generated from untranslated regions or other genomic regions. Their formation may be influenced by complementary sequences in flanking introns, RNA-binding proteins, and the activity of the cellular splicing machinery.
- The detection of circRNAs generally begins with RNA sequencing (RNA-seq). Because circRNAs do not have the polyadenylated tails commonly found in many linear messenger RNAs, conventional poly(A)-selected RNA-seq may fail to detect a substantial proportion of circRNAs. Ribosomal RNA-depleted total RNA-seq is therefore widely used for circRNA discovery. Another important experimental approach involves RNase R treatment, which preferentially degrades many linear RNAs while enriching circular RNA molecules. However, RNase R resistance is not absolute: some structured linear RNAs may resist digestion, and certain circular RNAs may be sensitive to the treatment. Using untreated and RNase R-treated samples together can provide stronger evidence for circRNA identification.
- After sequencing, raw reads undergo quality control and preprocessing. FastQC and MultiQC are commonly used to evaluate sequencing quality, while Cutadapt and fastp remove adapter sequences and low-quality bases. High-quality reads are then aligned to a reference genome using splice-aware aligners such as STAR, BWA, or other suitable mapping tools. CircRNA detection requires specialized computational approaches because the defining feature is a back-splice junction (BSJ), where a downstream genomic region is joined to an upstream region in the circular transcript. Reads spanning this junction provide important evidence that a circular RNA is present.
- Back-splice junction detection is the central computational step in circRNA identification. Unlike conventional linear RNA sequencing reads, which generally align in the expected genomic order, back-spliced reads contain sequences that map to genomic positions in a reverse order relative to normal transcription. Specialized algorithms identify these unusual alignments, reconstruct candidate circular junctions, and distinguish them from conventional splice junctions. Since sequencing errors, repetitive sequences, and incorrect read mapping can produce false-positive junctions, reliable detection requires careful filtering and validation.
- Several specialized tools have been developed for circRNA detection. CIRI2 identifies circRNAs by analysing paired-end and junction-spanning sequencing reads and is widely used for circRNA discovery. CIRCexplorer2 detects back-splice junctions and supports circRNA annotation and characterization. find_circ is an early computational method designed to identify circular RNA junctions from RNA-seq data. DCC provides a framework for circRNA detection and quantification, while CIRIquant supports detection, expression quantification, and differential analysis. Other tools, including circRNA_finder and KNIFE, offer alternative approaches to identifying circular RNA molecules. Different tools may produce different candidate sets because they use different alignment strategies, filtering criteria, and detection algorithms. Combining results from multiple tools can improve confidence, although it does not automatically eliminate false positives.
- Once candidate circRNAs have been identified, their genomic coordinates and host genes are determined through annotation. A circRNA host gene is the gene from which the circular transcript originates. Tools and resources such as Ensembl, RefSeq, GENCODE, and UCSC Genome Browser help identify genomic locations, exon boundaries, and relationships with annotated genes. CircRNAs may originate from protein-coding genes, noncoding RNA genes, or other genomic regions. Their annotation includes chromosome location, strand orientation, back-splice junction coordinates, exon composition, and transcript structure. Comparing candidate circRNAs with existing circRNA databases helps distinguish previously reported molecules from potentially novel circRNAs.
- Several databases provide information about known circular RNAs. circBase is an established resource containing circRNA annotations from different studies and organisms. circAtlas provides extensive circRNA information across multiple species, while CIRCpedia offers circRNA annotations and expression information. circBank provides information on human circRNAs and their potential regulatory relationships. These resources help researchers compare newly detected circRNAs with previously reported sequences, examine conservation, and investigate potential functions. Database coverage and annotation quality vary, so reference resources should be checked for their organism, genome assembly, and evidence standards.
- CircRNA expression quantification is another important stage of computational analysis. Expression can be estimated by counting sequencing reads that span back-splice junctions. Because conventional linear transcript quantification methods do not necessarily account for circular junctions, circRNA-specific tools are often required. Expression values may be normalized according to sequencing depth and other relevant factors. Differential expression analysis can then identify circRNAs whose abundance changes between tissues, developmental stages, disease conditions, or experimental treatments. Tools such as CIRIquant and DCC support aspects of circRNA quantification and differential analysis. Researchers should also consider the expression of the corresponding linear transcripts because changes in circRNA abundance may be related to changes in host-gene transcription or alternative splicing.
- Bioinformatics analysis can also investigate the potential functions of circRNAs. Some circRNAs interact with microRNAs and may influence their availability to regulate target messenger RNAs. This proposed mechanism is commonly described as the microRNA sponge effect, although it is relevant only to circRNAs with suitable binding sites, sufficient abundance, and appropriate cellular localization. Other circRNAs interact with RNA-binding proteins and may influence transcription, RNA stability, splicing, or cellular signalling. A small number of circRNAs have also been reported to undergo translation through noncanonical mechanisms, particularly when they contain suitable internal ribosome entry sites or other translation-promoting elements. Computational prediction can identify potential microRNA-binding sites, RNA-binding protein interactions, and open reading frames, but these predictions require experimental confirmation.
- Interaction analysis can integrate circRNA, microRNA, and messenger RNA expression data to construct competing endogenous RNA (ceRNA) networks. These networks are used to explore possible regulatory relationships among different RNA molecules. Tools such as miRanda, TargetScan, RNAhybrid, and other RNA interaction prediction methods can help identify candidate binding sites. Cytoscape can be used to visualize and analyse circRNA–microRNA–mRNA networks. Functional enrichment analysis using Gene Ontology (GO) and pathway databases such as KEGG can suggest biological processes associated with circRNA host genes and predicted target genes. However, computationally predicted networks should be interpreted cautiously because predicted binding does not necessarily demonstrate a functional regulatory interaction.
- Experimental validation is essential to confirm circRNA candidates identified through bioinformatics. Reverse transcription polymerase chain reaction (RT-PCR) using divergent primers is commonly used to amplify back-splice junctions. Sanger sequencing of the resulting products can verify the expected junction sequence. Quantitative PCR (qPCR) can measure circRNA abundance, while RNase R treatment can help distinguish circular from many linear RNA molecules. Northern blotting provides information about transcript size and abundance. Long-read sequencing can support the characterization of full-length circular transcripts, although specialized library preparation and analysis may be required. Additional experiments, including RNA fluorescence in situ hybridization (RNA-FISH), RNA pull-down, and RNA immunoprecipitation (RIP), can investigate cellular localization and molecular interactions.
- Circular RNA detection has applications in cancer research, neurological disorders, cardiovascular diseases, immune regulation, developmental biology, and biomarker discovery. Because some circRNAs are relatively stable and show tissue-specific expression, they are being investigated as potential diagnostic and prognostic biomarkers. Their roles in cellular differentiation, proliferation, apoptosis, and disease progression are also receiving increasing attention. In agriculture and animal science, circRNAs are studied in livestock species to investigate muscle development, fat deposition, growth, reproduction, lactation, immune responses, and disease resistance. Comparative analysis of circRNAs across breeds, tissues, and developmental stages may help identify regulatory molecules associated with economically important traits. Integration with quantitative trait loci (QTL), genome-wide association studies (GWAS), transcriptomics, and genomic selection research may further support investigations into complex animal traits.
- Despite improvements in sequencing and computational methods, circRNA detection remains challenging. False-positive back-splice junctions can arise from mapping errors, repetitive genomic regions, trans-splicing, genomic rearrangements, or template-switching artefacts during reverse transcription. Low-abundance circRNAs may be difficult to detect, while different sequencing protocols can produce substantial differences in detection sensitivity. In addition, circRNA databases may contain redundant entries or annotations based on different genome assemblies. Differences among computational tools can also make results difficult to compare. Reliable analysis therefore requires appropriate sequencing design, high-quality reads, careful alignment, suitable detection algorithms, stringent filtering, consistent annotation, and experimental validation.
- Overall, circular RNA detection is an important component of modern transcriptomics and noncoding RNA bioinformatics. Computational methods enable researchers to identify back-splice junctions, characterize circRNA structures, determine their genomic origins, quantify expression, and predict potential regulatory functions. Integration of RNA sequencing, specialized detection tools, circRNA databases, interaction analysis, and experimental validation continues to improve the understanding of circular RNA biology in humans, animals, plants, and other organisms.