Comprehensive analysis of circRNA expression profiles in humans by RAISE

Abstract

Circular RNAs (circRNAs) are pervasively expressed circles of non-coding RNAs. Even though many circRNAs have been reported in humans, their expression patterns and functions remain poorly understood. In this study, we employed a pipeline named RAISE to detect circRNAs in RNA-seq data. RAISE can fully characterize circRNA structure and abundance. We evaluated inter-individual variations in circRNA expression in humans by applying this pipeline to numerous non-poly(A)-selected RNA-seq data. We identified 59,128 circRNA candidates in 61 human liver samples, with almost no overlap in the circRNA of the recruited samples. Approximately 89% of the circRNAs were detected in one or two samples. In comparison, 10% of the linear mRNAs and non-coding RNAs were detected in each sample. We estimated the variation in other tissues, especially the circRNA high-abundance tissues, in advance. Only 0.5% of the 50,631 brain circRNA candidates were shared among the 30 recruited brain samples, which is similar to the proportion in liver. Moreover, we found inter- and intra-individual diversity in circRNAs expression in the granulocyte RNA-seq data from seven individuals sampled 3 times at one-month intervals. Our findings suggest that careful consideration of inter-individual diversity is required when extensively identifying human circRNAs or proposing their use as potential biomarkers and therapeutic targets in disease.

CircRNAwrap – a flexible pipeline for circRNA identification, transcript prediction, and abundance estimation

“…Alignment tools used to map chimeric reads include STAR, Tophat-fusion, and bwa mem, while tools like CIRI, circRNA_finder, and CIRCexplorer can align and extract backsplice junctions reads. Across the methods compared in this study, no one tool was demonstrated to be universally superior for identification and analysis of circRNAs [28,47,48].…”

“…Across the methods compared in this study, no one tool was demonstrated to be universally superior for identification and analysis of circRNAs [28,47,48]. Identification of cir-cRNAs from RNA-Seq data will be an integral step in the continued study of these RNA isoforms.…”

“…Several mathematical models are used to numerically describe the construction of transcripts, quantify abundance, and analyze differential expression. For circRNA whole transcript prediction, similar to the methods used for linear RNA transcript construction, paired-end mate reads, and splice reads are used to identify the circRNA inner structure [26][27][28][29]. These tools, such as CIRI-AS, FUCHS, and RAISE, use the splice and paired-end mate reads information to build whole coverage of circular transcripts [26,29].…”

“…CircRNA sequence prediction tools included CIRI-AS [26], CIRC-explorer2 [27], and RAISE [28]. To estimate the circRNA abundance with the merged circRNA tools, we used Sailfish-cir [30].…”

“…Our improved methods allow for miRNA seed analysis and differential motif enrichment, permitting the user to identify potential regulators of the backsplicing process. Several methods have been developed for cir-cRNA whole transcript prediction [26][27][28]38]. Due to the diversity of alternative splicing, it is necessary to build whole coverage of the circular transcripts.…”