Fig. 4

Analysis of reverse complementary sequences and repeat elements for circRNA formation. a Three kinds of repeat element density in the introns adjacent to end-joining exons of circRNAs. b CircRNAs are divided into three groups according to whether the upstream and downstream introns contain reverse complementary (RC) sequences and whether these sequences belong to repeat elements (RC-repeats and RC-Nonrepeats). The circRNA expression levels are normalized by the host gene FPKM. CircRNAs with adjacent introns containing RC sequences shows higher expression than those without RC sequences (NonRC). Moreover, the circRNA expression shows positive correlation with the number of RC sequence pairs (Fig. S8c in Additional file 1). c We calculated the distance between the nearest pair of RC sequences by summing up their distances to the circRNA splicing sites. When the distance is less than 5 kb, the RC sequences seem not to help circRNA formation. d CircRNAs are classed by the strength of the upstream splicing site (up5′ and up3′) of the first exon forming the end-joining site and those with strong up3′ (top) show higher expression than those with weak up3′, while no differences caused by the strength of up5′ were observed. Then when classing the circRNAs by strength of downstream splicing motif (down5′ and down 3′), only circRNAs with strong down5′ and weak down3′ show higher expression than all the other groups