- Expression normalization
1. Expression normalization
To accurately calculate the expression level of circRNAs, RNA-seq libraries with RbioMinus treatment were used, which provides an unbiased quantification of both linear and circular RNAs, as well as the calculation of circRNA junction ratio. For circRNAs that are not present in these libraries but are identified in the RbioMinus/RNase R treated libraries, we used the digestibility coefficient to estimate their expression levels. Specifically, shared circRNAs in the same tissues between the two types of libraries were extracted, and then the RNase R treatment coefficient was calculated by dividing the circRNA expression level of RbioMinus treated library to that of RbioMinus/RNase R treated library. After computing this parameter for each tissue, we obtained the normalized expression level of circRNAs only present in the RbioMinus/RNase R treated libraries by multiplying its expression by median RNase R treatment coefficient in a certain tissue. Please note that this type of circRNAs (only present in the RbioMinus/RNase R treated library) is marked in gray in the table browser.
Users can select one of the three species (left panel) to browse circRNAs. After choosing species, a table will be shown in the right panel, where each row represents a circRNA and columns represent the features of this circRNA, including type, parental gene (ENSEMBL id and gene name), evolutionary conservation, tissue specificity, GO annotation, expression and junction ratio profiles among tissues and whole sequence. After users click on the ‘more’ button, the detailed information of this circRNA will be shown at the bottom of the page.
Here is an example of the detailed information of a circRNA:
Select species and enter any search items (circRNA id, host gene id) into the search bar. The keyword of host gene is allowed, including ENSEMBL id or gene name. After clicking the search button, results for the search will be displayed.
Users can prioritize circRNAs at different levels, including species, tissues, type, ortholog, full length, tissue specificity and GO id.
- Species: human, macaque and mouse
- Expression: expression levels (normalized to the number of back-spliced reads per million mapped reads), high (>10), medium [5, 10), low [2, 5) and extremely low (0, 2)
- Tissues: selecting one or more tissues
- Type: the type of circRNAs (antisense, exon, intergenic, intron, non_repeat, 3utr and 5utr)
- Ortholog: selecting OO-type circRNAs
- Full-length: selecting circRNAs with whole-sequences
- Tissue specificity: the range between 0 and 1, which represents how strongly the expression is dominated by a single tissue (0 stands for circRNAs present in all tissues and 1 stands for those only present in one tissue)
- GO id: entering certain GO biological process term
Why the expression level of some circRNAs is ‘n/a’ ?
These circRNAs are generally in low abundance with a very limited number of BSJ reads, and we did not calculate their normalized expression level.