Professional Documents
Culture Documents
Assessing The Impacts of Various Factors On Circular RNA Reliability
Assessing The Impacts of Various Factors On Circular RNA Reliability
Circular RNAs (circRNAs) are non-polyadenylated RNAs with a further demonstrated to be evolutionarily conserved across spe-
continuous loop structure characterized by a non-colinear cies (3, 10, 11, 12). Accumulating evidence demonstrated that
back-splice junction (BSJ). Although millions of circRNA can- circRNAs can participate in various aspects of biological processes,
didates have been identified, it remains a major challenge for including competing with their host gene expression and splicing,
determining circRNA reliability because of various types of regulating activities of miRNAs and RNA-binding proteins
false positives. Here, we systematically assess the impacts of (RBPs), and acting as translation templates (2, 13). CircRNAs were
numerous factors related to circRNA identification, conservation, reported to be especially abundant in brain and neuronal tissues
biogenesis, and function on circRNA reliability by comparisons of (11, 14) and regulatory important in nervous system development
circRNA expression from mock and the corresponding colinear/ and aging (15). Loss or dysregulation of circRNAs may affect brain
polyadenylated RNA–depleted datasets based on three different function (12, 15, 16, 17), indicating the roles of circRNAs in patho-
RNA treatment approaches. Eight important indicators of circRNA genesis and progression of neurological diseases. In cancer studies,
reliability are determined. The relative contribution to variability alternation of circRNA expression was shown to affect cell apoptosis,
explained analyses reveal that the relative importance of these invasion, migration, and proliferation, revealing the potential of
factors in affecting circRNA reliability in descending order is the circRNAs for serving as biomarkers and therapeutic targets in cancer
conservation level of circRNA, full-length circular sequences, (18, 19). These findings suggest that circRNAs are an ancient and fine-
supporting BSJ read count, both BSJ donor and acceptor splice tuned class of functional transcripts.
sites at the same colinear transcript isoforms, both BSJ donor and Varying bioinformatic tools have been developed to identify BSJs
acceptor splice sites at the annotated exon boundaries, BSJs using high-throughput transcriptome sequencing (RNA-seq) data
detected by multiple tools, supporting functional features, and (20), leading to a tremendous number of potential circRNA candidates in
both BSJ donor and acceptor splice sites undergoing alternative human and other species (21, 22, 23). However, circRNA detections are
splicing. This study thus provides a useful guideline and an im- hampered by various types of false positives. An NCL junction that
portant resource for selecting high-confidence circRNAs for originates from sequencing error, in vitro artifact, ambiguous alignment,
further investigations. trans-splicing, or genetic rearrangement is often misjudged as a BSJ
event by computational detections (1, 2). This reflects the fact of limited
DOI 10.26508/lsa.202201793 | Received 1 November 2022 | Revised 15
overlap among the circRNA candidates identified by different tools (1, 24,
February 2023 | Accepted 15 February 2023 | Published online 27 February
2023 25, 26, 27) or databases (23), highlighting the issue of circRNA reliability.
Although millions of circRNA candidates were deposited in varied da-
tabases, only a very limited number of candidates have been carefully
Introduction curated to date (23). It requires a guideline for evaluating the reliability of
the detected circRNA candidates before performing time- and cost-
CircRNAs are an emerging class of RNAs formed by cis-backsplicing consuming experimental validations.
with a covalently closed loop structure that are endogenously Because circRNAs are NCL RNAs and lack 39polyadenylated tails,
expressed as non-polyadenylated circular molecules (1, 2). Such a many circRNA candidates were detected in RNA-seq datasets de-
loop structure is characterized by a non-colinear (NCL) back-splice rived from total RNAs (the ribosomal RNA–depleted RNAs without
junction (BSJ) between a downstream splice donor site and up- poly(A)-selection; “mock RNAs” for simplicity), which contain RNA-
stream splice acceptor site. Despite most circRNAs are expressed at seq reads from both colinear and NCL RNAs (28). Because most
a much lower level compared with linear mRNAs (3, 4, 5), some are colinear RNAs are digested by RNase R (3, 28, 29, 30), a detected
highly expressed (3, 6), even more abundant than their colinear circRNA candidate is more likely to be real if its normalized BSJ read
counterparts (7, 8). In addition, circRNAs are more stable than their count is enriched or not reduced after the treatment. Therefore,
corresponding colinear mRNA isoforms (1, 9). Some circRNAs were comparisons of BSJ read counts detected from mock and RNase
Correspondence: trees@gate.sinica.edu.tw.
individuals (or samples). We found that the conservation levels of around the BSJs (see see the Materials and Methods section) were
circRNAs at the three levels (Fig 3A–D) were all significantly higher in generally higher in not-depleted circRNAs than in depleted ones
not-depleted circRNAs than in depleted ones, regardless of the RNA (Fig S2). These results suggested that the BSJ events widely detected
treatment approaches. The percentages of non-depleted circRNAs across species/samples were more reliable than those detected in
increased with increasing numbers of the conserved species (Fig a limited number of species/samples.
3A), tissues (Fig 3B), and samples (Fig 3D). Of note, in contrast with
the number of tissues (Fig 3B), the levels of tissue specificity index Factors related to circRNA biogenesis
were significantly lower in not-depleted circRNA than in depleted
ones (Fig 3C). We further showed that the evolutionary rates (as We proceeded to examine the relationships between circRNA re-
determined by phyloP (45) or phastCons (46)) of the sequences liability and the factors related to circRNA biogenesis. Because
circRNAs were suggested to be generated by canonical spliceo- (Fig 4E). In addition, a number of RBPs were demonstrated to serve as
somal machinery (1, 5, 47, 48, 49), we examined whether the BSJs a role for regulating circRNA biogenesis by binding to cis elements in
of not-depleted circRNA candidates tended to agree to well- the sequences flanking circularized exons (1, 2). We then examined
annotated exon boundaries of colinear transcripts or be subject whether the factor of BSJs with RBPs binding to the flanking regions
to alternative splicing (AS) based on previously annotated colinear showed enrichment for not-depleted circRNAs and found that this
transcripts (e.g., the Ensembl annotation) as compared with those factor was slightly enriched for not-depleted circRNAs only (Fig 4F). We
of depleted circRNA ones. Indeed, the BSJs that agreed to annotated also examined the BSJs with RCSacross or RBP-binding sites in flanking
exon boundaries (either donor or acceptor splice sites) showed a regions and did not observe a significant enrichment for not-depleted
significant enrichment for not-depleted circRNAs, regardless of the circRNAs (Fig 4G). Moreover, considering the minimum distance of the
RNA treatment approaches (Figs 4A and S3A). The BSJs with both detected RBP-binding sites to the BSJs, the distance was only slightly
donor and acceptor splice sites matching annotated exon shorter in not-depleted circRNAs than in depleted ones (Fig 4H). These
boundaries from the same colinear transcript isoforms were also results suggested that the factors of RCS or RBPs binding to the flanking
enriched for not-depleted circRNAs (Fig 4B). We also observed that regions were not good indicators for evaluating circRNA reliability.
the splice site strength of BSJs was generally stronger in not-
depleted circRNAs than in depleted ones, regardless of the tools Factors related to circRNA function
used for estimating the splice site strength (Fig S3B). Regarding the
BSJs matching previously annotated exon boundaries of colinear We then examined the impact of functional features (miRNA binding, RBP
transcripts, the BSJs (donor or acceptor splice sites) that were binding, and seven types of evidence for circRNA coding potential) on
subject to AS were significantly enriched for not-depleted circRNAs circRNA reliability (see the Materials and Methods section). For each BSJ
across the mock-treated sample pairs examined (Fig S3C). Such a event, because the circularized sequence is shared with its colinear
trend exhibited particularly enriched when both donor and ac- counterpart except for the BSJ donor and acceptor splice sites, we con-
ceptor splice sites of the BSJs were subject to AS (Fig 4C). These sidered the predicted miRNA/RBP-binding sites and ORFs spanning the
observations support that most circularized exons are recognized BSJ. Integration of these nine types of functional features, we observed that
by the spliceosome during canonical splicing. numbers of supporting functional features were significantly higher in not-
Next, previous studies demonstrated that backsplicing can be depleted circRNAs than in depleted ones in all mock-treated sample pairs
facilitated by RCSs residing in the sequences flanking circularized (all FDR < 0.05; Fig 5). The percentages of non-depleted circRNAs increased
exons (3, 8, 17, 29) and affected by the competition of RCSs across with increasing numbers of supporting functional features (Fig 5). These
flanking regions (RCSacross) or within individual regions (RCSwithin) (29). We results suggested that the strength of functional evidence for circRNA
thus asked whether the BSJs with RCSacross in flanking regions were candidates were positively correlated with the level of circRNA reliability.
enriched for not-depleted circRNA candidates. However, we did not
observe this trend (Fig 4D). We further examined the numbers of Relative effect of each individual factor on circRNA reliability
(RCSacross - RCSwithin) for the detected circRNAs and found no signif-
icant differences between not-depleted and depleted circRNAs in According to the above assessment results, we observed eight
terms of the percentages of the BSJs with (RCSacross - RCSwithin) ≥ 1 important indicators of circRNA reliability, each of which can
effectively distinguish between non-depleted and depleted circRNA reliability for each mock-treated sample pair (see the
circRNAs in all mock-treated sample pairs examined. The eight Materials and Methods section) (Fig 6A and Table S4). We
factors were (i) supporting BSJ read count; (ii) BSJs detected by ranked the RCVE scores for each mock-treated sample pair and
multiple tools; (iii) full-length circular sequences; (iv) conservation found that on average the most important features in descending
level of circRNA (number of samples observed the circRNAs); (v) order were the conservation level of circRNA, full-length circular
both BSJ donor and acceptor splice sites at the annotated exon sequences, supporting BSJ read count, both BSJ donor and acceptor
boundaries; (vi) both BSJ donor and acceptor splice sites at the splice sites at the same colinear transcript isoforms, supporting
same colinear transcript isoforms; (vii) both BSJ donor and functional features, both BSJ donor and acceptor splice sites at the
acceptor splice sites undergoing annotated AS events; and (viii) annotated exon boundaries, BSJs detected by multiple tools, and
supporting functional features. We then employed a general- both BSJ donor and acceptor splice sites undergoing annotated AS
ized linear model (GLM) with the eight factors and measured events (Fig 6B).
the relative contributions to variability explained (RCVE) (50) to Particularly, our result revealed that the conservation level of
evaluate the relative effect of individual factor in affecting circRNA (number of samples observed the circRNAs) is the most
dominant determinant of circRNA reliability common to all the one cell type/tissue only. Of the examined mock-treated sample
examined mock-treated sample pairs except for the sample pair in pairs, the RNA-seq data of five pairs (HeLa_1, HeLa_2, HeLa_3,
blood (Fig 6B). Of note, the samples considered were the integration HeLa_KCl, and HeLa_A) were all derived from HeLa cells, which were
of samples across species, tissue, and individuals in circAtlas. It is generated from different studies or conditions (Table S2). The five
known that circRNAs are often expressed in a cell type-/tissue- HeLa-based RNA-seq datasets can be regarded as replicates. As
specific manner (4, 51). Actually, we found that most (61%; 293,152 expected, the percentage of not-depleted circRNAs was strongly
circRNAs) of the 480,471 circAtlas circRNAs were detected in one positively correlated with the number of samples observed the
sample only (Table S3). We were curious about whether the positive circRNAs in all the five HeLa-based pairs (Fig 6C, left). The similar
correlation between the percentage of not-depleted circRNAs and the trend was observed for the two K562-based pairs (Fig 6C, middle).
number of samples observed the circRNAs held when considering We also examined the HeLa_1 and HeLa_2 mock-treated sample
pairs, which were two replicates generated from the same each of the five HeLa-based mock-treated sample pairs, the HeLa-
group but different studies (Table S2) and showed the similar specific circRNAs were divided into circRNAs detected in single replicate
results (Fig 6C, right). Furthermore, regarding the mock samples only and circRNAs detected in multiple replicates. Our results revealed
of the 19 mock-treated sample pairs, we extracted the HeLa- that the percentages of not-depleted circRNAs were significantly
specific circRNAs that were detected in the examined HeLa higher in circRNAs detected in multiple HeLa replicates than in
samples only but not in the non-HeLa cell lines/tissues. For circRNAs detected in single replicate only for all five HeLa-based
pairs (Fig 6D, left). Such a trend was held for the K562-based pairs whether the presence of G-quadruplexes across the BSJs affected
(Fig 6D, right). the percentages of not-depleted circRNAs and found no significant
differences between circRNAs with and without removing
G-quadruplexes across BSJs (Fig S4 and Table S3). Moreover, we
extracted BSJs from two high-confidence circRNA datasets, in which
Discussion the BSJ events had passed multiple experimental validations. For
the first dataset (8), BSJs were supported by both avian myelo-
The reliability of computationally detected circRNAs is not guar- blastosis virus- and Moloney murine leukemia virus-derived
anteed because of varied types of false positives. It is particularly RNA-seq reads (designated as “RT-independent circRNAs”). RT-
difficult to pick out spurious BSJ (or circRNA) events derived from independent circRNAs are regarded to be highly accurate be-
in vitro artifacts among a tremendous number of previously cause comparisons of different RTase products can effectively
identified NCL junctions. Such in vitro artifacts may frequently arise detect RT-based artifacts (1, 10, 52, 53). For the second dataset (55),
from template switching by random hexamer primed reverse BSJs (or their functions) were previously validated by RT-based
transcriptase (RT) reactions (52), which can switch from one tem- experiments and at least one type of non-RT–based experiments
plate to another with short homologous sequences at the NCL simultaneously in at least one human tissues or cell lines (des-
junction sites (53, 54). Of note, some NCL junctions passing RT- ignated as “RT-/non-RT–validated circRNAs”; see the Materials and
based validations were finally confirmed to be generated from Methods section). We observed that the percentages of both RT-
in vitro artifacts by further validations (10). As RNA-seq data are independent (Fig 7A) and RT-/non-RT–validated (Fig 7B) circRNAs
generated from RT-based sequencing strategies, it is unavoidable were significantly higher in not-depleted circRNAs than in depleted
that a number of spurious NCL events masquerade as genuine BSJs circRNAs in all mock-treated sample pairs examined. Moreover, we
because of such RT-based artifacts (1). We asked whether the compared circAtlas circRNAs with circRNAs from eight other pub-
approaches based on comparisons of paired mock and treated licly accessible circRNA databases and identified circAtlas-specific
samples can reflect circRNA reliability to a certain extent. We circRNAs. We presumed that circAtlas-specific circRNAs were less
emphasized that, for the robustness of our analyses, 19 mock- probably to be reliable as compared with circRNAs collected in
treated sample pairs of different samples/studies based on three multiple databases. Indeed, the percentages of circAtlas-specific
types of RNA treatment approaches (Table 1): RNase R approach circRNAs were significantly lower in not-depleted circRNAs than in
(group 1), A-tailing RNase R approach (group 2), and non-poly(A) depleted circRNAs (Fig 7C).
approach (group 3) were performed for the mock-treated com- For the relative influence of each individual factor on circRNA
parisons. Because highly structured 3’ ends or G-quadruplexes may reliability, our RCVE analysis revealed that “conservation level of
affect the effect of the RNase R treatment (32), we also examined circRNA” and “full-length circular sequences” were the top two
corresponding treated samples. For each RNA-seq dataset, dupli- and (4) within +1 to +10 nucleotides of the donor site. The evolu-
cated RNA-seq reads were removed using FastUniq (77). tionary rate of each region was measured by the average value of
the phyloP (or phastCons) scores of the considered nucleotides
(10 bp) within the region. The annotated exon boundaries and
Treat/mock ratio calculation
alternative splicing (AS) patterns were derived from the Ensembl
annotation (version 100). The splice site strength of BSJs was es-
For accuracy, only the circRNAs detected in the mock samples with
timated using MaxEntScan based on three scoring models (maxi-
the support of at least two BSJ reads were considered in the
mum entropy model, first-order Markov model, and weight matrix
analysis for each mock-treated sample pair. The expression levels
model) (78). The numbers of reverse complementary sequences
of circRNAs were calculated using the BSJ reads per million raw
(RCSs) across the flanking sequences (RCSacross; ±20,000 nucleo-
reads (RPM). For each considered circRNA candidate, we calculated
tides of the BSJs) or within individual flanking sequences (RCSwithin)
the ratio of the circRNA expression detected in the treated samples
of the BSJs were calculated using CircMiMi (79) with the command
to that detected in the corresponding mock samples (i.e., treat/
line: circmimi_tools check RCS –dist 20,000 genome.fa circRNAs.tsv
mock ratio). The not-depleted and depleted circRNAs are defined as
out.tsv. RBPs binding to the flanking regions (±1,000 nucleotides)
follows:
of BSJs was determined according to CLIP-supported RBP-binding
(
≥ 1; Not − depleted circRNAs sites downloaded from ENCORI (80) at http://starbase.sysu.edu.cn/.
treat=mock ratio G-quadruplexes across the BSJs were determined using the QGRS
< 1; Depleted circRNAs:
mapping algorithm with default parameters (81, 82). The predicted
G-quadruplexes should span the BSJ boundaries by ≥5 bp on both
sides of the BSJs. miRNA-binding sites across the BSJs were de-
Extractions of various factors termined using miRanda 3.3a (83) with pairing score ≥ 155. RBP-
binding sites across the BSJs were determined using RBPmap (84)
The factors of CIRI-full–identified full-length circular sequence, with a high stringency level and conservation filter. The predicted
number of circRNA detection tools, conservation patterns at the miRNA (or RBP)-binding sites should span the BSJ boundaries
individual (or sample), tissue, and species levels (including tissue by ≥5 (or ≥2) bp on both sides of the BSJs, respectively. The seven
specificity index) were extracted from circAtlas 2.0 (22). The BSJ types of evidence for coding potential of more than 320,000
events with the direct support of full-length circular sequences human circRNAs (including circAtlas circRNAs) were extracted
from nanopore long RNA-seq reads were downloaded from Liu from TransCirc (85). The seven types of evidence included
et al’s study (44). The evolutionary rates were determined by phyloP ribosome/polysome-binding evidence, experimentally supported
(45) or phastCons (46) scores, which were downloaded from the translation initiation site, internal ribosome entry site, predicted
UCSC Genome Browser at https://genome.ucsc.edu/. For each BSJ, m6A modification site, circRNA-specific ORF, sequence composition
we considered the four regions around the BSJ: (1) within +1 to +10 score for the ORF, and mass spectrometry data–supported peptide
nucleotides of the acceptor site; (2) within −10 to −1 nucleotides of across the BSJ. The translatable scores of circRNAs were down-
the acceptor site; (3) within −10 to −1 nucleotides of the donor site; loaded from TransCirc.
Statistics analysis This work was supported by T-J Chuang, Genomics Research Center, Aca-
demia Sinica, Taiwan and the National Science and Technology Council,
Taiwan (MOST 111-2311-B-001-021). We thank the related organizations for
For enrichment analyses, we compared not-depleted circRNAs with generating and providing the RNA-seq data used in this study. We also thank
depleted ones in terms of various factors related to circRNA other members of T-J Chuang laboratory for helpful discussions.
identification, conservation, biogenesis, and function. P-values
were determined using two-tailed Fisher’s exact test or Wilcoxon Author Contributions
rank-sum test (WRST; greater and less) (see also the figure legends).
P-values were further FDR adjusted across 19 mock-treated sample T-J Chuang: conceptualization, formal analysis, investigation,
pairs for each examined feature using Benjamini–Hochberg cor- methodology, and writing—original draft, review, and editing.
rection. The Wilcoxon effect sizes and the corresponding 95% T-W Chiang and C-Y Chen: software and methodology.
confidence intervals were calculated with the wilcox_effsize
function implemented in the rstatix R package (93). To measure the
Conflict of Interest Statement
relative effect of individual factor in determining circRNA reliability,
we first employed a GLM with all the examined factors using the
The authors declare that they have no conflict of interest.
statsmodels package (0.13.2) (94) in Python and then performed the
RCVE (50) as follows:
In the GLM, y indicated whether a circRNA candidate was a not- 2 Chen LL (2020) The expanding regulatory mechanisms and cellular
functions of circular RNAs. Nat Rev Mol Cell Biol 21: 475–490. doi:10.1038/
depleted circRNA (i.e., not-depleted circRNA = 1; depleted circRNA =
s41580-020-0243-y
0). f1~f8 represented the eight factors examined in the mode:
3 Jeck WR, Sorrentino JA, Wang K, Slevin MK, Burd CE, Liu J, Marzluff WF,
supporting BSJ read count; whether BSJs were detected by multiple
Sharpless NE (2013) Circular RNAs are abundant, conserved, and
tools; whether BSJs can be reconstructed full-length circRNAs; associated with ALU repeats. RNA 19: 141–157. doi:10.1261/rna.035667.112
number of samples observed the circRNAs; whether BSJs agreed to 4 Salzman J, Chen RE, Olsen MN, Wang PL, Brown PO (2013) Cell-type
annotated exon boundaries; whether BSJs had both donor and specific features of circular RNA expression. PLoS Genet 9: e1003777.
acceptor splice sites matching annotated exon boundaries from the doi:10.1371/journal.pgen.1003777
14 You X, Vlatkovic I, Babic A, Will T, Epstein I, Tushev G, Akbalik G, Wang M, 33 Zhang Y, Yang L, Chen LL (2016) Characterization of circular RNAs.
Glock C, Quedenau C, et al (2015) Neural circular RNAs are derived from Methods Mol Biol 1402: 215–227. doi:10.1007/978-1-4939-3378-5_17
synaptic genes and regulated by development and plasticity. Nat 34 Rabin A, Zaffagni M, Ashwal-Fluss R, Patop IL, Jajoo A, Shenzis S, Carmel L,
Neurosci 18: 603–610. doi:10.1038/nn.3975 Kadener S (2021) SRCP: A comprehensive pipeline for accurate
15 Gasparini S, Licursi V, Presutti C, Mannironi C (2020) The secret garden of annotation and quantification of circRNAs. Genome Biol 22: 277.
neuronal circRNAs. Cells 9: 1815. doi:10.3390/cells9081815 doi:10.1186/s13059-021-02497-7
16 Piwecka M, Glažar P, Hernandez-Miranda LR, Memczak S, Wolf SA, Rybak- 35 Chen CY, Chuang TJ (2019) NCLcomparator: Systematically post-
Wolf A, Filipchyk A, Klironomos F, Cerda Jara CA, Fenske P, et al (2017) Loss screening non-co-linear transcripts (circular, trans-spliced, or fusion
of a mammalian circular RNA locus causes miRNA deregulation and RNAs) identified from various detectors. BMC Bioinformatics 20: 3.
affects brain function. Science 357: eaam8526. doi:10.1126/ doi:10.1186/s12859-018-2589-0
science.aam8526 36 Petkovic S, Muller S (2015) RNA circularization strategies in vivo and in
17 Mai TL, Chen CY, Chen YC, Chiang TW, Chuang TJ (2022) Trans-genetic vitro. Nucleic Acids Res 43: 2454–2465. doi:10.1093/nar/gkv045
effects of circular RNA expression quantitative trait loci and potential 37 Hansen TB (2018) Improved circRNA identification by combining
causal mechanisms in autism. Mol Psychiatry 27: 4695–4706. doi:10.1038/ prediction algorithms. Front Cell Dev Biol 6: 20. doi:10.3389/
s41380-022-01714-4 fcell.2018.00020
18 Liu J, Zhang X, Yan M, Li H (2020) Emerging role of circular RNAs in cancer. 38 Gao Y, Wang J, Zhao F (2015) CIRI: An efficient and unbiased algorithm for
Front Oncol 10: 663. doi:10.3389/fonc.2020.00663 de novo circular RNA identification. Genome Biol 16: 4. doi:10.1186/
19 Rajappa A, Banerjee S, Sharma V, Khandelia P (2020) Circular RNAs: s13059-014-0571-3
Emerging role in cancer diagnostics and therapeutics. Front Mol Biosci 7: 39 Gao Y, Wang J, Zheng Y, Zhang J, Chen S, Zhao F (2016) Comprehensive
577938. doi:10.3389/fmolb.2020.577938 identification of internal structure and alternative splicing events in
20 Chen L, Wang C, Sun H, Wang J, Liang Y, Wang Y, Wong G (2021) The circular RNAs. Nat Commun 7: 12060. doi:10.1038/ncomms12060
bioinformatics toolbox for circRNA discovery and analysis. Brief 40 Mercer TR, Clark MB, Andersen SB, Brunck ME, Haerty W, Crawford J, Taft
Bioinform 22: 1706–1728. doi:10.1093/bib/bbaa001 RJ, Nielsen LK, Dinger ME, Mattick JS (2015) Genome-wide discovery of
21 Xia S, Feng J, Chen K, Ma Y, Gong J, Cai F, Jin Y, Gao Y, Xia L, Chang H, et al human splicing branchpoints. Genome Res 25: 290–303. doi:10.1101/
(2018) CSCD: A database for cancer-specific circular RNAs. Nucleic Acids gr.182899.114
Res 46: D925–D929. doi:10.1093/nar/gkx863 41 Li Y, Wang F, Teng P, Ku L, Chen L, Feng Y, Yao B (2022) Accurate
22 Wu W, Ji P, Zhao F (2020) CircAtlas: An integrated resource of one million identification of circRNA landscape and complexity reveals their pivotal
highly accurate circular RNAs from 1070 vertebrate transcriptomes. roles in human oligodendroglia differentiation. Genome Biol 23: 48.
Genome Biol 21: 101. doi:10.1186/s13059-020-02018-y doi:10.1186/s13059-022-02621-1
80 Li JH, Liu S, Zhou H, Qu LH, Yang JH (2014) starBase v2.0: decoding miRNA- 89 Glazar P, Papavasileiou P, Rajewsky N (2014) circBase: a database for
ceRNA, miRNA-ncRNA and protein-RNA interaction networks from large- circular RNAs. RNA 20: 1666–1670. doi:10.1261/rna.043687.113
scale CLIP-Seq data. Nucleic Acids Res 42: D92–D97. doi:10.1093/nar/ 90 Chen X, Han P, Zhou T, Guo X, Song X, Li Y (2016) circRNADb: A
gkt1248 comprehensive database for human circular RNAs with protein-coding
81 Frees S, Menendez C, Crum M, Bagga PS (2014) QGRS-conserve: A annotations. Sci Rep 6: 34985. doi:10.1038/srep34985
computational method for discovering evolutionarily conserved G- 91 Lai H, Li Y, Zhang H, Hu J, Liao J, Su Y, Li Q, Chen B, Li C, Wang Z, et al (2022)
quadruplex motifs. Hum Genomics 8: 8. doi:10.1186/1479-7364-8-8 exoRBase 2.0: An atlas of mRNA, lncRNA and circRNA in extracellular
vesicles from human biofluids. Nucleic Acids Res 50: D118–D128.
82 Menendez C, Frees S, Bagga PS (2012) QGRS-H predictor: A web server for
doi:10.1093/nar/gkab1085
predicting homologous quadruplex forming G-rich sequence motifs in
nucleotide sequences. Nucleic Acids Res 40: W96–W103. doi:10.1093/nar/ 92 Hinrichs AS, Karolchik D, Baertsch R, Barber GP, Bejerano G, Clawson H,
gks422 Diekhans M, Furey TS, Harte RA, Hsu F, et al (2006) The UCSC genome
browser database: Update 2006. Nucleic Acids Res 34: D590–D598.
83 Enright AJ, John B, Gaul U, Tuschl T, Sander C, Marks DS (2003) MicroRNA
doi:10.1093/nar/gkj144
targets in Drosophila. Genome Biol 5: R1. doi:10.1186/gb-2003-5-1-r1
93 Tomczak M, Tomczak E (2014) The need to report effect size estimates
84 Paz I, Kosti I, Ares M Jr., Cline M, Mandel-Gutfreund Y (2014) RBPmap: A
revisited. An overview of some recommended measures of effect size.
web server for mapping binding sites of RNA-binding proteins. Nucleic Trends Sport Sci 1: 19–25.
Acids Res 42: W361–W367. doi:10.1093/nar/gku406
94 Seabold S, Perktold J (2010) statsmodels: Econometric and statistical
85 Huang W, Ling Y, Zhang S, Xia Q, Cao R, Fan X, Fang Z, Wang Z, Zhang G modeling with python. In Proceedings of the 9th Python in Science
(2021) TransCirc: An interactive database for translatable circular RNAs Conference. Austin, TX: SciPy Organizers. doi:10.25080/Majora-92bf1922-011
based on multi-omics evidence. Nucleic Acids Res 49: D236–D242.
doi:10.1093/nar/gkaa823
86 Dong R, Ma XK, Li GW, Yang L (2018) CIRCpedia v2: An updated database License: This article is available under a Creative
for comprehensive circular RNA annotation and expression comparison. Commons License (Attribution 4.0 International, as
Genomics Proteomics Bioinformatics 16: 226–233. doi:10.1016/ described at https://creativecommons.org/
j.gpb.2018.08.001 licenses/by/4.0/).