Assessment of different enrichment methods revealed the optimal approach to identify bovine circRnas

Figures & data

Figure 1. The workflow of circRNAs enrichment and sequencing. Total RNA was extracted from three biological replicate sample of cattle liver and rumen tissue. circRNA enriched libraries were constructed using the methods: Ribo: rRNA depletion using Ribo-zero (Green line); R: RNase R (Brown line); RTP: RNase R + Tailing + Poly (A) depletion (Blue line); Ribo-R: Ribo-zero + RNase R (Yellow line); Ribo-RP: Ribo-zero + RNase R + Poly (A) depletion (Red line); Ribo-RTP: Ribo-zero + RNase R + Tailing + Poly (A) depletion (Purple line). The libraries were sequenced using Novaseq 6000 S4 PE100 platform (Illumina). After quality control and alignment, remaining reads were used for circRNAs identification. BSJ: Back-splicing junction.

Figure 2. The ratio of uniquely mapped reads (a proportion of the reads uniquely mapped to the reference genome out of all clean reads) from different circRNA enrichment methods. The Chi-Square value and the adjusted p value (Asymptotic Sig) of the Kruskal-Wallis H test are shown. Ribo: rRNA depletion using Ribo-zero (Green box); R: RNase R (Brown box); RTP: RNase R + Tailing + Poly (A) depletion (Blue box); Ribo-R: Ribo-zero + RNase R (Yellow box); Ribo-RP: Ribo-zero + RNase R + Poly (A) depletion (Red box); Ribo-RTP: Ribo-zero + RNase R + Tailing + Poly (A) depletion (Purple box).

Table 1. Summary of sequencing data and the number of identified circRnas.

Tissue	Method^a	Sample ID	Raw reads	Clean reads	Total reads on BSJ^b	CircRNAs with BSJ = 1^c	CircRNAs with BSJ ≥ 2^d
Liver	Ribo	Li_Ribo_1	32,386,032	29,904,109	2,153	1,186	305
Liver	Ribo	Li_Ribo_2	30,155,098	27,827,542	1,973	1,165	253
Liver	Ribo	Li_Ribo_3	33,838,111	31,537,720	2,189	1,211	305
Liver	R	Li_R_1	81,121,434	61,042,504	11,696	2,714	1,979
Liver	R	Li_R_2	94,699,576	80,605,591	11,129	2,442	1,904
Liver	R	Li_R_3	62,265,781	84,480,012	15,157	2,931	2,493
Liver	RTP	Li_RTP_1	89,391,711	76,786,092	16,999	3,421	2,706
Liver	RTP	Li_RTP_2	82,486,126	87,011,943	18,342	3,309	2,964
Liver	RTP	Li_RTP_3	83,799,180	70,011,904	6,078	1,285	1,184
Liver	Ribo-R	Li_Ribo-R_1	94,603,289	73,263,710	28,147	4,491	4,088
Liver	Ribo-R	Li_Ribo-R_2	81,451,996	85,015,973	26,233	4,179	3,877
Liver	Ribo-R	Li_Ribo-R_3	69,089,549	80,190,913	25,792	4,059	3,853
Liver	Ribo-RP	Li_Ribo-RP_1	80,316,499	67,832,188	24,306	3,915	3,690
Liver	Ribo-RP	Li_Ribo-RP_2	98,163,995	81,833,234	25,071	3,942	3,795
Liver	Ribo-RP	Li_Ribo-RP_3	92,896,427	75,564,054	27,540	3,857	4,177
Liver	Ribo-RTP	Li_Ribo-RTP_1	65,634,347	61,779,606	21,544	2,952	3,521
Liver	Ribo-RTP	Li_Ribo-RTP_3	60,754,526	49,684,853	21,435	2,362	3,584
Rumen	Ribo	Ru_Ribo_1	34,779,311	32,289,158	3,858	1,960	594
Rumen	Ribo	Ru_Ribo_2	34,266,770	31,763,222	5,322	2,444	836
Rumen	Ribo	Ru_Ribo_3	33,056,329	30,092,472	4,588	2,260	696
Rumen	R	Ru_R_1	75,526,801	75,440,120	10,640	2,924	1,966
Rumen	R	Ru_R_2	91,056,217	70,140,131	6,049	1,861	1,195
Rumen	R	Ru_R_3	84,654,269	76,087,344	10,673	2,169	2,058
Rumen	RTP	Ru_RTP_1	77,400,233	54,265,001	12,594	2,932	2,395
Rumen	RTP	Ru_RTP_2	91,810,264	82,620,794	11,247	2,566	2,134
Rumen	RTP	Ru_RTP_3	67,738,941	66,075,494	15,626	2,825	2,802
Rumen	Ribo-R	Ru_Ribo-R_1	97,359,928	82,336,274	27,906	5,269	4,404
Rumen	Ribo-R	Ru_Ribo-R_2	91,506,493	74,382,418	18,066	3,868	3,037
Rumen	Ribo-R	Ru_Ribo-R_3	88,451,557	80,074,908	28,966	4,477	4,527
Rumen	Ribo-RP	Ru_Ribo-RP_1	92,962,733	73,106,595	24,625	4,631	4,039
Rumen	Ribo-RP	Ru_Ribo-RP_2	98,378,031	83,056,975	18,667	3,717	3,233
Rumen	Ribo-RP	Ru_Ribo-RP_3	88,094,371	58,998,847	24,745	3,449	4,083
Rumen	Ribo-RTP	Ru_Ribo-RTP_2	60,786,354	45,322,371	10,401	1,984	2,099

^aRibo (Ribo-zero); R (RNase R); RTP (RNase R & Tailing & Poly (A) depletion); Ribo-R (Ribo-zero & RNase R); Ribo-RP (Ribo-zero & RNase R & Ploy (A) depletion); Ribo-RTP (Ribo-zero & RNase R & Tailing & Ploy (A) depletion). ^bThe total number of reads mapped to the back splicing junction site; ^cThe number of circRNAs with only one read mapped to the back splicing junction; ^dThe number of circRNAs with at least two reads mapped to the back splicing junction site.

Figure 3. The quality and number of identified circRNA. The false-positive rate and the number of circRNA per million clean reads were calculated for liver (A) and rumen (B), respectively. The Chi-Square value and adjusted p value (Asymptotic Sig) of the Kruskal-Wallis H test are shown. The value on the left is for false-positive rate and the value on the right is for the number of circRNA per million clean reads. Different lowercase letters indicate significant differences for pairwise comparison of false-positive rate (Adjusted p < 0.05), and uppercase letters indicate significant differences for pairwise comparison of the number of circRNA per million clean reads (Adjusted p < 0.05). Ribo: rRNA depletion using Ribo-zero (Green); R: RNase R (Brown); RTP: RNase R + Tailing + Poly (A) depletion (Blue); Ribo-R: Ribo-zero + RNase R (Yellow); Ribo-RP: Ribo-zero + RNase R + Poly (A) depletion (Red); Ribo-RTP: Ribo-zero + RNase R + Tailing + Poly (A) depletion (Purple).

Figure 4. Identified circRNAs intersection between different methods. The upset plots of the liver (a) and rumen (b) were generated based on the circRnas with at least two reads mapped to the back-splicing junction (BSJ) position and identified in all biological replicates within the group. The longitudinal axis represents the intersection size of the circRNA number. The horizontal axis represents the input circRNA number of each group. The dark circles indicate sets that are part of the intersection. The clustered bar on the left bottom presented the total number of exonic circRnas (filled) and intronic circRnas (frame) identified by different method. Ribo: rRNA depletion using Ribo-zero (Green); R: RNase R (Brown); RTP: RNase R + Tailing + Poly (A) depletion (Blue); Ribo-R: Ribo-zero + RNase R (Yellow); Ribo-RP: Ribo-zero + RNase R + Poly (A) depletion (Red); Ribo-RTP: Ribo-zero + RNase R + Tailing + Poly (A) depletion (Purple).

Figure 5. CircRNA conservation between cattle and human. (a) The workflow used to identify circRNAs conserved between cattle and human. Cattle circRNAs were divided into six types (marked with roman number) and schematic diagrams of conserved circRNAs were represented (Blue: human circRNAs, Red: cattle circRNAs). (b) The proportion of different types of cattle circRNAs when compared with human circRNAs. (c) The density of conserved circRNAs on cattle chromosomes. The number of circRNAs was corrected with chromosome length (Mb).

Figure 6. Circos plot of circRNAs conserved between bovine and human genome. The outermost layer is the chromosome name (Bovine: B1 - BX; Human: H1 – HX). The conserved circRNAs identified in different enrichment methods were labelled as black lines in the frame (Ribo-RTP, purple frame; Ribo-RP, red frame; Ribo-R, yellow frame; RTP, blue frame; R, brown frame; Ribo, green frame). The relative length of cattle and human chromosomes were represented by deep purple and deep brown boxes, respectively. The conserved cattle circRNAs and corresponding human circRNAs are linked using lines. Green lines represent links between non-homologous conserved circRNAs, and red lines represent homologous circRNAs.

Data availability statement

The five sequencing datasets of Ribo method were downloaded from Sun et al. projects (GEO accession number GSE116775) [43]. All the sequencing datasets generated from current study have been deposited at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE226717.