how to calculate or extract Circular RNA expression via RNA-seq data

Question

[modarzi]

Hi,

Through RNA-seq data set, I extract Circular RNA.Now, I want to know the expression level of Circular RNA. But really I don't have any idea for doing that. I don't know is it possible that calculating Circular RNA expression via RNA-seq data or not? if it is possible, so how can I do it?

I appreciate if anybody share his/her idea for solving my problem.

Best Regards,

Mohammad

[modarzi]

Dear Dr. Blighe

Thanks for your comment. really I have seen these paper and know I use CIRI for extracting Circular RNA. my question is how can I find expression level of Circular RNA via RNA-seq Data? In the other word, is it possible to find expression level of Circular RNA from RNA-seq or not? if possible,so how and by which method?

I appreciate if you share your comment with me.

Best wishes,

Mohammad

[Kevin Blighe]

Yes, use one of these 'tools' and they will both identify the circular RNA for you and also determine their level of expression.

[Kevin Blighe]

If any of these do not quantify gene expression for you, then you can just obtain the circRNAs co-ordinates as GTF or GFF format, and then quantify expression over these with HTseq, Kallisto, featureCounts, or Salmon

[modarzi]

Dear Dr. Blighe

Thanks for you comment. as I wrote, I extract Circular RNA via CIRI and the result of that for one of my sample is as below:

circRNA_ID chr circRNA_start circRNA_end #junction_reads SM_MS_SMS #non_junction_reads junction_reads_ratio circRNA_type gene_id strand junction_reads_ID

chrX:139865340|139866824 chrX 139865340 139866824 7 2_2_0 2 0.875 intron ENSG00000101974.14, + SRR1427487.33684899.1,SRR1427487.11777859.1,SRR1427487.24302263.1,SRR1427487.26442679.1,SRR1427487.30395973.1,SRR1427487.48129543.1,SRR1427487.48467004.1,

As you see, for each CircRNA we have 12 attributes. Is it possible for you explain how can I use these Information for finding expression level of CircRNA in HTseq, Kallisto, featureCounts, or Salmon. In other words, are these information enough as input for that software for calculating expression level of CircRNA?

I appericiate if you share your comment with me.

Best Regards,

Mohammad

[Kevin Blighe]

Your best hope is probably featureCounts, in this case, but you will require the alignment BAM files for your samples, too.

With featureCounts, you also require a GTF file that contains the co-ordinates of your transcripts. To this GTF, you should add the entries for your identified circRNAs. GENCODE currently has some GTF files for download (for all currently known genes): https://www.gencodegenes.org/releases/current.html

So, the process is:

1. identify potential circRNAs
2. merge the co-ordinates of your potential circRNAs to those of existing known transcripts (in GTF format)
3. determine transcript abundance over your circRNAs and known transcripts using featureCounts
4. normalise data in a RNA-seq differential expression analysis tool, e.g., EdgeR or DESeq2

[modarzi]

Dear Dr Blighe

thanks for your comment.I have some questions that I appreciate if you share your comment with them: 1- for identification circRNAs, I used whole genome annotation file through GTF file.I got it via UCSC. Now, you recommend me that I have to "merge the co-ordinates of your potential circRNAs to those of existing known transcripts (in GTF format)". So, really for this step of process, I cant understand your mean.because based on up description, in extracting process of circRNA I used GTF file. 2-why should normalize data in a RNA-seq differential expression analysis tool? should I prepare DE analysis for calculating expression level of circRNA?

Best Regards,

Mohammad

[Kevin Blighe]

Hello again Mohammad. What is the ultimate aim of your study? How many samples do you have? Are you interested in differential expression between your sample groups (if you have groups)?

[modarzi]

Dear Dr Blighe

Hi,

In my study I have 2 groups: Case and Control. I have 2 candidate data set that one of them has 44 sample and another has 83 sample. So, I prefer to work on 83 samples but it needs dbGaP access permission and I don't have it. any way, first of all in case and control groups I have to extract CircRNAs and then find differential expression of CircRNAs in two groups.

This is all my research design.I appreciate if you share your comment with me.

Best Regards,

Mohammad

[Kevin Blighe]

Yes, then you should follow the approach that I mentioned:

1. identify potential circRNAs
2. merge the co-ordinates of your potential circRNAs to those of existing known transcripts (in GTF format)
3. determine transcript abundance over your circRNAs and known transcripts using featureCounts
4. normalise data in a RNA-seq differential expression analysis tool, e.g., EdgeR or DESeq2

For the GTF (part 2), download one of the GTFs from GENCODE: https://www.gencodegenes.org/releases/current.html

When editing the GTF to add your novel cirRNAs, just put in as minimum amount of information as required to adhere to the GTF format. The most important fields relate to the chromosome, base positions, and gene ID.

For determining transcript abundance, use something like Kallisto: https://pachterlab.github.io/kallisto/download

Does this help?

[abdul.suboor123]

My problem is that, using CIRI2 for circRNA prediction is ok but when I try CIRI_AS and try this command

perl CIRI_AS_v1.2.pl -S Splitmap.sam -C CIRI2out.ciri -O CIRI2ASout -F Zea_mays.AGPv4.dna.chromosome.1.fa -A annotation.gtf

So the following error come, what is the solution for it. I will be very thankful.

"Please make sure the SAM is the same one used for circRNA detection by CIRI without modification!"

[Kevin Blighe]

Is the SAM file the exact same as you used for CIRI?

[inu_juliocesar]

currently i am using CIRI2 when a use my sam file i use this argument "-I"

for example:

perl CIRI_AS_v1.2.pl -I Splitmap.sam.......

[Kevin Blighe]

and, do you have a question?

[abdul.suboor123]

Yes @Kevin Blighe the SAM file is the same that i have used for CIRI.

Answer 1

[Kevin Blighe]

عزيزي محمد ،

It is of course possible to infer circular RNAs from RNA-seq data, and there are a few algorithms / programs that can do this.

Two excellent resources for you are:

• Detecting circular RNAs: bioinformatic and experimental challenges (see Table 1)
• Comparison of circular RNA prediction tools

Table 1 from the second publication is here:

Table 1 from the first publicaion is too large to display here.

Both of these were published in 2016.

شكرا لك سيدي.

Kevin