CircRNA Identification and Integration Pipeline

The pipeline provides an easy and reproducible way to detect circRNA from pair-end FASTQ files using four methods: CIRIquant, Circexplorer2, find_circ, and circRNA_finder.

The scripts and logs for handling TCCIA cohorts are available under the run_batch_from_qc path.

Step 1. Install the required conda environment

1. (Optional) Set an independent linux account circrna for deploying and running circRNA identification pipeline.

2. Install miniconda3 to default path, i.e., ~/miniconda3. When using recommended setting above, conda should be available at /home/circrna/miniconda3.

3. Install mamba to base env with conda install -n base --override-channels -c conda-forge mamba 'python_abi=*=*cp*'.

4. Install just with curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s -- --to ~/bin. Please add ~/bin here to your $PATH. You can change the ~/bin to anywhere, but you need to make the just available when you enter the terminal.

5. Install rush and add its path to $PATH, similar to just.

6. (Optional) Set registry of conda and pypi (pip) if necessary. For example, if you are in China, I recommend https://mirrors.tuna.tsinghua.edu.cn/help/anaconda/ & https://mirrors.tuna.tsinghua.edu.cn/help/pypi/.

7. Now, clone this repo (git clone git@42.192.87.178:coco/circrna-pipeline.git).

8. Install the conda environments one by one.

   cd circrna-pipeline
   cd CIRIquant
   just install
   cd ../FindCirc
   just install
   cd ../Circexplorer2
   just install
   cd ../circRNA_finder
   just install

Please make sure all the conda environments have been created with required softwares.

$ conda env list
# conda environments:
#
base                  *  /home/circrna/miniconda3
CIRIquant                /home/circrna/miniconda3/envs/CIRIquant
Circexplorer2            /home/circrna/miniconda3/envs/Circexplorer2
FindCirc                 /home/circrna/miniconda3/envs/FindCirc
circRNA_finder           /home/circrna/miniconda3/envs/circRNA_finder

Step 2. Prepare align index, reference file and config file

For running the pipeline, many files are required.

1. Prepare genome fasta file and gtf file. We use GRCh38.primary_assembly.genome.fa and gencode.v34.annotation.gtf.

2. For Circexplorer2, you need to download the reference file hg38_ref_all.txt (should be corresponding to your reference genome) with fetch_ucsc.py script from the Circexplorer2 environment.

3. Prepare align index, config_zhou.sh has recorded the commands. Please note, you need to activate corresponding environment before run index commands.

   For example, to prepare index for CIRIquant.

   source activate CIRIquant
   bwa index -a bwtsw -p /path/to/GRCh38.primary_assembly.genome.fa /path/to/GRCh38.primary_assembly.genome.fa
   hisat2-build -p 40 /path/to/GRCh38.primary_assembly.genome.fa /path/to/GRCh38.primary_assembly.genome.fa

4. For CIRIquant, a yml file is required to set the paths of softwares and files, e.g., hg38.yml. You need to modify the contents to fit your setting (You can also create another yml file).

5. Set a config.sh file, it sets all required setting with SHELL variables, config_zhou.sh is a good reference (Of cource, you can modify its contents to fit your needs).

Step 3. Preprocess pair-end fastq files

You need to preprocess your pair-end fastq files (QC, cut adapters, etc.). fastp is a one-stop solution for this.

Currently, we only support file names with the postfix _1.fastq.gz and _2.fastq.gz. Please make sure you output your processed fastq files in such a format.

Step 4. Test and run pipeline

Create a shell script with following settings and commands.

fqfile=./sample_list.txt
indir=/path/include/paired/fastq/files
oudir=/path/to/output
nthreads=20
config=/path/to/your/config.sh

common/ll_fq.py ${indir} --output ${fqfile}

nohup bash caller.sh ${fqfile} ${indir} ${oudir} ${nthreads} ${config} &> run.log &

The script has to be executed in conda base environment (or python3 is installed). If you have prepared the sample_list.txt file by your own. You can comment the common/ll_fq.py row, and you can run the script in bash without any other requirement (e.g., no python3 is required from the base environment).

The directory run_batch have examples for running our TCCIA cohorts.

I recommend testing the pipeline with 4 samples. If it goes well, run all the data files you have. The pipeline will skip samples with result files already generated.

After getting the detection results from 4 methods, you can use the ensemble approach (code under aggr to get final results.

An example is given as:

workdir=/home/zhou/raid/IO_RNA/circRNA/PHS003316

bash -c "../../aggr/aggr_beds.R ${workdir} ${workdir}/aggr && ../../aggr/aggr_dataset.R ${workdir}/aggr ${workdir}/aggr ./PHS003316.txt" &> PHS003316_aggr.log

Output

The output directory contains result files with names combined from sample names and methods.

$ ls *.bed
GO28753_ngs_rna_targrna_rnaaccess_EA_0f0fda909f_20150820.circexplorer2.bed
GO28753_ngs_rna_targrna_rnaaccess_EA_0f0fda909f_20150820.circRNA_finder.bed
GO28753_ngs_rna_targrna_rnaaccess_EA_0f0fda909f_20150820.CIRI.bed
GO28753_ngs_rna_targrna_rnaaccess_EA_0f0fda909f_20150820.find_circ.bed

The result file usually contains the postion and count value of circRNAs.

$ head GO28753_ngs_rna_targrna_rnaaccess_EA_0f0fda909f_20150820.circexplorer2.bed
chr3    9750944 9751949 +       31
chr3    11331339        11348035        +       16
chr3    12489783        12489989        +       19
chr3    12496517        12503784        +       1
chr3    15016151        15034809        +       11
chr3    15210801        15212212        +       12
chr3    15232518        15233973        +       1
chr3    15563357        15573242        -       3
chr3    17372074        17384015        -       2
chr3    18415110        18420991        -       9

Reference

We would like to thank all contributors of the following two projects. Our pipeline was inspired by the two works and could not be built without them.

• https://github.com/Yelab2020/ICBcircSig
• https://github.com/nf-core/circrna

Copyright

© 2023 - OncoHarmony Network by Shixiang Wang, Yi Xiong, and Jian-Guo Zhou.