This repository contains data and scripts for reproducing the results of the circular RNA assembler CircPlex. The official CircPlex repository is available here.
Reproducing the results of CircPlex requires C++11 and python3.
The data folder contains integrated information from CircPlex experiments. There are two subdirectories data/testis and data/brain that contain information from the testis and brain sample respectively.
data/{tissue}/{tissue}_features_A_combined.tsv - Details of circRNAs sseqeunces predicted by CircPlex from rolling circular nanopore long reads.
data/{tissue}/{tissue}_features_AAbar_combined.tsv - Details of seqeunces containing circRNA and its partial reverse complement from rolling circular nanopore long reads.
data/{tissue}/{tissue}_bsj_A.tsv - Details of back-splicing junctions from circRNAs predicted by CircPlex.
data/{tissue}/{tissue}_bsj_A_0.1.tsv - Details of back-splicing junctions from circRNAs predicted by CircPlex that have absolute over-prediction ratio ≤ 1.
data/{tissue}/{tissue}_bsj_cluster_A.tsv - Details of clustered back-splicing junctions from circRNAs predicted by CircPlex.
data/{tissue}/{tissue}_bsj_cluster_A_0.1.tsv - Details of clustered back-splicing junctions from circRNAs predicted by CircPlex that have absolute over-prediction ratio ≤ 1.
The scripts directory contain scripts to reproduce the results from the CircPlex experiments.
scripts/plot.py - To generate over-prediction ratio distributions of circRNA seqeunces from CircPlex and seqeunces containing circRNA and its partial reverse complement.
scripts/compare.cpp - To compare overlaps of BSJs predicted by CircPlex and isoCirc with circBase.
For generating plots, run the following command.
python3 scripts/plot.py data/{tissue}/{tissue}_features_A_combined.tsv data/{tissue}/{tissue}_features_AAbar_combined.tsv <output-driectory>
Output files {tissue}_A_hist.pdf and {tissue}_AAbar_hist.pdf will be generated inside <output-directory>.
The cpp file scripts/compare.cpp takes as input a BSJ file from the /data/{tissue} folder, the isocirc annotaion (isocirc.out) file for the tissue and circBase annotation (bed) file. An output file {prefix}_venn_counts.txt is generated in the specified output directory that summarizes the overlap counts between CircPlex, isoCirc and circBase. The {prefix} will be automiatically generated based on the input BSJ file.
To generate isoCirc annotations, download the corresponding long reads data for human testis and brain sampples from GEO accession number: GSE141693. Then run the tool isoCirc on the testis and brain reads to get respective isoCirc annotations. Note that the isoCirc annotations used in CircPlex evaluation are filtered to remove duplicate BSJs and partial circRNAs.
To install isoCirc (license), visit (https://github.com/Xinglab/isoCirc).
The circBase annotation bed file can be downloaded from here.
Compile the scripts/compare.cpp file.
g++ -O3 scripts/comapre.cpp -o scripts/compare
Run the follwoing command.
scripts/compare /data/{tissue}/{tissue}_{bsj/cluster}_{A/A_0.1}.tsv <tissue-isocirc-annotation (filtered isocirc.out file)> <circBase-bed-file> <output-directory>
Output file {tissue}_{bsj/cluster}_{A/A_0.1}_venn_counts.txt will be generated inside <output-directory> with the overlap information.