jsacco1 / igwos

iGWOS: integrated Genome-Wide Off-target cleavage Search

iGWOS

integrated Genome-Wide Off-target cleavage Search platform

Introduction

iGWOS is designed for the optimal genome-wide off-target sites (OTS) prediction by integrating the available OTS prediction tools in a complementary way. iGWOS integrates OTS prediction tools with an Adaboost framework, supports conventional NGG-PAM OTS prediction with mismatches up to 6 in human species.

iGWOS searches the candidate off-target sites with Cas-OFFinder and predict CRISPR/Cas9-induced off-target cleavage sites by integrating distinct OTS prediction tools (CRISPRoff, uCRISPR, DeepCRISPR, CFD, MIT, CROP-IT and CCTop). The genome encode way can be referred to DeepCRISPR (https://github.com/bm2-lab/DeepCRISPR)

By inputting the gRNA(s) sequence file and related restrictions, iGWOS precisely predicts the genome-wide OTS list of given gRNAs, and visualizes the top 200 risky genome-wide off-target profile with a Circos plot. The iGWOS score of a off-target site denotes its cleavage probability.

Requirement

• Docker==18.09.6
• python==3.7
• pandas==0.20.1
• numpy==1.14.5
• pyfaidx==0.4.8.4
• tensorflow==1.8.0
• sonnet==1.33
• uCRISPR==0.1
• RNAstructure==6.2
• python==2.7
• biopython==1.73
• ViennaRNA==2.4.12
• RIsearch==2.1
• circos==0.69-6

Installation

Please refer to file "Dockerfile" in the code to build the performing environment for iGWOS. Users could easily perform the iGWOS under the docker image.

# build image based on Dockerfile
docker build -t igwos:latest ./

# run container
docker run --name [container_name] -d igwos:latest

Usage

python3 main.py [-h] [-v] [-gRNA GRNA] [-g GENOME] [-m {0,1,2,3,4,5,6}]
         [-cell CELL] [-cid CID] [-e ENCODE] [-circos {0,1}] [-gpu GPU]
         [-o OUTPUT]

Predict genome-wide CRISPR-Cas9 off-target sites with iGWOS.

optional arguments:
-h, --help        show this help message and exit
-v, --version     show program's version number and exit
-gRNA GRNA        gRNAs file in Fasta format
-g GENOME         genome folder for candidate off-target searching,
                    default=genome/hg19
-m {0,1,2,3,4,5,6}  maximum mismatch allowed in off-target prediction,
                    default=5
-cell CELL        cell-type of gRNAs
-cid CID          cell-id file, formed like data/encode_hg19.tab
-e ENCODE         epigenomic encode folder, default=/data/genome/encode/fa/
-circos {0,1}     whether to draw a circos plot to visualize the top 200
                  risky predicted off-target profile, default=1
-gpu GPU          select a gpu device to perform cas-offinder and/or
                    deepcrispr, default=0
-o OUTPUT         output folder, default=output/

Example

python3 main.py -gRNA data/grna.fa -g genome/hg19 -m 5 -cell K562 -cid data/encode_hg19.tab -e /data/genome/encode/fa/ -circos 1 -gpu 1 -o output

gRNA file format

>sg1
GCCTCCCCAAAGCCTGGCCAGGG
>sg2
GGCCAGGCTTTGGGGAGGCCTGG

cell-id file format: [cid] [cell]

h1	MCF-7
h2	GM12878
h3	HepG2
h4	LNCaP clone FGC
h5	HCT116
h6	HeLa-S3
h7	K562

encode folder format (take K562 cell as an example)

Format: [encode_path]/[cid]_[epi].fa

/data/genome/encode/fa/h7_ctcf.fa
/data/genome/encode/fa/h7_dnase.fa
/data/genome/encode/fa/h7_h3k4me3.fa
/data/genome/encode/fa/h7_rrbs.fa

output format

sgID	gRNA	OTS	Chr	Strand	Start	Mismatch	iGWOS
sg1	GCCTCCCCAAAGCCTGGCCAGGG	GCCTCCCCAAAAGCTGAGCAGGG	chr1	+	929401	4	0.3254589692131059
sg1	GCCTCCCCAAAGCCTGGCCAGGG	TGCTCCCCAGAGCCTAGCCGTGG	chr7	-	1900836	5	0.46115958387343875
sg1	GCCTCCCCAAAGCCTGGCCAGGG	ACCTCCCCATAGCCTGGCCAGGG	chr11	-	44986455	2	0.529510001952807

OTS Circos visualization format

Citation

Jifang Yan, Qi Liu et al. Benchmarking and integrating genome-wide CRISPR off-target detection and prediction. Nucleic Acids Research, 2020 (Manuscript accepted)

Contacts

[email protected] or [email protected]