Deep Learning Models for off-target predicitons in CRISPR-Cas9 gene editing

This repository includes a deep convolutional neural network for predicting the off-targets in CRISPR-Cas9 gene editing. The CNN_std conducted by TensorFlow were trained using CRISPOR dataset. The CNN_std conducted by Keras were trained on the largest sgRNA off-target dataset up to date from [1].

PUBLICATION

Please cite this paper if using our preditive model:

Jiecong, Lin. & Ka-Chun, Wong. (2018). Off-target predictions in CRISPR-Cas9 gene editing using deep learning (ECCB 2018 Proceeding Special Issue). Bioinformatics, 34(17), i656–i663. http://doi.org/10.1093/bioinformatics/bty554

CRISPOR dataset

The dataset to plot ROC curves in Figure 3 and 4 of our paper are collected from CRISPOR paper [2].

The descriptions of this dataset can be found in Fig. 2 of the CRISPOR paper: "For each scoring method, shown are the True positive rate (TPR)/False positive rate (FPR) when classifying 143 validated off-targets with a mismatch count of up to four, one of the PAMs NAG/NGA/NGG, and a minimum modification frequency of 0.1 %.".

You can get this dataset from the github of CRISPOR paper (https://github.com/maximilianh/crisporPaper) through running the python script -- "plotRoc.py".

PREREQUISITE

The models for off-target predicitons were conducted by using Python 2.7.13 and TensorFLow v1.4.1. Following Python packages should be installed:

• scipy

• numpy

• pandas

• scikit-learn

• TensorFlow

The Keras version of CNN were conducted by Python 3.6, TensorFlow 1.9.0, Keras 2.2.0. Following Python packages should be installed:

• scipy

• numpy

• pandas

• Keras

• TensorFlow

REFERENCE

[1] Hui Peng, Yi Zheng, Zhixun Zhao, Tao Liu, Jinyan Li; Recognition of CRISPR/Cas9 off-target sites through ensemble learning of uneven mismatch distributions, Bioinformatics, Volume 34, Issue 17, 1 September 2018, Pages i757–i765, https://doi.org/10.1093/bioinformatics/bty558

[2] Haeussler, M., Schönig, K., Eckert, H. et al. Evaluation of off-target and on-target scoring algorithms and integration into the guide RNA selection tool CRISPOR. Genome Biol 17, 148 (2016). https://doi.org/10.1186/s13059-016-1012-2

CONTAINS:

• CFDScoring/cfd-score-calculator.py : Python script to run CFD score

• predictions/cnn_std_prediction_TF.py : CNN_std conducted by TensorFlow

• predictions/cnn_std_keras.py : CNN_std conducted by Keras used for off-target prediction

---

Jiecong Lin

[email protected]

January 27 2018