CRISPR-ONT & CRISPR-OFFT

Overview

CRISPR-ONT and CRISPR-OFFT are attention-based convolution neural networks (CNNs) frameworks to accurately predict CRISPR/Cas9 sgRNA on- and off-target activities, respectively. By integrating CNNs with attention mechanism, both CRISPR-ONT and CRISPR-OFFT are able to capture the intrinsic characteristics of Cas9-sgRNA binding and cleavage, thus improving the accuracy and interpretability.

Dependencies

• Ubuntu 16.04 or later
• Anaconda 3-5.2.0 or later
• Python packages:
  numpy
  pandas
• Keras
• Tensorflow

File Description

• crispr_ont.h5: the weights for the CRISPR-ONT model
• crispr_offt.h5: the weights for the CRISPR-OFFT model
• crispr_on_prediction.py: CRISPR-ONT code for sgRNA on-target activity prediction
• crispr_offt_prediction.py: CRISPR-OFFT code for sgRNA off-target activity prediction

Testing CRISPR-ONT and CRISPR-OFFT with test set

python crispr_on_prediction.py (for on-target activity prediction)
python crispr_offt_prediction.py (for off-target activity prediction)

Docker

Alternatively, you can start a Docker container and exec into it.

sudo docker build -t crispr .  
sudo docker run crispr