Pytorch implementation and trained model for AcrNET: Predicting Anti-CRISPR with Deep Learning.
git clone https://github.com/banma12956/AcrNET.git
cd AcrNET
conda env create -f environment.yml
conda activate AcrNET
We provide a trained model and a simple dataset for demonstration. Run the following code and try:
python test.py
We also provide the five-fold cross-validation training code we used in our experiment:
python train_five_fold.py
The data we used in experiments can be downloaded at https://drive.google.com/file/d/1LK6y9g75ktlJEOy3CXZcPpZjQ4h4l-Ws/view?usp=sharing
You can also upload your own protein sequence data and use our trained model to make prediction.
DeepAcr needs structure information, evolutionary information and Transformer feature as input features, as shown in folder data. RaptorX, POSSUM, ESM-1b can provide corresponding calculation.