CRISPRclassify-CNN-Att is a deep learning-based method that utilizes Convolutional Neural Networks (CNNs) and self-attention mechanisms to classify CRISPR-Cas systems based on repeat sequences.

• data/: Contains data files .

  • repeats_all.csv: csv file containing all repeats and their corresponding subtypes.

• source/: Contains all the script files.

  • CNN_Att.py: CNN and self-attention mechanism models.
  • dataselect.py: selecting datasets.
  • repeatsEncoder.py: encoding repeats.
  • structure_features.py: calculating additional features (including k-mer frequency, GC content, and sequence length).
  • transferlearning.py: fine-tuning for classifying less abundant subtypes .
  • typeEncoder.py: encoding subtypes.
  • stacking.py: model stacking.

• model/:

  • cnn_att_large.pth: pre-trained model-large

  • cnn_att_less.pth：pre-trained model-less

  • CRISPRclassify_CNN_Att.pkl : stacking model

    [!NOTE]

    Due to file size limitations, we have stored the files cnn_att_large.pth,cnn_att_less.pth,andCRISPRclassify_CNN_Att.pkl at the following link: Google Drive Folder.

  • test.py : model testing

  • test.xlsx: test data

• README.md: project description and instructions.

• requirements.txt: list of dependencies for the project.

1. Clone the repository:

   git clone https://github.com/Xingyu-Liao/CRISPRclassify-CNN-Att.git
   cd CRISPRclassify-CNN-Att

2. Create and activate a virtual environment :

   conda create --name crisprclassify-cnn-att
   conda activate crisprclassify-cnn-att

3. Install dependencies:

   pip install -r requirements.txt

1. Train model-large for the subtype with abundant samples:

   cd source
   python CNN_Att.py

2. Train model-less for the subtype with fewer samples:

   cd source
   python transferlearning.py

3. Model stacking:

   cd source
   python stacking.py

4. Model testing:

   cd model
   python test.py

CRISPRclassify-CNN-Att/
├── data/
│ ├── repeats_all.csv
├── source/
│ ├── CNN_Att.py
│ ├── dataselect.py
│ ├── repeatsEncoder.py
│ ├── structure_features.py
│ ├── transferlearning.py
│ ├── typeEncoder.py
│ ├── stacking.py
├── model/
│ ├── cnn_att_large.pth │ ├── cnn_att_less.pth │ ├── CRISPRclassify_CNN_Att.pkl │ ├── test.py
│ ├── test.xlsx
├── README.md
├── requirements.txt