DeepLigType: Predicting Ligand Types of Protein-Ligand Binding Sites Using a Deep Learning Model

The paper "DeepLigType: Predicting Ligand Types of Protein-Ligand Binding Sites Using a Deep Learning Model" is currently under review. More details will be provided shortly. Stay tuned!

Citation

Stay tuned! (Under Review)

Files

Link - Download Files => The Molcache file and trained models (.pth) can be found via the link.

Requirements & Installation

Please note that certain libraries, such as libmolgrid, are not compatible with Windows. To run the program, you'll need an operating system with a Linux kernel. Additionally, It should be executed on a GPU-equipped machine.

The below libraries and their versions ensure the program functions without issues. The latest library versions will also work without any problems. If you face any issues, report them by opening an issue.

Python 3.10.12
torch==2.1.0
torchvision==0.16.0
torchaudio==2.1.0
biopython==1.81
molgrid==0.5.3
scikit-learn==1.3.1
fpocket==4.1
tqdm==4.66.1

As a second option, Google Colab is a good choice if you want to use our program without needing a specific operating system or GPU on your local machine. If you use Google Colab, you can open 'Run_Colab.ipynb' from your Google Drive and follow the instructions within.

Prediction

python '.../DeepLigType/predict.py' -p '.../DeepLigType/xxxx.pdb' -t 'trainedmodel'

Example:

 python '.../DeepLigType/predict.py' -p '.../DeepLigType/5xpp.pdb' -t "CNN_2023-08-30_acc_0.717969_59.26.pth" 

Evaluation

!python '.../DeepLigType/evaluation.py' -t ".../trained_model_pth"

Example:

 python '.../DeepLigType/evaluation.py' -t "CBAM_2023-08-29_acc_0.915781_74.16.pth" 

Training

python '.../DeepLigType/train.py' -m "deeplearning_model_name"

Example:

 python '.../DeepLigType/train.py' -m "cbam" 

Visualization

Visualization is coming! Stay tuned!

Running DeepLigType: Video