MDML_PeptPerm is a tool for predicting the membrane permeability coefficient of cyclic peptides by combining descriptors obtained from molecular dynamics simulations with 2D descriptors.
This repository provides the implementation used in the associated publication.
MDML_PeptPerm/
├── README.md
├── requirements.txt
├── environment.yml
├── environment_cuda.yml
├── main.ipynb # Model training and cross-validation
├── ex_test.ipynb # External test evaluation
├── data/ # Training and external datasets
├── predicted/ # Model prediction outputs
├── weight/ # Saved model weights
└── utils/ # Utility scripts and helper functions
```sh
git clone https://github.com/akiyamalab/MDML_PeptPerm.git
cd MDML_PeptPerm
```
mamba env create -f environment.yml
mamba activate MDMLpeptmamba create -n MDMLpept python=3.9.6 -y
mamba activate MDMLpept
mamba install -c conda-forge $(cat requirements.txt) -yAfter activating the environment:
jupyter notebook- Run
main.ipynbfor model training and cross-validation. - Run
ex_test.ipynbfor external evaluation.
If you use this code in your research, please cite:
Author(s). Title. Journal. Year. DOI
See the LICENSE file for details.