Skip to content

akiyamalab/MDML_PeptPerm

Repository files navigation

MDML_PeptPerm

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

Installation

1. Clone the repository

```sh
git clone https://github.com/akiyamalab/MDML_PeptPerm.git
cd MDML_PeptPerm
```

2. Create and activate a conda/mamba environment (recommended)

Option A — Using environment.yml

mamba env create -f environment.yml
mamba activate MDMLpept

Option B — Manual installation

mamba create -n MDMLpept python=3.9.6 -y
mamba activate MDMLpept
mamba install -c conda-forge $(cat requirements.txt) -y

Usage

After activating the environment:

jupyter notebook
  • Run main.ipynb for model training and cross-validation.
  • Run ex_test.ipynb for external evaluation.

Citation

If you use this code in your research, please cite:

Author(s). Title. Journal. Year. DOI

License

See the LICENSE file for details.

About

Tool for predicting the membrane permeability coefficient of cyclic peptides by combining descriptors obtained from molecular dynamics simulations with 2D descriptors.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors