DeepPath

Physics-guided deep learning framework that generates protein transition pathway by active learning.

---

This repository contains code and pretrained models for exploring and constructing protein trajectories as demonstrated in the paper: https://www.biorxiv.org/content/10.1101/2025.02.27.640693v1 The main entry point is run_experiment.py, which loads a saved model from a specific experiment folder and generates predicted structural pathways.

Installation

We recommend using conda to manage dependencies.

Clone the repo and create the environment:

git clone https://github.com/yourusername/protein-path-explorer.git
cd protein-path-explorer
conda env create -f environment.yml
conda activate protein-path-explorer

Usage

1. Navigate into an experiment folder:

cd exps/exp1

2. Run the script, pointing to the models in the current experiment directory:

python ../../run_experiment.py --iter ITER [--npath NPATH]

• --iter (required): models from which iteration of training (e.g., 50)
• --npaths (optional): how many paths to sample (default: 1)

3. The predicted paths will be output as .dcd trajectory files named:

explorer{iter}-0.dcd, explorer{iter}-1.dcd, ...

License

The pretrained models are provided with MIT License.