Code to train a neural network to identify mutation "hotspots" in a protein sequence. Details on the dataset and experiments can be found in src/experiment_driver.py.

1. Clone the repository

git clone https://gitlab.com/dellacortelab/mutation-hotspot-detection.git

2. Build the docker container

cd mutation-hotspot-detection && docker build -t mutation-hotspot -f docker/Dockerfile .

3. Run an interactive docker container (replacing /path/to/data and /path/to/logs with the desired local paths)

docker run -it --rm --name myname-hotspot -v $(pwd):/code -v /path/to/data:/data -v /path/to/logs:/logs mutation-hotspot

4. Inside the docker container, run

cd code/src 
python3 experiment_driver.py

There is no command line interface set up yet, so to change options, you can tweak them in the definition of the hotspot_experiment function in src\experiment_driver.py.