Repository of "Towards improving full-length ribosome density prediction by bridging sequence and graph-based representations". The dataset files can be found here: https://zenodo.org/records/12180308

Inside the interpret subfolder one can find the code to run the Plotly server to intepret the RiboGL predictions.

In order to run the server:

1. Install the dependecies in a conda envirorment ribogl (arbitrary name):

   conda env create -n ribogl -f interpret/requirements.yaml

2. Download the data from https://zenodo.org/records/12180308 into the data subfolder.
3. Extract RiboGL_Attributions.h5.gz into RiboGL_Attributions.h5 using gzip -d RiboGL_Attributions.h5.gz. The uncompressed size will be around 43 GB.
4. Run the server:

   conda activate ribogl
   python interpret/app.py --port 8050

   The server will be available at https://localhost:8050/. The port can be changed with the port argument.