---

🦕 STEGGosaurus-Docker

This is a working Docker implementation of STEGG. The long-term goal is to make the Docker image available via Docker Hub and host a web server so others can use the tool easily—without ever needing to look at any code.

---

Quick Start

1. Download Required Files

Before building the Docker container, you must download two sets of files:

MHCFlurry2.0 Frequency Matrices

Download the frequency matrix CSV from this link: MHCFlurry2.0 Frequencies (Box)

Place the CSV file into the following directory:

Ape-Gen2.0-main/helper_files/

⚙️ KORP Binary

Download korp6Dv1.bin from this link: Download KORP

Copy the binary file to the following directory:

RCD_required_files/

---

Pytorch models

If you want to pack the models into the docker image, download them and place them in the models/ directory.

wget -O models/esm_ppi_650m_tcr.pth https://zenodo.org/record/12602915/files/esm_ppi_650m_tcr.pth && \
wget -O models/tfold_pmhc_trunk.pth https://zenodo.org/record/12602915/files/tfold_pmhc_trunk.pth && \
wget -O models/tfold_tcr_trunk.pth https://zenodo.org/record/12602915/files/tfold_tcr_trunk.pth

🐳 Building & Running the Docker Container

Once dependencies are in place:

1. Build the Docker container

   run build_image.sh and start_image.sh

2. Run the model inside the container:

   Navegate to:

   /home/STEGG_controler/
   

   And run

   python3 model_complex.py demo.json

3. Model Additional Complexes To model other TCR–pMHC complexes, create a new JSON input file following the same structure as demo.json, and run:

   python3 model_complex.py your_input.json

---