feat: Add Docker support with PLIP and PyMOL integration

[jolakopec]

Summary

This PR adds Docker support to STCRpy, making it easier to run the complete pipeline including PLIP interaction profiling and PyMOL visualizations
without complex dependency management.

Changes

• Dockerfile: Base image with STCRpy + PLIP + PyMOL support
• Dockerfile.ml: ML variant with PyTorch, scikit-learn, and all visualization tools
• docker-compose.yml: Service orchestration for easy container management
• .dockerignore: Optimized build context
• Bug fix: Improved PLIP binding site ID tracking in interaction data

Features

• ✅ Full STCRpy functionality (fetch, load, save, geometry calculations)
• ✅ PLIP interaction profiling
• ✅ PyMOL 3D visualizations and heatmap generation
• ✅ ML dataset creation (Dockerfile.ml)
• ✅ All dependencies pre-installed (OpenBabel, Qt5, ANARCI models)

Testing

Tested against existing test suite:

• 52 out of 53 tests passing (98% pass rate)
• All README examples working
• Successfully generated interaction heatmaps and visualizations
• Verified with 5 TCR:pMHC complexes (8gvb, 7pb2, 6eqa, 4ftv, 4jff)

Usage

# Build and run base image
docker build -t stcrpy .
docker run --rm \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/output:/app/output \
  stcrpy python your_script.py

# Build ML variant
docker build -t stcrpy-ml -f Dockerfile.ml .
docker run --rm \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/output:/app/output \
  stcrpy-ml python your_script.py

# Or use docker-compose
docker-compose up

Example: Run interaction profiling
docker run --rm \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/output:/app/output \
  stcrpy python -c "
import stcrpy
tcr = stcrpy.fetch_TCRs('8gvb')[0]
tcr.profile_peptide_interactions()
tcr.get_interaction_heatmap(plotting_kwargs={'save_as': '/app/output/heatmap.png'})
"
# Heatmap saved to ./output/heatmap.png


Benefits

- No manual dependency installation
- Consistent environment across platforms
- Easy CI/CD integration
- Simplified onboarding for new users

[benjiemc]

Thank you for the PR Jola!

I have added a few commits to clean up the Docker setup a bit:

• There is now only 1 Dockerfile (regular and ML were merged)
• UV was removed as virtual environments aren't really necessary with Docker
• The documentation files were merged into the main README

Let us know if you disagree with any of these changes, and we can discuss.

@npqst could you review the changes in this commit concerning the PLIP IDs. I am less familiar with the interaction profiling.