XenoSignal

Code and notebooks for the paper: “XenoSignal: Investigating Intra- and Inter-Species Ligand–Receptor Interactions Using AlphaFold3.” Aly O. Abdelkareem, Courtney F. Hall, Athish Marutharaj, Kiran Narta, Theodore B. Verhey & A. Sorana Morrissy bioRxiv: https://www.biorxiv.org/content/10.1101/2025.08.13.670200v1

---

What this repo does

XenoSignal builds cross-species ligand–receptor (LR) pairs from CellChat interactions, retrieves the corresponding protein sequences/IDs, and prepares AlphaFold3 (AF3) jobs to model LR complexes. It then computes binding features (BindCraft, PRODIGY, ipSAE) and analyzes them to produce the manuscript’s supplementary figures and tables.

---

Repository layout

XENOSIGNAL/
├─ Data/
│  ├─ cellchat_data/        # CellChat LR interactions (input)
│  ├─ orthologs/            # gene orthology tables (input)
│  └─ xenosignal_jobs.zip   # AF3 job package (output of N1)
├─ IPSAE/
│  ├─ ipsae.py
│  ├─ run_ipsae.ps1         # PowerShell runner for ipSAE
│  └─ README.md
├─ notebooks/
│  ├─ N1_CellChat_XenoSignal.ipynb
│  ├─ N2_binding_features_generate.ipynb
│  └─ N3_Binding_Features_Analysis.ipynb
├─ xeno/                    # Python helpers (I/O, features, plotting, utils)
│  ├─ biopython_utils.py
│  ├─ features.py
│  ├─ io.py
│  ├─ pl.py
│  └─ pyrosetta_utils.py    # optional utilities (not required for main workflow)
├─ LICENSE
└─ README.md

---

Requirements

• Python ≥ 3.10
• Python packages: pandas, polars, numpy, scipy, scikit-learn, biopython, matplotlib
• Optional (for some utilities): pyrosetta (not required for reproducing the figures)
• AlphaFold3 (external; supply your own AF3 binaries/resources)
• BindCraft and PRODIGY (called via wrappers under ./Xeno source code/)
• ipSAE (invoked via IPSAE/run_ipsae.ps1; requires PowerShell on Windows or adapt for your OS)

External tools (AF3, BindCraft, PRODIGY, ipSAE) are not bundled here. Please install/obtain them per their licenses and set paths accordingly.

---

Quickstart

1. Create environment

   conda create -n xenosignal python=3.10 -y
   conda activate xenosignal
   pip install pandas polars numpy scipy scikit-learn biopython matplotlib

2. Prepare inputs

   • Place CellChat interaction tables under ./Data/cellchat_data/.
   • Place ortholog mapping tables under ./Data/orthologs/.
   • Ensure wrappers and binaries for BindCraft/PRODIGY are accessible under ./Xeno source code/.
   • Make ipSAE available (e.g., run via IPSAE/run_ipsae.ps1).

3. Run the notebooks in order

   • N1_CellChat_XenoSignal.ipynb Builds cross-species LR pairs (mouse→human, human→mouse) using orthologs, resolves protein IDs/sequences, and packages AF3 jobs to ./Data/xenosignal_jobs.zip.

   • N2_binding_features_generate.ipynb Runs BindCraft & PRODIGY via ./Xeno source code/, executes ipSAE via IPSAE/run_ipsae.ps1, parses outputs, and merges features to ./Data/binding_features.csv.

   • N3_Binding_Features_Analysis.ipynb Produces the manuscript visuals:

     • Supplementary Figure 1: clustering dendrogram + feature–feature correlation heatmap
     • Supplementary Figure 2: feature distributions colored by cluster assignment Saved to ./Figures/binding_features/.

---

Configuration tips

• If your tool paths aren’t on PATH, set an environment variable, e.g.:

  export XENO_TOOLS_DIR="/path/to/Xeno source code"

• AlphaFold3 inputs are generated as a zipped job bundle: ./Data/xenosignal_jobs.zip. Unzip/submit to your AF3 infrastructure as appropriate.

• The ipSAE step can be run outside the notebook:

  PowerShell -ExecutionPolicy Bypass -File IPSAE/run_ipsae.ps1

  Make sure it writes outputs where N2_binding_features_generate.ipynb expects them.

---

Citing XenoSignal

If you use this code or figures, please cite:

Abdelkareem AO, Hall CF, Marutharaj A, Narta K, Verhey TB, Morrissy AS. XenoSignal: Investigating Intra- and Inter-Species Ligand–Receptor Interactions Using AlphaFold3. bioRxiv (2025). doi:10.1101/2025.08.13.670200

BibTeX

@article{Abdelkareem2025XenoSignal,
  title   = {XenoSignal: Investigating Intra- and Inter-Species Ligand–Receptor Interactions Using AlphaFold3},
  author  = {Abdelkareem, Aly O. and Hall, Courtney F. and Marutharaj, Athish and Narta, Kiran and Verhey, Theodore B. and Morrissy, A. Sorana},
  journal = {bioRxiv},
  year    = {2025},
  doi     = {10.1101/2025.08.13.670200},
  url     = {https://www.biorxiv.org/content/10.1101/2025.08.13.670200v1}
}

---

License

See LICENSE in this repository.

---

Contact

For questions or issues, please open a GitHub issue or contact the corresponding author listed in the manuscript.