Phytotherapy Discovery — Ontology Work Toolkit

This repository accompanies the analysis for “Phytotherapy Discovery through Ontology-Guided Machine Learning.”
It is intended to be used with the upstream phytotherapy ontology maintained in RomanoLab/poppy (the source of truth for the ontology itself):

• Upstream ontology + primary project context: https://github.com/RomanoLab/poppy

This repository also includes a modified CTAPred pipeline used to predict protein–phytochemical interactions, supporting downstream mechanism inference and repurposing analyses.

---

Table of Contents

1. Background
2. Repository Layout
3. Quick Start
4. Command-line Reference
5. Workflow Details
6. Extending the Toolkit
7. Contributing
8. Citation

---

Background

Researchers in natural-product drug discovery often work with ontologies linking plants → chemicals → targets → therapeutic roles. In this project, the ontology is maintained upstream in RomanoLab/poppy, while this repository contains analysis companion materials and the CTAPred workflow used in the ontology-guided ML portion of the study.

At a high level, the workflow supports:

• Starting from an ontology-derived phytochemical set (from poppy)
• Predicting likely protein targets for phytochemicals (CTAPred)
• Using predictions for downstream analysis (e.g., candidate prioritization, mechanism inference)

---

Repository Layout

.
├── README.md                    # you are here
├── requirements.txt             # repo-level environment dependencies (if applicable)
└── CTAPred/                     # modified CTAPred pipeline for protein–phytochemical predictions
    ├── predict_targets.py
    ├── SharedFunc.py
    ├── requirements.txt
    ├── QueryList1_smiles.csv
    └── README.md                # CTAPred usage instructions

---

| File / Dir             | Purpose                                                                      |
| ---------------------- | -----------------------------------------------------------------------------|
| `CTAPred/`             | Modified CTAPred pipeline used to predict protein–phytochemical interactions |
| `CTAPred/README.md     | Canonical usage instructions for running CTAPred in this repo                |
| `Crequirements.txt`    | Repo-level dependencies used by analysis utilities                           |

---

Quick Start

1. Clone

git clone https://github.com/RomanoLab/phytotherapy-discovery.git
cd phytotherapy-discovery

2. (Recommended) Create a virtual environment

python -m venv .venv
source .venv/bin/activate      # Windows: .venv\Scripts\activate

3. Install dependencies

If you are running CTAPred, install CTAPred’s pinned dependencies:

pip install -r CTAPred/requirements.txt

If you are running other repo-level analysis utilities, you may also install the root requirements:

pip install -r requirements.txt

4. Run CTAPred

Follow the instructions in CTAPred/README.md. To see the CLI options:

python CTAPred/predict_targets.py --help

CTAPred/QueryList1_smiles.csv is provided as an example/input artifact containing phytochemical SMILES used for predictions.

---

Command-line Reference

Because this repository’s executable workflow is centered on CTAPred, the authoritative CLI reference is:

python CTAPred/predict_targets.py --help

For usage examples, input expectations, and output files, see:

• CTAPred/README.md

---

Workflow Details

A typical end-to-end workflow looks like this:

1. Consult / generate ontology artifacts upstream Use RomanoLab/poppy to access the ontology and (if needed) derive a phytochemical set for ML analysis.
2. Prepare CTAPred inputs Provide a CSV of phytochemical SMILES (use CTAPred/QueryList1_smiles.csv as a template).
3. Run CTAPred predictions Execute CTAPred/predict_targets.py to generate predicted protein targets / ranked interactions.
4. Downstream analysis / integration Use predictions for prioritization, hypothesis generation, or reintegration into ontology/KG analyses (depending on your study design and tooling).

---

Extending the Toolkit

Ideas that keep this repository aligned with its current scope:

• Modularise – split into smaller modules and expose a proper Python API.
• Unit tests – add pytest and GitHub Actions for continuous integration.
• Visualisation – integrate network diagrams or heatmaps (e.g., via matplotlib).
• Docker – package the pipeline for reproducible runs on any machine.

---

Contributing

1. Fork → create a feature branch → commit + push → open a Pull Request.
2. Keep changes scoped and well-described (what/why/how).
3. If you change model inputs/outputs, update CTAPred/README.md and this README accordingly.

---

Citation

If you use this repository in research, please cite:

@software{hewryk_ontology_work_2025,
  author       = {Hewryk, Oresta S. I. and Pan, Ian Tong and Romano, Joseph D.},
  title        = {{Ontology Work}: A Python toolkit for phytotherapy ontology enrichment},
  year         = {2025},
  publisher    = {GitHub},
  url          = {https://github.com/RomanoLab/ontology-work}
}