OS-CVD Multi-Agent Explorer

A LangChain + LlamaIndex powered multi-agent system to investigate the biological, molecular, and pharmacological associations between Oxidative Stress (OS) and Cardiovascular Disease (CVD).

---

📌 Project Architecture

            ┌────────────────────────────┐
            │       User Query           │
            └────────────┬───────────────┘
                         ▼
            ┌────────────────────────────┐
            │       PubMed Agent         │
            │  ↳ Literature search + QA  │
            └────────────┬───────────────┘
                         ▼
     ┌──────────────┬──────────────┬──────────────┐
     ▼              ▼              ▼              ▼
Protein Agent   Pathway Agent   Drug Agent   (Future: Genomics, etc.)
     └──────────────┴──────────────┴──────────────┘
                         ▼
            ┌────────────────────────────┐
            │      Summarizer Agent      │
            │   ↳ Final scientific report│
            └────────────────────────────┘

Each agent is autonomous, uses domain-specific vector tools, and works with LLM-based reasoning and memory.

---

🧠 Agents

Agent Name	Role
pubmed-agent	Queries biomedical literature to extract OS–CVD associations
protein-agent	Identifies relevant proteins and their functional roles
pathway-agent	Maps involved signaling or metabolic pathways
drug-agent	Extracts drugs related to OS/CVD mechanisms
summarizer-agent	Synthesizes all agent outputs into a final scientific report

---

📁 Folder Structure

ros-cvd-multiagent/
├── agents/                # LangChain-compatible agent nodes
├── tools/                 # LlamaIndex tool + index builder
├── memory/                # Global + per-agent memory
├── llama\_indexes/         # FAISS-based LlamaIndex indexes
├── data/                  # Chunked documents for agents
├── raw-docs/              # Raw files to build indexes from
├── orchestrator.py        # Executes routing logic
├── run-simulation.py      # CLI entry point
├── requirements.txt
└── README.md

---

🚀 How to Run

1. Install Dependencies

pip install -r requirements.txt

2. Add Raw Docs

Put .txt, .pdf, or .md files under:

raw-docs/pubmed/
raw-docs/protein/
raw-docs/pathway/
raw-docs/drug/

3. Build Indexes

python tools/index_builder.py

4. Run the System

python run-simulation.py

---

✅ Example Query

How does ROS contribute to cardiomyopathy and what proteins, pathways, and drugs are involved?

---

📜 License

MIT License — free to use, modify, and extend.