Skip to content
BranchesTags

Repository files navigation

OriGene: A Self-Evolving Virtual Disease Biologist Automating Therapeutic Target Discovery

Important: This is the OriGene, an open‑source, self-evolving multi-agent system that acts as a virtual disease biologist. We also introduce the TRQA Benchmark — a benchmark of 1,921 expert-level questions for evaluating biomedical AI agents.

What's New

1.Public online Launch – OriGene is now live and available to try at https://origene.lglab.ac.cn/.

2.Open‑Source Release – The entire OriGene codebase and benchmark is now available, Fork away!

3.Officially presented at the 2025 World Artificial Intelligence Conference (WAIC).

1. OriGene Overview

Therapeutic target discovery remains one of the most critical yet intuition-driven stages in drug development. We present OriGene, a self-evolving multi-agent system that functions as a virtual disease biologist to identify and prioritize therapeutic targets at scale.

2. Getting Started

Running OriGene requires two components: the MCP Server (OrigeneMCP) and the OriGene agent itself. We recommend setting up the MCP Server first, then configuring and running OriGene.

Step 1: Deploy the MCP Server

OriGene relies on the MCP Server (OrigeneMCP), which aggregates more than 600 bioinformatics tools. The MCP server provides access to multiple databases including ChEMBL, PubChem, FDA, OpenTargets, NCBI, UniProt, PDB, Ensembl, UCSC, KEGG, STRING, TCGA, Monarch, ClinicalTrials, and more.

Step-by-step deployment:

a. Clone the OrigeneMCP repository:

git clone https://github.com/GENTEL-lab/OrigeneMCP.git
cd OrigeneMCP

b. Install dependencies (requires uv package manager):

# Install uv if you haven't already
pip install uv

# Create virtual environment
uv venv .venv --python=3.13
source .venv/bin/activate

# Install dependencies
uv sync

c. Deploy the MCP service:

# Copy configuration file
cp default.conf.toml local.conf.toml

# (Optional) Configure API keys in local.conf.toml if you need:
# - Tavily search: get your API key from https://tavily.com/
# - Jina search: get your API key from https://jina.ai/

# Deploy the service
export PYTHONPATH=`fab pypath`
uv run -m deploy.web

Note:

  • The MCP server runs on port 8788 by default. If this port is occupied, modify the port in local.conf.toml.
  • Keep the MCP server running in a separate terminal while using OriGene.
  • The server endpoint will be http://127.0.0.1:8788 (or your custom port).

For more detailed information about OrigeneMCP deployment and usage, please refer to the OrigeneMCP repository.

Step 2: Configure and Run OriGene

Both deployment methods require configuring API keys. Edit the configuration file src/local_deep_research/_settings/.secrets.toml and fill in your actual API keys and MCP server URL.

OriGene is model-agnostic, so you can freely switch between different base models.

Note: For LLM inference, you can configure either OpenAI/DeepSeek APIs or use CloseAI as a fallback. If OpenAI or DeepSeek is not configured, CloseAI will automatically be used instead. You can also configure all three if desired.

API Key Configuration

Important: All third-party API keys must be obtained directly from the respective service providers. This project does not provide any API keys. Users are responsible for managing their own API credentials and complying with each provider's terms of service.

Service Purpose Registration
SiliconFlow Text embedding siliconflow.cn
Volcano Engine Template embedding volcengine.com
OpenAI LLM inference openai.com
DeepSeek Reasoning model platform.deepseek.com
CloseAI LLM inference (fallback for OpenAI/DeepSeek) closeai-asia.com

Configuration template (src/local_deep_research/_settings/.secrets.toml):

[mcp]          
server_url = "Enter your mcp url"

[embedding]          
api_key = "Enter your api key (match url : https://api.siliconflow.cn/v1/embeddings)"
cache   = "embedding_cache.pkl"

[template]
api_base = "https://ark.cn-beijing.volces.com/api/v3"
api_key  = "Enter your api key"


[openai]             
api_base = "https://api.openai.com/v1"
api_key  = "Enter your api key"

[deepseek]
api_base = "https://api.deepseek.com"
api_key  = "Enter your api key"

# CloseAI (optional, fallback for OpenAI and DeepSeek)
# If OpenAI or DeepSeek API is not configured, CloseAI will be used as fallback
# You can configure only CloseAI, or configure both CloseAI and OpenAI/DeepSeek
[closeai]
api_base = "https://api.openai-proxy.org/v1"
api_key  = "Enter your api key"

OriGene supports two deployment methods: Docker (recommended) and Native Installation. Choose the one that suits your environment.

Option A: Quick Start with Docker (Recommended)

For the easiest setup experience, we provide Docker-based deployment that handles all dependencies automatically.

Prerequisites:

One-Command Deployment:

# Clone and setup in one command
git clone https://github.com/GENTEL-lab/OriGene.git
cd OriGene
./setup.sh

Or step by step:

  1. Clone the repository:

    git clone https://github.com/GENTEL-lab/OriGene.git
cd OriGene

  2. Configure API keys and MCP connection:

    # Edit the configuration file with your API keys
nano src/local_deep_research/_settings/.secrets.toml
# or use your preferred editor (vim, code, etc.)

    Update the [mcp] section with your MCP server URL:

    [mcp]          
server_url = "http://host.docker.internal:8788"

    MCP Server Access in Docker: Since OriGene runs inside a Docker container, it cannot use 127.0.0.1 to reach the host MCP server. Update the MCP server URL as follows:

    • For Mac/Windows: Use http://host.docker.internal:8788
    • For Linux: Use http://172.17.0.1:8788
    • Or deploy MCP in Docker and use service name

  3. Build and verify environment:

    # Quick check script (recommended for first-time setup)
./docker-check.sh

# Or manually build and run checks
docker-compose build
docker-compose run --rm origene python -m local_deep_research.test.check_all

  4. Run OriGene:

    Using Makefile (Recommended):

    # Run environment check
make check

# Interactive mode
make start

# Quick research
make quick QUERY="What are therapeutic targets for Alzheimer's disease?"

# Detailed research
make detailed QUERY="Analyze molecular mechanisms of EGFR in lung cancer"

# Run tests
make test-e2e

# See all available commands
make help

    Using docker-compose directly:

    # Interactive mode
docker-compose run --rm origene

# Quick research (CLI)
docker-compose run --rm origene python -m local_deep_research.main "your query here" --mode quick

# Detailed research (CLI)
docker-compose run --rm origene python -m local_deep_research.main "your query here" --mode detailed

# Run end-to-end test
docker-compose run --rm origene python -m local_deep_research.test.test_example

    Important Notes:

    • All logs will be saved to the ./logs directory
    • Embedding cache will be stored in the ./cache directory

Option B: Native Installation (uv / conda)

If you prefer native installation without Docker, follow the instructions below.

  1. Clone the repository:

    git clone https://github.com/GENTEL-lab/OriGene.git
cd OriGene

  2. Configure MCP connection and API keys:

    Edit the configuration file src/local_deep_research/_settings/.secrets.toml and fill in your MCP server URL and API keys (see API Key Configuration above).

    [mcp]          
server_url = "http://127.0.0.1:8788"  # Use your actual MCP server URL and port

    If you changed the port during MCP deployment, make sure to update the server_url accordingly.

  3. Install dependencies:

    cd src
uv sync

  4. Activate the virtual environment:

    source ./.venv/bin/activate

  5. (Optional) Add the project root to PYTHONPATH:

    export PYTHONPATH=$(pwd):$PYTHONPATH

  6. System Check & Testing:

    Before running OriGene, you can verify your environment setup using the built-in check scripts:

    Run all environment checks (recommended before first use):

    uv run -m local_deep_research.test.check_all

    This will check:

    • Python version and all dependencies
    • Configuration files and API keys
    • MCP server connection
    • Core module imports

    Run individual checks:

    # Check dependencies only
uv run -m local_deep_research.test.check_deps

# Check API keys and configuration
uv run -m local_deep_research.test.check_config

# Check MCP server connection
uv run -m local_deep_research.test.check_mcp

# Check module imports
uv run -m local_deep_research.test.check_modules

    Run end-to-end test (includes MCP tool call and LLM query):

    uv run -m local_deep_research.test.test_example

    Example output:

    ======================================================================
                    OriGene Environment Check
======================================================================

> testing dependencies...
> ✓ Dependencies ready (58 packages checked)

> checking LLM API keys...
> ✓ LLM API keys ready (5 APIs configured)

> checking MCP server connection...
> ✓ MCP tool list ready (4 checks passed)

> checking module imports...
> ✓ All modules ready (15 modules checked)

======================================================================
> ✓ SUCCESS! You can run OriGene now!

  7. Run OriGene:

    Launch the interactive assistant:

    uv run -m local_deep_research.main

    You will see a prompt similar to the following:

    Welcome to the Advanced Research System
Type 'quit' to exit

Select output type:
1) Analysis (few minutes, answers questions, summarizes findings)
2) Detailed Report (more time, generates a comprehensive report with deep analysis)
Enter number (1 or 2):

    After selecting an output type, enter your research query and OriGene will return the results.

Benchmark: Running and Scoring

Using Docker:

# Run benchmark evaluation
docker-compose run --rm origene python -m local_deep_research.evaluate_local

# Score the results (example for TRQA-lit-choice)
docker-compose run --rm origene python -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/TRQA_lit_choice/agent_answers_test.txt \
  --original_data benchmark/TRQA_lit_choice/TRQA-lit-choice-172-coreset.csv \
  --model_name "OriAgent"

Using Native Installation:

Run the benchmark to generate agent answers (you can use either command):

# From the project root (this directory), after activating the venv
uv run -m local_deep_research.evaluate_local

# Or using python
python -m local_deep_research.evaluate_local

Then score the generated results (replace paths if you changed dataset/output names):

# Example: score TRQA-lit-choice core set results
python local_deep_research/score_evaluation_results.py \
  --agent_results benchmark/TRQA_lit_choice/agent_answers_test.txt \
  --original_data benchmark/TRQA_lit_choice/TRQA-lit-choice-172-coreset.csv \
  --model_name "OriAgent"

# Or using uv
uv run -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/TRQA_lit_choice/agent_answers_test.txt \
  --original_data benchmark/TRQA_lit_choice/TRQA-lit-choice-172-coreset.csv \
  --model_name "OriAgent"

Other Benchmarks:

Using Docker:

# DbQA benchmark
docker-compose run --rm origene python -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/DbQA/agent_answers_testv2.txt \
  --original_data benchmark/DbQA/DbQA.csv \
  --model_name "OriAgent"

# GPQA benchmark
docker-compose run --rm origene python -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/GPQA/agent_answers_test.txt \
  --original_data benchmark/GPQA/GPQA-lit-choice.csv \
  --model_name "OriAgent"

# TRQA-lit short answer benchmark
docker-compose run --rm origene python -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/TRQA_lit_short_ans/agent_answers_test.txt \
  --original_data benchmark/TRQA_lit_short_ans/TRQA-lit-short-answer-1108.csv \
  --model_name "OriAgent"

Using Native Installation:

# DbQA benchmark
uv run -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/DbQA/agent_answers_testv2.txt \
  --original_data benchmark/DbQA/DbQA.csv \
  --model_name "OriAgent"

# GPQA benchmark
uv run -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/GPQA/agent_answers_test.txt \
  --original_data benchmark/GPQA/GPQA-lit-choice.csv \
  --model_name "OriAgent"

# TRQA-lit short answer benchmark
uv run -m local_deep_research.score_evaluation_results \
  --agent_results benchmark/TRQA_lit_short_ans/agent_answers_test.txt \
  --original_data benchmark/TRQA_lit_short_ans/TRQA-lit-short-answer-1108.csv \
  --model_name "OriAgent"

3. License

This code repository is licensed under the Creative Commons Attribution-Non-Commercial ShareAlike International License, Version 4.0 (CC-BY-NC-SA 4.0) (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at https://github.com/GENTEL-lab/OriGene/blob/main/LICENSE.

4. Contact

If you have any questions, please raise an issue or contact us at shuangjia.zheng@sjtu.edu.cn or zhongyuezhang@sjtu.edu.cn.

5. Acknowledgements

Thanks to DeepSeek, ChatGPT, Claude, and Gemini for providing powerful language models that made this project possible.

Special thanks to the human experts who assisted us in benchmarking and evaluating the agent's performance!

About

OriGene: A Self-Evolving Virtual Disease Biologist for Mechanism-Guided Therapeutic Target Discovery

Resources

Readme
License
Activity
Custom properties

Stars

218 stars

Watchers

5 watching

Forks

11 forks
Report repository

Releases

Packages

Contributors

Languages