AstroAgents: Multi-Agent AI for Hypothesis Generation from Mass Spectrometry Data

Overview

AstroAgents is a large language model-based, multi-agent AI system for hypothesis generation from mass spectrometry data. With upcoming sample return missions across the solar system and the increasing availability of mass spectrometry data, AstroAgents addresses the urgent need for methods that analyze such data within the context of existing astrobiology literature and generate plausible hypotheses regarding the emergence of life on Earth.

<div align="center"> <img src="img/Fig1.png" alt="AstroAgents System Architecture" width="800px"> <p><em>Figure 1: AstroAgents System Architecture</em></p> </div>

System Architecture

AstroAgents is structured around eight collaborative agents:

Data Analyst: Interprets the mass spectrometry data
Planner: Delegates specific data segments to the scientist agents
Domain Scientists (3): Perform in-depth exploration of data segments
Accumulator: Collects and deduplicates the generated hypotheses
Literature Reviewer: Identifies relevant literature using Semantic Scholar
Critic: Evaluates the hypotheses, offering rigorous suggestions for improvement

Results

An astrobiology expert evaluated the novelty and plausibility of more than a hundred hypotheses generated from data obtained from eight meteorites and ten soil samples. Surprisingly, 36% were identified as plausible, and among those, 66% were novel.

<div align="center"> <img src="img/hypotheses.png" alt="Generated Hypotheses" width="800px"> <p><em>Generated Hypotheses from Mass Spectrometry Data</em></p> </div>

Usage Instructions

Prerequisites

Install the required Python packages:

pip install -r requirements.txt

Paper Context Preparation

Before running AstroAgents, you need to prepare paper context from research papers:

Create the necessary directories:

mkdir -p papers/md

Place your PDF research papers in the papers/ directory
Run the extraction script:

python extract_text.py

This will convert all PDFs to markdown format in the papers/md/ directory, which will be used by AstroAgents.

Running AstroAgents

Basic usage:

python AstroAgents.py --llm_model claude --iterations 10 --anthropic_api_key YOUR_ANTHROPIC_API_KEY --semantic_scholar_api_key YOUR_SEMANTIC_SCHOLAR_API_KEY

Command-line arguments:

| Argument | Description | Default | |----------|-------------|---------| | --paper_context_file | Path to the paper context file | paper_context.md | | --input_prompt_file | Path to the input prompt file | prompt.txt | | --llm_model | LLM model to use (claude or gemini) | claude | | --iterations | Number of iterations to run | 10 | | --anthropic_api_key | Anthropic API key | ANTHROPIC_API_KEY | | --google_api_key | Google API key | GOOGLE_API_KEY | | --semantic_scholar_api_key | Semantic Scholar API key | SEMANTIC_SCHOLAR_API_KEY |

Python Requirements

anthropic>=0.7.0
langchain>=0.1.0
langchain-anthropic>=0.1.0
langchain-google-genai>=0.0.5
google-generativeai>=0.3.0
colorama>=0.4.6
requests>=2.31.0
pymupdf4llm>=0.1.0

Citation

If you use AstroAgents in your research, please cite our paper:

@misc{saeedi2025astroagentsmultiagentaihypothesis,
      title={AstroAgents: A Multi-Agent AI for Hypothesis Generation from Mass Spectrometry Data}, 
      author={Daniel Saeedi and Denise Buckner and Jose C. Aponte and Amirali Aghazadeh},
      year={2025},
      eprint={2503.23170},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2503.23170}, 
}

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

The AstroAgents project website: https://astroagents.github.io/
Semantic Scholar for their API access

AstroAgents

Install / Use

README