About Prometheus
Prometheus is a citizen science initiative that uses AI and machine learning to accelerate scientific discovery. We believe that modern AI tools — protein language models, neural networks, large language models — should be pointed at real scientific problems.
Our Approach
We build automated research pipelines that can screen millions of data points for interesting signals. Then we share everything: code, data, results, and methodology.
Active Projects
🧬 AMPHunter
Antimicrobial resistance is one of the biggest threats to global health. We use Meta's ESM-2 protein language model to scan metagenomic sequences from extreme environments — hot springs, permafrost, deep-sea vents — for novel antimicrobial peptide candidates.
So far, we've identified 2,500+ computational candidates across 5 biomes. Each candidate goes through a multi-stage pipeline: ESM-2 scoring, biophysical filtering, hemolysis prediction, novelty assessment, and structure prediction.
🪐 ExoHunter
We're building a machine learning pipeline to detect exoplanet transit signals in TESS (Transiting Exoplanet Survey Satellite) light curve data. Our hybrid approach combines feature engineering with XGBoost classifiers and convolutional neural networks.
🌌 ExoExplorer
An interactive dashboard for exploring 5,700+ confirmed exoplanets. Visualize habitable zones, compare planetary systems, and design AI-generated alien ecosystems based on real planetary parameters.
Team
Prometheus is built by a small team using AI agents as research assistants. We use Claude (Anthropic), GPT (OpenAI), and various open-source models as collaborative tools in our research workflow.
Open Source
Everything is open source and publicly available:
- AMPHunter Pipeline — ESM-2 AMP screening pipeline
- ExoHunter — Exoplanet transit detection
- ExoExplorer — Interactive exoplanet dashboard
Contact
Interested in our work? Have questions about our candidates? Want to collaborate on experimental validation?
Reach out via GitHub or check our blog for the latest updates.