Last updated: December 10, 2025
Phonos incorporates and uses various third-party machine learning models, libraries, and software components. We are grateful to the researchers, developers, and organizations who have made their work available to the community. This page provides attribution and license information for all third-party components used in our platform.
Description: Protein structure prediction from amino acid sequences
Creators: Meta Fundamental AI Research Protein Team (FAIR)
License: Code is licensed under Apache 2.0; Model parameters are licensed under CC BY 4.0
Citation: Lin, Zeming et al. "Evolutionary-scale prediction of atomic-level protein structure with a language model." Science 379.6637 (2023): 1123-1130.
Repository: https://github.com/facebookresearch/esm
Description: Molecular docking for predicting protein-ligand binding poses
Creators: Gabriele Corso, Hannes Stark, Bowen Jing, Regina Barzilay, and Tommi Jaakkola (MIT)
License: MIT License
Citation: Corso, Gabriele et al. "DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking." International Conference on Learning Representations (ICLR) 2023.
Repository: https://github.com/gcorso/DiffDock
Description: High-accuracy protein structure prediction
Creators: DeepMind (Google)
License: Code is licensed under Apache License 2.0; Model parameters are licensed under CC BY 4.0 (changed from CC BY-NC 4.0 to allow commercial use)
Citation: Jumper, John et al. "Highly accurate protein structure prediction with AlphaFold." Nature 596.7873 (2021): 583-589.
Repository: https://github.com/google-deepmind/alphafold
Description: Deep learning-based protein sequence design
Creators: Justas Dauparas et al., Baker Lab, University of Washington
License: Code is licensed under MIT License; Published work is licensed under CC BY 4.0
Citation: Dauparas, Justas et al. "Robust deep learning-based protein sequence design using ProteinMPNN." Science 378.6615 (2022): 49-56.
Repository: https://github.com/dauparas/ProteinMPNN
Description: Metal ion binding site prediction using 3D convolutional neural networks
Creators: Laboratory of Computational Biology and Chemistry (LCBC), EPFL
License: Code is licensed under MIT License; Network weights are licensed under CC BY 4.0
Citation: Torrisi, Mirko et al. "Metal3D: a general deep learning framework for accurate metal ion location prediction in proteins." Nature Communications 14 (2023): 2606.
Repository: https://github.com/lcbc-epfl/metal-site-prediction
Description: Biomolecular structure prediction for proteins, nucleic acids, and small molecules
Creators: MIT CSAIL (Computer Science and Artificial Intelligence Laboratory)
License: MIT License
Citation: Wohlwend, Jeremy et al. "Boltz-1: Democratizing Biomolecular Interaction Modeling." bioRxiv (2024).
Repository: https://github.com/jwohlwend/boltz
Description: DNA sequence generation and analysis using foundation models
Creators: Arc Institute and Together AI
License: Apache License 2.0
Citation: Nguyen, Eric et al. "Sequence modeling and design from molecular to genome scale with Evo." Science (2024).
Repository: https://github.com/evo-design/evo
Description: GPU compute infrastructure for AI/ML model execution
Provider: Nebius AI
Purpose: Provides GPU compute resources necessary for running AI/ML models including Boltz2, Evo2, and other computational models
Location: European Union
Website: https://nebius.ai/
Description: GPU compute infrastructure used exclusively for demonstration and testing purposes
Provider: NVIDIA Corporation
Purpose: Demo environment for showcasing AI/ML capabilities. Not used for production user data processing.
Location: USA
Website: https://www.nvidia.com/en-us/data-center/dgx-cloud/
Description: Conversational AI assistance for project chat, file analysis, and AI-powered search functionality (optional feature, activated via @gemini mentions or search usage)
Provider: Google LLC
Terms of Service: https://policies.google.com/terms
Privacy Policy: https://policies.google.com/privacy
Usage: This is an optional feature. When you use Gemini features (by @mentioning gemini in chats or using the search functionality), your query text, search queries, project context, and optionally file content (limited to 10,000 characters) are sent to Google's Gemini API for processing. Google processes this data according to their terms and privacy policy.
Data Location: Google may process requests on servers worldwide, including outside the European Union
User Control: You have full control over this feature. Simply avoid using @gemini mentions or the search functionality if you prefer to keep all data within our EU-based infrastructure. All core protein analysis features work without Gemini integration.
Description: Website analytics for understanding user behavior and improving the service (optional, requires cookie consent)
Provider: Google LLC
Privacy Policy: https://policies.google.com/privacy
Data Collected: Anonymized IP addresses, page views, browser information, session data
User Control: Google Analytics is only loaded if you accept cookies via the cookie consent banner. You can withdraw consent at any time.
Description: Open-source machine learning framework
Creators: Meta Platforms, PyTorch Foundation, The Linux Foundation
License: Modified BSD License (BSD 3-Clause)
Website: https://pytorch.org/ | GitHub
Description: State-of-the-art Natural Language Processing and Machine Learning library
Creators: Hugging Face Team
License: Apache License 2.0
Website: https://huggingface.co/transformers | GitHub
Description: Open-source cheminformatics and machine learning software
Creators: RDKit Contributors
License: BSD 3-Clause License
Website: https://www.rdkit.org/ | GitHub
Description: Lightweight WSGI web application framework
Creators: Pallets Projects
License: BSD 3-Clause License
Website: https://flask.palletsprojects.com/
Description: In-memory data structure store, used as database, cache, and message broker
Creators: Redis Ltd.
License: BSD 3-Clause License
Website: https://redis.io/
A permissive license that allows you to use, modify, and distribute the software for both commercial and non-commercial purposes. It provides an express grant of patent rights from contributors. Attribution is required.
A permissive license that is short and to the point. It lets people do almost anything they want with your project, including making and distributing closed source versions, as long as they include the original copyright and license notice.
A permissive license similar to the MIT License but with an additional clause that prohibits others from using the name of the project or its contributors to promote derived products without written consent.
This license lets others distribute, remix, adapt, and build upon the work, even commercially, as long as they credit the original creation. This is the most accommodating of licenses offered for creative works.
Citation Requirements: If you use outputs from our service in academic publications or research, you should cite the relevant model papers as listed above, in addition to any citation of the Phonos platform itself.
License Compliance: Your use of the Service and its outputs must comply with all applicable third-party license terms. We make no representations or warranties regarding third-party models and technologies.
Commercial Use: All models listed above are available for commercial use under their respective licenses. However, you are responsible for ensuring compliance with all license terms.
For questions about licensing or attribution, please contact us at: help@fyreflysystems.de