Show HN: An open-source implementation of AlphaFold3
AlphaFold3, an open-source project by Ligo Biosciences, aims to improve biomolecular structure prediction for single-chain proteins, with future expansions planned. It is currently in early development and not production-ready.
Read original article
AlphaFold3 is an open-source project developed by Ligo Biosciences aimed at enhancing biomolecular structure prediction, particularly for single-chain proteins, with future plans to include ligands, multimers, and nucleic acids. Currently in its early development stages, the project is not yet ready for production use. The repository features a demo video that illustrates the model training process and emphasizes optimizations for speed and memory efficiency to tackle training challenges. It acknowledges contributions from the AlphaFold3 team at Google DeepMind, the OpenFold project, and the ProteinFlow library. Users can experiment with the model by loading checkpoint weights, and beta testing is available for those interested in ligand-protein and nucleic acid predictions. The project encourages community involvement through contributions, bug reporting, and pull requests. It is licensed under the Apache License 2.0.
- AlphaFold3 focuses on predicting biomolecular structures, starting with single-chain proteins.
- The project is in early development and not yet production-ready.
- It includes optimizations for model efficiency and a demo video for training dynamics.
- Community contributions and beta testing opportunities are encouraged.
- The project is licensed under the Apache License 2.0.
Related
Are AlphaFold's new results a miracle?
AlphaFold 3 by DeepMind excels in predicting molecule-protein binding, surpassing AutoDock Vina. Concerns about data redundancy, generalization, and molecular interaction understanding prompt scrutiny for drug discovery reliability.
The Illustrated AlphaFold
The article discusses AlphaFold3's architecture for predicting protein structures, including Input Preparation, Representation Learning, and Structure Prediction. It highlights improvements like predicting complexed proteins and enriching representations with MSA and templates.
Rapid protein evolution by few-shot learning with a protein language model
Researchers introduce EVOLVEpro, a method merging PLMs and predictors to speed up protein evolution. It outperforms existing techniques, achieving significant property enhancements in just four rounds. The approach benefits various applications, showcasing AI's potential in protein engineering.
Accurate structure prediction of biomolecular interactions with AlphaFold 3
AlphaFold 3 introduces a diffusion-based architecture that enhances biomolecular interaction predictions, outperforming existing tools and improving data efficiency, with potential implications for drug discovery and structural biology.
Creating the largest protein-protein interaction dataset in the world
A-Alpha Bio, founded in 2017, is creating a large protein-protein interaction dataset using its AlphaSeq method, which employs genetically engineered yeast cells to measure interactions, enhancing drug development and biotechnology.
- There is excitement about the open-source nature of AlphaFold3, especially in contrast to DeepMind's closed-source direction.
- Questions arise about the verification of predictions and the necessity of experimental techniques post-prediction.
- Some commenters suggest the need for formal publication of the implementation to enhance credibility and reproducibility.
- Concerns are raised about the use of the "AlphaFold3" name and potential legal issues with DeepMind.
- Discussion includes the challenges of training data and compute resources, highlighting Google's advantages in this area.
12 commentsDeepMind and AlphaFold are clearly moving in a closed-source direction, since they created Isomorphic Labs as a division of Alphabet essentially focused on doing this stuff closed source. In theory it seems nice for academic tools to have an open source version, although I'm not familiar enough with this field to point to a specific benefit of it.
So what's your plan for the company itself, do you intend to continue working on this open source project as part of your business model, or was it more of a one-off? Your website seems very nonspecific about what exactly you intend to be selling.
Folding@Home https://en.wikipedia.org/wiki/Folding@home :
> making it the world's first exaflop computing system
Google has access to training compute on a scale perhaps nobody else has.
Related
Are AlphaFold's new results a miracle?
AlphaFold 3 by DeepMind excels in predicting molecule-protein binding, surpassing AutoDock Vina. Concerns about data redundancy, generalization, and molecular interaction understanding prompt scrutiny for drug discovery reliability.
The Illustrated AlphaFold
The article discusses AlphaFold3's architecture for predicting protein structures, including Input Preparation, Representation Learning, and Structure Prediction. It highlights improvements like predicting complexed proteins and enriching representations with MSA and templates.
Rapid protein evolution by few-shot learning with a protein language model
Researchers introduce EVOLVEpro, a method merging PLMs and predictors to speed up protein evolution. It outperforms existing techniques, achieving significant property enhancements in just four rounds. The approach benefits various applications, showcasing AI's potential in protein engineering.
Accurate structure prediction of biomolecular interactions with AlphaFold 3
AlphaFold 3 introduces a diffusion-based architecture that enhances biomolecular interaction predictions, outperforming existing tools and improving data efficiency, with potential implications for drug discovery and structural biology.
Creating the largest protein-protein interaction dataset in the world
A-Alpha Bio, founded in 2017, is creating a large protein-protein interaction dataset using its AlphaSeq method, which employs genetically engineered yeast cells to measure interactions, enhancing drug development and biotechnology.