1
AlphaGo
Shares tags: ai
AI Tool
AlphaFold 2 Review
AlphaFold 2 is an AI system developed by DeepMind that significantly advanced protein structure prediction accuracy, turning molecular biology into an optimization playground for AI.
1AlphaFold 2 achieved a median score of 92.4 out of 100 on the global distance test (GDT) in the CASP14 competition.
2The AlphaFold Protein Structure Database provides open access to over 200 million protein structure predictions.
3AlphaFold 3, released in May 2024, expanded prediction capabilities to include complexes with DNA, RNA, and various ligands.
4The AlphaFold Protein Structure Database has over three million users from more than 190 countries.
⚡
AlphaFold 2 at a Glance
Connect
</>Embed "Featured on Stork" Badge▼
<a href="https://www.stork.ai/en/alphafold-2" target="_blank" rel="noopener noreferrer"><img src="https://www.stork.ai/api/badge/alphafold-2?style=dark" alt="AlphaFold 2 - Featured on Stork.ai" height="36" /></a>
[](https://www.stork.ai/en/alphafold-2)
overview
What is AlphaFold 2?
AlphaFold 2 is a deep learning AI system developed by DeepMind that enables scientists and researchers to predict the three-dimensional structure of proteins from their amino acid sequences. It leverages neural networks to achieve atomic-level accuracy in its predictions. This capability is crucial because a protein's structure dictates its function, impacting fields from drug discovery to understanding disease mechanisms.
quick facts
Quick Facts
| Attribute | Value |
|---|---|
| Developer | DeepMind |
| Business Model | Freemium |
| Pricing | Free (AlphaFold Protein Structure Database, AlphaFold 2 open-source code); Free for non-commercial use (AlphaFold Server); Commercial licensing via Isomorphic Labs (AlphaFold 3) |
| Platforms | Web (AlphaFold Protein Structure Database, AlphaFold Server), Codebase (Python, TensorFlow) |
| API Available | Yes (via open-source code for AlphaFold 2; AlphaFold Server provides an interface for predictions) |
| Founded | AlphaFold 2 released 2020 (DeepMind founded 2010) |
features
Key Features of AlphaFold 2
AlphaFold 2 provides a suite of capabilities for biomolecular structure prediction, leveraging advanced AI techniques to deliver high-accuracy results for a wide range of biological molecules.
- 1Predicts 3D protein structures from linear amino acid sequences with atomic-level accuracy.
- 2Facilitates understanding of protein-protein interactions through its AlphaFold-Multimer extension.
- 3Identifies Intrinsically Disordered Regions (IDRs) using its local confidence metric (pLDDT).
- 4Accelerates drug discovery processes by providing accurate protein structures for target identification and rational drug design.
- 5Supports structural biology research by offering starting points for experimental structure determination and aiding in the interpretation of low-resolution data.
- 6Contributes to understanding disease mechanisms by analyzing pathogenic mutations in the context of protein structure.
- 7Offers open access to over 200 million predicted protein structures via the AlphaFold Protein Structure Database.
- 8AlphaFold 3 extends capabilities to predict structures of complexes involving proteins, DNA, RNA, and various ligands.
use cases
Who Should Use AlphaFold 2?
AlphaFold 2 is primarily utilized by the scientific and research community for its capabilities in molecular biology and drug development.
- 1Scientists: For fundamental research in biological sciences, understanding protein function, and elucidating disease mechanisms.
- 2Researchers in Drug Discovery: To identify new drug targets, enhance virtual screening, and aid in rational drug design for novel therapeutic compounds.
- 3Structural Biologists: To facilitate experimental structure determination (e.g., X-ray crystallography, cryo-EM, NMR spectroscopy) and interpret low-resolution structural data.
- 4Biomedical Researchers: To analyze pathogenic mutations, understand protein folding errors linked to diseases such as Alzheimer's and Parkinson's, and develop new treatments and vaccines.
pricing
AlphaFold 2 Pricing & Plans
AlphaFold 2 operates on a freemium model, with significant resources available at no cost for research and non-commercial use. The original AlphaFold 2 source code was released under an Apache 2.0 license, making it freely available for academic and commercial use. The AlphaFold Protein Structure Database, developed in collaboration with EMBL-EBI, provides open access to over 200 million predicted protein structures. The AlphaFold Server, powered by AlphaFold 3, offers customized biomolecular structure prediction for non-commercial research. AlphaFold 3's source code was released with a restricted license in November 2024, making it less amenable to general tweaking and integration into protein design pipelines compared to AlphaFold 2. Commercial applications and partnerships utilizing AlphaFold 3 are managed through Isomorphic Labs, a DeepMind subsidiary.
- 1AlphaFold Protein Structure Database: Free (open access)
- 2AlphaFold 2 Source Code: Free (open-source, Apache 2.0 license)
- 3AlphaFold Server: Free (for non-commercial research use)
- 4AlphaFold 3 Source Code: Restricted license (less amenable to general tweaking)
competitors
AlphaFold 2 vs Competitors
The field of protein structure prediction has seen significant advancements, with several tools offering comparable or specialized capabilities alongside AlphaFold 2.
1
RoseTTAFold↗
It's a deep learning network that achieved similar accuracy to AlphaFold 2 and has evolved into RoseTTAFold All-Atom, capable of modeling more complex biological assemblies including proteins, nucleic acids, small molecules, and metals.
RoseTTAFold demonstrated comparable accuracy to AlphaFold 2 in protein structure prediction, particularly in the CASP14 competition. Its 'All-Atom' version extends its capabilities beyond just proteins to a wider range of biomolecules, whereas AlphaFold 2 primarily focused on single protein chains and multimers, though AlphaFold 3 has expanded to predict interactions with DNA, RNA, and ligands.
2
ESMFold↗
It predicts protein structures from a single amino acid sequence using a protein language model, offering significantly faster prediction speeds compared to methods relying on multiple sequence alignments.
ESMFold offers comparable accuracy to AlphaFold 2 but with a major advantage in speed, as it doesn't require computationally expensive multiple sequence alignments (MSAs). AlphaFold 2 relies heavily on MSAs for its predictions. ESMFold is also available for free.
3
OpenFold↗
It is a fast, memory-efficient, and fully trainable open-source implementation of AlphaFold 2, designed to match its accuracy.
OpenFold aims to replicate and provide an open-source alternative to AlphaFold 2's capabilities and accuracy. While AlphaFold 2's code was open-sourced, OpenFold was built from the ground up to be a robust and generalizable implementation, addressing some of the challenges of using the original AlphaFold 2 code.
4
NVIDIA BioNeMo↗
It's a comprehensive cloud service and framework for drug discovery that integrates various AI models, including protein language models and OpenFold, for biomolecular prediction and generation.
BioNeMo is a broader platform for drug discovery that includes protein structure prediction capabilities (e.g., via OpenFold and other large language models) rather than being solely a protein structure predictor like AlphaFold 2. It offers a framework for training and deploying large-scale biomolecular models, providing a more extensive toolkit for researchers.
❓
Frequently Asked Questions
+What is AlphaFold 2?
AlphaFold 2 is a deep learning AI system developed by DeepMind that enables scientists and researchers to predict the three-dimensional structure of proteins from their amino acid sequences. It leverages neural networks to achieve atomic-level accuracy in its predictions. This capability is crucial because a protein's structure dictates its function, impacting fields from drug discovery to understanding disease mechanisms.
+Is AlphaFold 2 free?
Yes, AlphaFold 2 operates on a freemium model. The AlphaFold Protein Structure Database is freely accessible, and the original AlphaFold 2 source code is open-source under an Apache 2.0 license. The AlphaFold Server, powered by AlphaFold 3, is also free for non-commercial research use. Commercial applications of AlphaFold 3 are managed through DeepMind's subsidiary, Isomorphic Labs.
+What are the main features of AlphaFold 2?
AlphaFold 2's main features include predicting 3D protein structures with atomic-level accuracy, understanding protein-protein interactions, identifying intrinsically disordered regions, accelerating drug discovery, supporting structural biology research, and contributing to understanding disease mechanisms. Its AlphaFold Protein Structure Database provides open access to over 200 million predicted structures, and AlphaFold 3 extends predictions to complexes with DNA, RNA, and ligands.
+Who should use AlphaFold 2?
AlphaFold 2 is primarily intended for scientists and researchers in fields such as biological sciences, drug discovery, structural biology, and biomedical research. It assists in fundamental research, identifying drug targets, facilitating experimental structure determination, and analyzing disease-related protein mutations.
+How does AlphaFold 2 compare to alternatives?
AlphaFold 2 is a leading protein structure prediction tool. Compared to RoseTTAFold, AlphaFold 2 focuses on high-accuracy protein prediction, while RoseTTAFold All-Atom extends to a broader range of biomolecules. ESMFold offers comparable accuracy but with significantly faster prediction speeds by avoiding multiple sequence alignments. OpenFold is an open-source implementation designed to replicate AlphaFold 2's accuracy. NVIDIA BioNeMo is a broader drug discovery platform that integrates various AI models, including structure prediction capabilities.