Protein folding
Also written: Protein structure prediction, Protein-folding problem, Protein structures
How a protein chain folds into its specific 3D shape, which determines its function.
Protein folding is the process where a long chain of amino acids, called a polypeptide, twists and bends to form a unique three-dimensional structure. This specific shape is crucial because it determines what the protein does in the body, like carrying oxygen or fighting infections.
Think of it like a long piece of string that needs to be tied in a very specific knot to become a functional tool. AI models like Google's AlphaFold can now predict this final 3D shape from the amino acid sequence, which used to take a lot of lab work.
Protein folding is the physical process by which a polypeptide chain acquires its native three-dimensional structure. This process is driven by thermodynamic principles, primarily hydrophobic interactions, hydrogen bonding, and van der Waals forces, leading to a stable, functional conformation. The accuracy of predicting this structure from sequence alone has been a grand challenge in biology, with AI models like AlphaFold achieving near-experimental accuracy.
Protein folding is a key area where AI has shown significant scientific impact, as highlighted by Google DeepMind's AlphaFold. Recent discussions focus on the limitations of current AI approaches and the need for new methods to fully understand biological processes, moving beyond just predicting structure to understanding function and discovery.
- Why AlphaFold Didn't Solve Protein Folding — Pushmeet Kohli, Google DeepMind & Sal Candido, BiohubGoogle · Latent Space ·
- From games to biology and beyond: 10 years of AlphaGo’s impactGoogle · Google DeepMind Blog ·
- Strengthening our partnership with the UK government to support prosperity and security in the AI eraGoogle · Google DeepMind Blog ·
- AlphaFold: Grand Challenge to Nobel Prize with John JumperGoogle · Google DeepMind: The Podcast ·
- AlphaFold: Five years of impactGoogle · Google DeepMind Blog ·
- Google Gemini
- Google Gemini has been a leader in protein folding with its AlphaFold system, which has revolutionized protein structure prediction and accelerated biological research.
- Protein synthesis
- This is the process of creating the initial amino acid chain,
- Protein denaturation
- This is the unfolding of a protein from its native structu
The study of protein folding has been a central topic in biochemistry and molecular biology since the mid-20th century. The term itself refers to the physical process observed in biological systems.
You can explore protein structures predicted by AI. Visit the AlphaFold Protein Structure Database online and search for a protein you're interested in, like hemoglobin, to see its 3D model. This allows you to visualize the complex shapes AI can now predict.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- What makes a protein fold correctly?
- A protein folds based on the sequence of its amino acids and the physical and chemical forces in its environment, like water. The goal is to reach the most stable, lowest energy 3D shape.
- Can AI make mistakes in predicting protein folding?
- While AI models like AlphaFold are highly accurate, they are not perfect. Predictions can sometimes have lower confidence in certain regions of the protein structure.
First seen on Subvolts Jan 25, 2022. Updated Oct 17, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.