AlphaFold

Introduction

Proteins are essential molecules that carry out nearly every function in the human body, such as building tissues to fight infections. Protein folding is the process by which a protein chain twists and bends into a specific 3D shape, which can determine how the protein functions in the body. Amino acids have a diverse range of chemical properties that influences protein folding, such as being hydrophobic or hydrophilic, as well as varying in their electric charge. For decades, scientists have struggled to find an accurate way to predict protein folding, and AlphaFold has solved this issue. 

AlphaFold

AlphaFold is an artificial intelligence system developed by DeepMind that has the ability to predict a protein’s 3D structure from its amino acid sequence with high accuracy. The AI model was developed from the Protein Data Bank, which includes all 215,000 macromolecular structures that are publicly available. After being introduced to a new protein, it compares the amino acid sequence with similar proteins and predicts how these molecules will interact. A digital 3D model is then generated from this information. This technology is also available for the general public to use, allowing anyone to conduct independent research on proteins. 

Drawbacks

Since AlphaFold is trained on proteins with known sequences and structures, it may struggle to accurately predict the structures of “orphan” proteins that have few related sequences in existing databases. In addition, the model is unable to predict the structural changes that proteins undergo when performing their functions. Like other AI models, AlphaFold is prone to hallucinations, which leads it to generate incorrect protein structures. 

Conclusion

Protein folding is driven by interactions within the amino acid chain, resulting in a unique 3D structure that determines each protein’s specific function. This process is highly complex, and for decades, scientists struggled to predict protein structures accurately. That changed with the development of AlphaFold, an AI system that uses machine learning to model a protein’s 3D shape from its amino acid sequence. While AlphaFold represents a major breakthrough and outperforms previous methods, several limitations still prevent it from achieving complete accuracy. 

References:

Database, AlphaFold Protein Structure. “Alphafold Protein Structure Database.” AlphaFold Protein Structure Database, alphafold.ebi.ac.uk/. Accessed 17 Oct. 2025.

Embl-Ebi. “An Introductory Guide to Alphafold’s Strengths and Limitations.” An Introductory Guide to AlphaFold’s Strengths and Limitations | AlphaFold, www.ebi.ac.uk/training/online/courses/alphafold/an-introductory-guide-to-its-strengths-and-limitations/. Accessed 17 Oct. 2025.

Embl-Ebi. “What Is Alphafold?” What Is AlphaFold? | AlphaFold, www.ebi.ac.uk/training/online/courses/alphafold/an-introductory-guide-to-its-strengths-and-limitations/what-is-alphafold/. Accessed 17 Oct. 2025.

Jumper, John, et al. “Highly Accurate Protein Structure Prediction with Alphafold.” Nature News, Nature Publishing Group, 15 July 2021, www.nature.com/articles/s41586-021-03819-2.

Quiz: