Prediction of protein structure and AI

Shiho Ohno,Noriyoshi Manabe,Yoshiki Yamaguchi

DOI: https://doi.org/10.1038/s10038-023-01215-4

2024-01-04

Journal of Human Genetics

Abstract:AlphaFold, an artificial intelligence (AI)-based tool for predicting the 3D structure of proteins, is now widely recognized for its high accuracy and versatility in the folding of human proteins. AlphaFold is useful for understanding structure-function relationships from protein 3D structure models and can serve as a template or a reference for experimental structural analysis including X-ray crystallography, NMR and cryo-EM analysis. Its use is expanding among researchers, not only in structural biology but also in other research fields. Researchers are currently exploring the full potential of AlphaFold-generated protein models. Predicting disease severity caused by missense mutations is one such application. This article provides an overview of the 3D structural modeling of AlphaFold based on deep learning techniques and highlights the challenges in predicting the pathogenicity of missense mutations.

genetics & heredity

What problem does this paper attempt to address?

The paper primarily explores the application of AlphaFold in protein structure prediction and its potential in assessing the impact of missense mutations. Specifically, the paper focuses on the following aspects:

1. **Application of AlphaFold**:
- AlphaFold is an AI-based tool used to predict the three-dimensional structure of proteins. Its high accuracy and versatility in human protein folding have been widely recognized.
- Protein models generated by AlphaFold help in understanding the relationship between protein structure and function and can serve as templates or references for experimental structural analyses such as X-ray crystallography, NMR, and cryo-EM.

2. **Prediction of the Impact of Missense Mutations**:
- AlphaFold can be used to predict the severity of diseases caused by missense mutations. Although most protein structures in the PDB are of the wild type rather than pathogenic mutants, AlphaFold still has potential in assessing the impact of missense mutations.
- AlphaMissense is a new tool developed by DeepMind that uses AlphaFold's structural database to evaluate the severity of mutation patterns of all amino acids in all human proteins (AlphaMissense score, AM score). This scoring system can help researchers predict the potential pathogenicity or benignity of variants.

3. **Impact on Disordered Regions**:
- AlphaFold performs well in predicting folded regions but finds it challenging to predict static conformations of loop regions and Intrinsically Disordered Regions (IDRs).
- Many mutations in disordered regions are usually benign, but a certain proportion of pathogenic mutations are located in these regions. Studying the impact of mutations in these regions can help improve the accuracy of mutation severity predictions.

4. **Future Prospects**:
- Although AI-based structure prediction and missense mutation severity prediction have reached a practical level, the specific mechanisms of the AlphaMissense score are not yet fully understood.
- Future research needs to further elucidate the mechanisms of protein folding, determine the impact of mutations on proteins from a physicochemical perspective, and classify the mechanisms of pathogenic variants in disordered regions.

In summary, this review paper provides a detailed overview of the advancements in the field of protein structure prediction by AlphaFold and focuses on its application and challenges in assessing the impact of missense mutations.