AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins
AlphaFold 2 (AF2) has revolutionised protein structure prediction but, like any new tool, its performance on specific classes of targets, especially those potentially under-represented in its training data, merits attention. Prompted by a highly confident prediction for a biologically meaningless, r...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | Computational and Structural Biotechnology Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2001037025000200 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832585130625990656 |
|---|---|
| author | Olivia S. Pratt Luc G. Elliott Margaux Haon Shahram Mesdaghi Rebecca M. Price Adam J. Simpkin Daniel J. Rigden |
| author_facet | Olivia S. Pratt Luc G. Elliott Margaux Haon Shahram Mesdaghi Rebecca M. Price Adam J. Simpkin Daniel J. Rigden |
| author_sort | Olivia S. Pratt |
| collection | DOAJ |
| description | AlphaFold 2 (AF2) has revolutionised protein structure prediction but, like any new tool, its performance on specific classes of targets, especially those potentially under-represented in its training data, merits attention. Prompted by a highly confident prediction for a biologically meaningless, randomly permuted repeat sequence, we assessed AF2 performance on sequences composed of perfect repeats of random sequences of different lengths. AF2 frequently folds such sequences into β-solenoids which, while ascribed high confidence, contain unusual and implausible features such as internally stacked and uncompensated charged residues. A number of sequences confidently predicted as β-solenoids are predicted by other advanced methods as intrinsically disordered. The instability of some predictions is demonstrated by molecular dynamics. Importantly, other deep learning-based structure prediction tools predict different structures or β-solenoids with much lower confidence suggesting that AF2 alone has an unreasonable tendency to predict confident but unrealistic β-solenoids for perfect repeat sequences. The potential implications for structure prediction of natural (near-)perfect sequence repeat proteins are also explored. |
| format | Article |
| id | doaj-art-63d5de881a024e22aeea0a1005d94bc0 |
| institution | Kabale University |
| issn | 2001-0370 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Computational and Structural Biotechnology Journal |
| spelling | doaj-art-63d5de881a024e22aeea0a1005d94bc02025-01-27T04:21:49ZengElsevierComputational and Structural Biotechnology Journal2001-03702025-01-0127467477AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteinsOlivia S. Pratt0Luc G. Elliott1Margaux Haon2Shahram Mesdaghi3Rebecca M. Price4Adam J. Simpkin5Daniel J. Rigden6Department of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United Kingdom; Department of Chemistry, University of Liverpool, Crown Street, Liverpool L69 7ZD, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United Kingdom; Computational Biology Facility, MerseyBio,University of Liverpool, Crown Street, Liverpool L69 7ZB, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United KingdomDepartment of Biochemistry, Cell and Systems, Biology, Institute of Structural, Molecular and Integrative Biology, University of Liverpool, Crown Street, Liverpool L69 7ZB, United Kingdom; Corresponding author.AlphaFold 2 (AF2) has revolutionised protein structure prediction but, like any new tool, its performance on specific classes of targets, especially those potentially under-represented in its training data, merits attention. Prompted by a highly confident prediction for a biologically meaningless, randomly permuted repeat sequence, we assessed AF2 performance on sequences composed of perfect repeats of random sequences of different lengths. AF2 frequently folds such sequences into β-solenoids which, while ascribed high confidence, contain unusual and implausible features such as internally stacked and uncompensated charged residues. A number of sequences confidently predicted as β-solenoids are predicted by other advanced methods as intrinsically disordered. The instability of some predictions is demonstrated by molecular dynamics. Importantly, other deep learning-based structure prediction tools predict different structures or β-solenoids with much lower confidence suggesting that AF2 alone has an unreasonable tendency to predict confident but unrealistic β-solenoids for perfect repeat sequences. The potential implications for structure prediction of natural (near-)perfect sequence repeat proteins are also explored.http://www.sciencedirect.com/science/article/pii/S2001037025000200AlphafoldStructure predictionBeta-solenoidModel confidenceRepeat proteins |
| spellingShingle | Olivia S. Pratt Luc G. Elliott Margaux Haon Shahram Mesdaghi Rebecca M. Price Adam J. Simpkin Daniel J. Rigden AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins Computational and Structural Biotechnology Journal Alphafold Structure prediction Beta-solenoid Model confidence Repeat proteins |
| title | AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins |
| title_full | AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins |
| title_fullStr | AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins |
| title_full_unstemmed | AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins |
| title_short | AlphaFold 2, but not AlphaFold 3, predicts confident but unrealistic β-solenoid structures for repeat proteins |
| title_sort | alphafold 2 but not alphafold 3 predicts confident but unrealistic β solenoid structures for repeat proteins |
| topic | Alphafold Structure prediction Beta-solenoid Model confidence Repeat proteins |
| url | http://www.sciencedirect.com/science/article/pii/S2001037025000200 |
| work_keys_str_mv | AT oliviaspratt alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT lucgelliott alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT margauxhaon alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT shahrammesdaghi alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT rebeccamprice alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT adamjsimpkin alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins AT danieljrigden alphafold2butnotalphafold3predictsconfidentbutunrealisticbsolenoidstructuresforrepeatproteins |