Using AlphaFold to predict the impact of single mutations on protein stability and function
Open accessPublished version of a preprint
The DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index; 3 of the 9 sources its claims cite were checked for support.
Abstract
AlphaFold changed the field of structural biology by achieving three-dimensional (3D) structure prediction from protein sequence at experimental quality. The astounding success even led to claims that the protein folding problem is “solved”. However, protein folding problem is more than just structure prediction from sequence. Presently, it is unknown if the AlphaFold-triggered revolution could help to solve other problems related to protein folding. Here we assay the ability of AlphaFold to predict the impact of single mutations on protein stability (ΔΔG) and function. To study the question we extracted the pLDDT and metrics from AlphaFold predictions before and after single mutation in a protein and correlated the predicted change with the experimentally known ΔΔG values. Additionally, we correlated the same AlphaFold pLDDT metrics with the impact of a single mutation on structure using a large scale dataset of single mutations in GFP with the experimentally assayed levels of fluorescence. We found a very weak or no correlation between AlphaFold output metrics and change of protein stability or fluorescence. Our results imply that AlphaFold may not be immediately applied to other problems or applications in protein folding.
Overview
CoChat Summary
TL;DR
The paper tests whether AlphaFold confidence metrics can predict the effect of single mutations on protein stability (ΔΔG) and GFP fluorescence, using experimental mutation datasets. It finds weak or no useful correlation, concluding AlphaFold outputs are not directly reliable for these prediction tasks.
Key findings
- For 1154 mutations in 73 proteins, the correlation between mutation-induced change in local confidence and experimental stability change was weak: PCC = -0.17 ± 0.03 for ΔpLDDT, while Δ<pLDDT> was 0.02 ± 0.03.
- A multiple linear regression combining ΔpLDDT and Δ<pLDDT> did not improve prediction much: 0.12 ± 0.05 on the training set and 0.20 ± 0.04 on the test set.
- For 796 GFP single mutants, AlphaFold metrics also showed only weak association with function: PCC = 0.17 ± 0.03 for ΔpLDDT and 0.16 ± 0.04 for Δ<pLDDT>.
- Among destabilizing mutations, 87% had negative ΔpLDDT, but this still did not produce a meaningful correlation with ΔΔG.
Methods
- Experimental stability data came from ThermoMutDB v1.3; after filtering for single-point mutations measured at pH 3–9, 293–300 K, and proteins <250 aa, the authors analyzed 1154 mutations in 73 proteins.
- GFP function data came from a published dataset; the authors randomly selected 796 single mutants and used fluorescence as the functional readout.
- Wild-type and mutant structures were obtained from the AlphaFold Protein Structure Database or predicted with standalone AlphaFold, and analysis focused on ΔpLDDT and Δ<pLDDT>; the stability dataset was split into training/test sets by a 50% sequence identity threshold and analyzed with multiple linear regression.
Limitations
- The paper states that AlphaFold was not designed or validated to predict mutation effects on stability or function, so direct use is not expected to work well.
- The authors note that the available ΔΔG data are limited in size (they describe experimentally measured ΔΔG values as being on the order of 10,000), which they argue constrains deep-learning prediction; this is an interpretation made in the paper.
- The evidence is limited to the datasets studied here: single-point mutations in ThermoMutDB proteins and GFP mutants, so broader generalization is not stated in the paper.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures


Verification
Claims the paper makes
3 Valid · 0 Not valid · 3 Not checked · 19 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Thus, using AlphaFold models instead of PDB structures does not make ΔΔG predictions more accurate [15]”
§ Using AlphaFold to predict the impact of single mutations on protein stability and function > Discussion
What the sources say
- Partly supportedcites [15] Pak MA, Ivankov DN. Best templates outperform homology models in predicting the impact of mutations on protein stability. Bioinformatics 2022 38(18): 4312– 4320. doi: 10.1093/bioinformatics/btac515 35894930 sourceExperimental — accuracy not yet measured
Section 3.6 (Protein stability change prediction with AlphaFold models) states that 'utilization of predicted protein structure does not improve the accuracy of protein stability change prediction by FoldX', and the Discussion ('Surprisingly, the conclusion holds true even for the models made by AlphaFold') supports the direction of the claim. However, the support is bounded in ways the citing sentence drops: it applies only to FoldX as the ΔΔG predictor, and rests on just two CASP14 AlphaFold models with 1200 random mutations, where correlation coefficients were statistically indistinguishable but prediction error was actually greater for the AlphaFold models (Figure 5); the authors explicitly caution that 'this conclusion is compromised by the limited amount of structures available for evaluation'. So the generalized, unhedged claim 'using AlphaFold models instead of PDB structures does not make ΔΔG predictions more accurate' overstates the scope and certainty of what this source shows. (LLM, conf=0.75)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The correlation coefficient is -0.17 ± 0.03 (p-value = 10-8) for ΔpLDDT and 0.02 ± 0.03 (p-value = 0.44) for the Δ<pLDDT>.”
§ Using AlphaFold to predict the impact of single mutations on protein stability and function > Results > Correlation between ΔΔG and ΔpLDDT values
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Expectedly, the destabilizing effect of mutations was associated with decreasing pLDDTs: 87% of destabilizing mutations had negative ΔpLDDT (p-value = 10−22).”
§ Using AlphaFold to predict the impact of single mutations on protein stability and function > Results > Relationship of ΔpLDDT and amino acid properties
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, there was no correlation between ΔpLDDT and ΔΔG for that 87% of mutations with negative ΔpLDDT.”
§ Using AlphaFold to predict the impact of single mutations on protein stability and function > Results > Relationship of ΔpLDDT and amino acid properties
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We found a very weak or no correlation between AlphaFold output metrics and change of protein stability or fluorescence.”
- Also stated in: § Correlation between ΔΔG and ΔpLDDT values · plain
“We did not observe a pronounced correlation between the mutation effect and the difference in confidence metrics (Fig 1).”
- Also stated in: § Discussion · plain
“We found a weak correlation of -0.17 ± 0.03 between ΔpLDDT and ΔΔG associated with specific mutations.”
§ Using AlphaFold to predict the impact of single mutations on protein stability and function > Abstract
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Correlation between ΔΔG and ΔpLDDT values · plain
14 more claims
Create a free account to see all 25 claims and to verify the 1 reference still open.
- Probably not a substantial claim. 2
- It may state two separate claims as one. 4
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“studies have reported an inability of AF2 to predict the effects of mutations on protein structure and stability14,49,50”
From The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins (2024) · cites this paper as [50]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Beyond predicting the predominant state of naturally occurring proteins, we wanted to test the ability of AF-Cluster to predict effects of point mutations, a task that AF2 in its default settings has not achieved40.”
From Predicting multiple conformations via sequence clustering and AlphaFold2 (2024) · cites this paper as [40]
Citations
Cites
Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing
Benjamini, Y., & Hochberg, Y. (1995) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1995.tb02031.x · Cited by 109,482 · cited 1×
“All correlations were adjusted for multiple hypotheses testing by Benjamini-Hochberg correction [22].”
MethodBasic local alignment search tool
Altschul, S. F., Gish, W., Miller, W., Myers, E. W., & Lipman, D. J. (1990) · Journal of Molecular Biology · doi:10.1006/jmbi.1990.9999 · Cited by 98,669 · cited 4×
“To identify the sequence identities between the proteins in the dataset of mutations we performed protein BLAST [18] search of protein sequences against themselves.”
MethodInfluentialHighly accurate protein structure prediction with AlphaFold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03819-2 · Cited by 43,764 · cited 3×
“AlphaFold is widely claimed to have revolutized protein 3D structure prediction from protein sequence, a 50-years long-standing challenge of protein physics and structural bioinformatics [1].”
BackgroundMethodInfluentialBerman, H. M. (2000) · Nucleic Acids Research · doi:10.1093/nar/28.1.235 · Cited by 40,196
Highly accurate protein structure prediction for the human proteome
Tunyasuvunakool, K., Adler, J., Wu, Z., Green, T., Zielinski, M., Žídek, A., Bridgland, A., Cowie, A., Meyer, C., Laydon, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03828-1 · Cited by 3,025 · cited 2×
“Next, AlphaFold in collaboration with EMBL-EBI constructed the structure models for the whole protein sequence space [8].”
MethodKnowledge‐based protein secondary structure assignment
Frishman, D., & Argos, P. (1995) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.340230412 · Cited by 2,523 · cited 1×
“Information on solvent accessibility was taken from Stride [19].”
BackgroundGuerois, R., Nielsen, J. E., & Serrano, L. (2002) · Journal of Molecular Biology · doi:10.1016/s0022-2836(02)00442-4 · Cited by 1,728 · cited 3×
“Similarly, the same AlphaFold metrics had a very weak correlation with the impact of single mutations on protein function, fluorescence, of GFP. Recent results show that the use of AlphaFold models instead of template structures does not improve ΔΔG prediction by FoldX [12] [13].”
ResultInfluentialHarnessing protein folding neural networks for peptide–protein docking
Tsaban, T., Varga, J. K., Avraham, O., Ben-Aharon, Z., Khramushin, A., & Schueler-Furman, O. (2022) · Nature Communications · doi:10.1038/s41467-021-27838-9 · Cited by 1,205
Harnessing protein folding neural networks for peptide-protein docking
(2021) · doi:10.21203/rs.3.rs-781411/v1 · Cited by 1,203
Mariani, V., Biasini, M., Barbato, A., & Schwede, T. (2013) · Bioinformatics · doi:10.1093/bioinformatics/btt473 · Cited by 1,138 · cited 1×
“LDDT ranges from 0 to 100 and is a superposition-free metric indicating to what extent the protein model reproduces the reference structure [23].”
Background
Cited by
Harnessing protein folding neural networks for peptide–protein docking
Tsaban, T., Varga, J. K., Avraham, O., Ben-Aharon, Z., Khramushin, A., & Schueler-Furman, O. (2022) · Nature Communications · doi:10.1038/s41467-021-27838-9 · Cited by 1,205
A structural biology community assessment of AlphaFold2 applications
Akdel, M., Pires, D. E. V., Pardo, E. P., Jänes, J., Zalevsky, A. O., Mészáros, B., Bryant, P., Good, L. L., Laskowski, R. A., Pozzati, G., et al. (2022) · Nature Structural & Molecular Biology · doi:10.1038/s41594-022-00849-w · Cited by 675
“However, some reports have indicated that AF2 does not appear to be well suited to predict the structures of mutated proteins 54,55 .”
BackgroundAlphaFold2 and its applications in the fields of biology and medicine
Yang, Z., Zeng, X., Zhao, Y., & Chen, R. (2023) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-023-01381-z · Cited by 577
“Although there are researches showing that AF2 could predict the phenotypic effect of missense mutations,(209,210) it was also observed that the performance of missense mutations prediction is not good in other studies,(211) and there were only weak or no correlations between the output metrics of AF2 and changes in p…”
BackgroundPredicting multiple conformations via sequence clustering and AlphaFold2
Wayment-Steele, H. K., Ojoawo, A., Otten, R., Apitz, J. M., Pitsawong, W., Hömberger, M., Ovchinnikov, S., Colwell, L., & Kern, D. (2024) · Nature · doi:10.1038/s41586-023-06832-9 · Cited by 545
“Beyond predicting the predominant state of naturally occurring proteins, we wanted to test the ability of AF-Cluster to predict effects of point mutations, a task that AF2 in its default settings has not achieved 40 .”
Sampling alternative conformational states of transporters and receptors with AlphaFold2
del Alamo, D., Sala, D., Mchaourab, H. S., & Meiler, J. (2022) · eLife · doi:10.7554/elife.75751 · Cited by 511
“Several preprints have provided evidence that AF2, despite its accuracy, likely does not learn the energy landscapes underpinning protein folding and function ( Saldaño et al., 2021 ; Pak et al., 2021 ; Akdel et al., 2021 ).”
BackgroundScaffolding protein functional sites using deep learning
Wang, J., Lisanza, S., Juergens, D., Tischer, D., Watson, J. L., Castro, K. M., Ragotte, R., Saragovi, A., Milles, L. F., Baek, M., et al. (2022) · Science · doi:10.1126/science.abn2100 · Cited by 487
“For almost all problems, we obtained designs that are closely recapitulated by AF with overall and motif (functional site) RMSD typically <2 Å and <1 Å respectively, with high model confidence (pLDDT > 80; Table S2); such recapitulation suggests the designed sequences encode the designed structures (although it should…”
BackgroundMega-scale experimental analysis of protein folding stability in biology and design
Tsuboyama, K., Dauparas, J., Chen, J., Laine, E., Mohseni Behbahani, Y., Weinstein, J. J., Mangan, N. M., Ovchinnikov, S., & Rocklin, G. J. (2023) · Nature · doi:10.1038/s41586-023-06328-6 · Cited by 350
Can AlphaFold2 predict the impact of missense mutations on structure?
Buel, G. R., & Walters, K. J. (2022) · Nature Structural & Molecular Biology · doi:10.1038/s41594-021-00714-2 · Cited by 332
Before and after AlphaFold2: An overview of protein structure prediction
Bertoline, L. M. F., Lima, A. N., Krieger, J. E., & Teixeira, S. K. (2023) · Frontiers in Bioinformatics · doi:10.3389/fbinf.2023.1120370 · Cited by 297
Machine learning for functional protein design
Notin, P., Rollins, N., Gal, Y., Sander, C., & Marks, D. (2024) · Nature Biotechnology · doi:10.1038/s41587-024-02127-0 · Cited by 270
AlphaFold2 models indicate that protein sequence determines both structure and dynamics
Guo, H.-B., Perminov, A., Bekele, S., Kedziora, G., Farajollahi, S., Varaljay, V., Hinkle, K., Molinero, V., Meister, K., Hung, C., et al. (2022) · Scientific Reports · doi:10.1038/s41598-022-14382-9 · Cited by 248
One-shot design of functional protein binders with BindCraft
Pacesa, M., Nickel, L., Schellhaas, C., Schmidt, J., Pyatova, E., Kissling, L., Barendse, P., Choudhury, J., Kapoor, S., Alcaraz-Serna, A., et al. (2025) · Nature · doi:10.1038/s41586-025-09429-6 · Cited by 236
Machine Learning-Guided Protein Engineering
Kouba, P., Kohout, P., Haddadi, F., Bushuiev, A., Samusevich, R., Sedlar, J., Damborsky, J., Pluskal, T., Sivic, J., & Mazurenko, S. (2023) · ACS Catalysis · doi:10.1021/acscatal.3c02743 · Cited by 209
“…of closely related protein variants because AlphaFold2 predictions, which rely on evolutionary information obtained from MSAs, appear to be insensitive to single-point mutations 139 and so may be unable to accurately capture the often subtle differences between closely related protein variants.”
Transfer learning to leverage larger datasets for improved prediction of protein stability changes
Dieckhaus, H., Brocidiacono, M., Randolph, N. Z., & Kuhlman, B. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2314853121 · Cited by 205
High-throughput prediction of protein conformational distributions with subsampled AlphaFold2
Monteiro da Silva, G., Cui, J. Y., Dalgarno, D. C., Lisi, G. P., & Rubenstein, B. M. (2024) · Nature Communications · doi:10.1038/s41467-024-46715-9 · Cited by 196
Modeling conformational states of proteins with AlphaFold
Sala, D., Engelberger, F., Mchaourab, H. S., & Meiler, J. (2023) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2023.102645 · Cited by 182
Whole genome sequencing in clinical practice
Bagger, F. O., Borgwardt, L., Jespersen, A. S., Hansen, A. R., Bertelsen, B., Kodama, M., & Nielsen, F. C. (2024) · BMC Medical Genomics · doi:10.1186/s12920-024-01795-w · Cited by 178
“Although, initial attempts were not entirely successful [83, 84], the recent AlphaMissense (AM) algorithm, represents a major leap forward [64].”
BackgroundThe impact of AlphaFold2 one year on
Jones, D. T., & Thornton, J. M. (2022) · Nature Methods · doi:10.1038/s41592-021-01365-3 · Cited by 159
“These two concerns will create a huge dilemma.”
Protein generation with evolutionary diffusion: sequence is all you need
Alamdari, S., Thakkar, N., van den Berg, R., Tenenholtz, N., Strome, R., Moses, A. M., Lu, A. X., Fusi, N., Amini, A. P., & Yang, K. K. (2023) · doi:10.1101/2023.09.11.556673 · Cited by 154
Navigating the landscape of enzyme design: from molecular simulations to machine learning
Zhou, J., & Huang, M. (2024) · Chemical Society Reviews · doi:10.1039/d4cs00196f · Cited by 191
“48 Whereas another research indicated that AlphaFold2 was able to predict the effect of single mutations on local structural deformation for a large range of proteins, using the measure of effective strain (ES).”
The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins
Agarwal, V., & McShan, A. C. (2024) · Nature Chemical Biology · doi:10.1038/s41589-024-01638-w · Cited by 155
“In support of this, studies have reported an inability of AF2 to predict the effects of mutations on protein structure and stability 14,49,50 , which may be due to a training bias on stable structures or an inability to extract signal from small mutations through MSAs.”
ResultAI-Based Protein Structure Prediction in Drug Discovery: Impacts and Challenges
Schauperl, M., & Denny, R. A. (2022) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.2c00026 · Cited by 122
Benchmarking Refined and Unrefined AlphaFold2 Structures for Hit Discovery
(2023) · doi:10.1021/acs.jcim.2c01219 · Cited by 136
BindCraft: one-shot design of functional protein binders
Pacesa, M., Nickel, L., Schellhaas, C., Schmidt, J., Pyatova, E., Kissling, L., Barendse, P., Choudhury, J., Kapoor, S., Alcaraz-Serna, A., et al. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.09.30.615802 · Cited by 109
Hold out the genome: a roadmap to solving the cis-regulatory code
de Boer, C. G., & Taipale, J. (2024) · Nature · doi:10.1038/s41586-023-06661-w · Cited by 107
“However, predicting mutation effects remains a challenge 165,166 .”
DeepSTABp: A Deep Learning Approach for the Prediction of Thermal Protein Stability
Jung, F., Frey, K., Zimmer, D., & Mühlhaus, T. (2023) · International Journal of Molecular Sciences · doi:10.3390/ijms24087444 · Cited by 98
“This is a common issue faced in protein property prediction that needs to be addressed and analyzed in the future [34].”
BackgroundMolecular determinants of condensate composition
Holehouse, A. S., & Alberti, S. (2025) · Molecular Cell · doi:10.1016/j.molcel.2024.12.021 · Cited by 97
RosettaDDGPrediction for high‐throughput mutational scans: From stability to binding
(2022) · doi:10.1002/pro.4527 · Cited by 90
Masked inverse folding with sequence transfer for protein representation learning
(2022) · doi:10.1093/protein/gzad015 · Cited by 110
Ranking Peptide Binders by Affinity with AlphaFold**
Chang, L., & Perez, A. (2023) · Angewandte Chemie International Edition · doi:10.1002/anie.202213362 · Cited by 100
AlphaFold2 Can Predict Single-Mutation Effects
McBride, J. M., Polev, K., Abdirasulov, A., Reinharz, V., Grzybowski, B. A., & Tlusty, T. (2023) · Physical Review Letters · doi:10.1103/physrevlett.131.218401 · Cited by 87
Qiu, X., Li, H., Steeg, G. V., & Godzik, A. (2024) · Biomolecules · doi:10.3390/biom14030339 · Cited by 118
“Together, these tools have effectively resolved a long-standing challenge in structural biology: the generation of atomic-level models for protein structures from sequence information alone [1–3].”
BackgroundMasters, M. R., Mahmoud, A. H., & Lill, M. A. (2025) · Nature Communications · doi:10.1038/s41467-025-63947-5 · Cited by 165
Koh, H. Y., Zheng, Y., Yang, M., Arora, R., Webb, G. I., Pan, S., Li, L., & Church, G. M. (2025) · Nature Reviews Bioengineering · doi:10.1038/s44222-025-00349-8 · Cited by 147
“In a different approach, PhiGNet 95 uses graph neural networks representing residues as nodes connected by co-evolutionary links to improve prediction through graph-based modelling.”
FoldX force field revisited, an improved version
Delgado, J., Reche, R., Cianferoni, D., Orlando, G., van der Kant, R., Rousseau, F., Schymkowitz, J. W. H., & Serrano, L. (2025) · Bioinformatics · doi:10.1093/bioinformatics/btaf064 · Cited by 108
“…protein structure has experienced a revolution with the appearance of sequence-based structure predictive methods like Alphafold (Park et al. 2022), it has shown a very weak or no correlation between its metrics and changes in protein stability or fluorescence on single mutations (Pak et al. 2023).”
InfluentialPrediction of multiple conformational states by combining sequence clustering with AlphaFold2
Wayment-Steele, H. K., Ovchinnikov, S., Colwell, L., & Kern, D. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.10.17.512570 · Cited by 69
Disordered–Ordered Protein Binary Classification by Circular Dichroism Spectroscopy
Micsonai, A., Moussong, É., Murvai, N., Tantos, Á., Tőke, O., Réfrégiers, M., Wien, F., & Kardos, J. (2022) · Frontiers in Molecular Biosciences · doi:10.3389/fmolb.2022.863141 · Cited by 69
Critical Assessment of Methods for Predicting the 3D Structure of Proteins and Protein Complexes
Wodak, S. J., Vajda, S., Lensink, M. F., Kozakov, D., & Bates, P. A. (2023) · Annual Review of Biophysics · doi:10.1146/annurev-biophys-102622-084607 · Cited by 67
Using AlphaFold Predictions in Viral Research
Gutnik, D., Evseev, P. V., Анатольевич, М. К., & Shneider, M. M. (2023) · Current Issues in Molecular Biology · doi:10.3390/cimb45040240 · Cited by 67
“Other have found that it is impossible to obtain a reliable evaluation of the impact of mutation on protein stability with the direct application of AI predictions [81].”
BackgroundKeskin Karakoyun, H., Yüksel, Ş. K., Amanoglu, I., Naserikhojasteh, L., Yeşilyurt, A., Yakıcıer, C., Timuçin, E., & Akyerli, C. B. (2023) · Frontiers in Genetics · doi:10.3389/fgene.2023.1052383 · Cited by 58
“Recently, the power of the AF2 computed structures in predicting stability changes was analyzed by addressing the correlation between the experimental ΔΔG and the change in pLDDT scores (Pak et al., 2021; Buel and Walters, 2022).”
MethodRevolutionizing peptide‐based drug discovery: Advances in the post‐AlphaFold era
Chang, L., Mondal, A., Singh, B., Martínez‐Noa, Y., & Perez, A. (2024) · WIREs Computational Molecular Science · doi:10.1002/wcms.1693 · Cited by 56
“47 In such scenarios, traditional docking tools retain their value.”
BackgroundA Perspective on Protein Structure Prediction Using Quantum Computers
Doğa, H., Raubenolt, B. A., Cumbo, F., Joshi, J., DiFilippo, F. P., Qin, J., Blankenberg, D. J., & Shehab, O. (2024) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.4c00067 · Cited by 81
“The discrepancies can be observed in both the 3D coordinates of the produced structure as well as the predicted thermodynamic stability of the mutated residues relative to their confidence scores [101, 102].”
Challenges and opportunities associated with rare-variant pharmacogenomics
Zhou, Y., Tremmel, R., Schaeffeler, E., Schwab, M. G., & Lauschke, V. M. (2022) · Trends in Pharmacological Sciences · doi:10.1016/j.tips.2022.07.002 · Cited by 57
Overview of AlphaFold2 and breakthroughs in overcoming its limitations
Wang, L., Wen, Z., Liu, S.-W., Zhang, L., Finley, C., Lee, H., & Fan, H. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108620 · Cited by 65
LectinOracle: A Generalizable Deep Learning Model for Lectin–Glycan Binding Prediction
(2021) · doi:10.1002/advs.202103807 · Cited by 50
Unmasking AlphaFold to integrate experiments and predictions in multimeric complexes
Mirabello, C., Wallner, B., Nystedt, B., Azinas, S., & Carroni, M. (2024) · Nature Communications · doi:10.1038/s41467-024-52951-w · Cited by 46
“AlphaFold is not trained to predict the effect of mutations on the folding of a protein, and it cannot predict the impact of single-point mutations on protein stability 21 .”
Background(2025) · doi:10.1186/s13059-025-03497-7 · Cited by 45
“This choice allows a fast analysis of deep mutational scanning of proteins and it is supported by previous studies that showed that AlphaFold2 cannot predict reliably the structure of proteins subjected to single-point mutations [72, 73].”
Artificial intelligence and machine learning applications for cultured meat
Todhunter, M. E., Jubair, S., Verma, R., Saqe, R., Shen, K., & Duffy, B. (2024) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2024.1424012 · Cited by 54
Liu, Y., Yeung, W. S. B., Chiu, P. C. N., & Cao, D. (2022) · Frontiers in Genetics · doi:10.3389/fgene.2022.981005 · Cited by 41
“The development of AlphaFold2 (Tunyasuvunakool et al., 2021), to an extent increases the protein structure coverage; but its capability to predict the impact of single mutation is questionable (Pak et al., 2021; Buel and Walters, 2022).”
BackgroundWang, Q. S., Hasegawa, T., Namkoong, H., Saiki, R., Edahiro, R., Sonehara, K., Tanaka, H., Azekawa, S., Chubachi, S., Takahashi, Y., et al. (2024) · Nature Genetics · doi:10.1038/s41588-024-01896-3 · Cited by 43
“…pQTLs by increasing the study size and the integration of more detailed machine learning-derived variant features, such as effects on folding 58,59 , we envision a future where functional priors specific for pQTLs similar to our previous work on eQTLs 60 could be leveraged to better understand…”
Brown, B. P., Stein, R. A., Meiler, J., & Mchaourab, H. S. (2024) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.3c01081 · Cited by 41
Karlsen, S. T., Rau, M. H., Sánchez, B. J., Jensen, K. L., & Zeidan, A. A. (2023) · FEMS Microbiology Reviews · doi:10.1093/femsre/fuad030 · Cited by 47
Membranome 3.0: Database of single‐pass membrane proteins with AlphaFold models
Lomize, A. L., Schnitzer, K. A., Todd, S. C., Cherepanov, S., Outeiral, C., Deane, C. M., & Pogozheva, I. D. (2022) · Protein Science · doi:10.1002/pro.4318 · Cited by 40
“15 Although AlphaFold 2 has a limited applicability in modeling protein dynamics, multiple conformational states, or effects of mutations, 16,17 the approach is thought to achieve a significant progress in predicting the structure of a single protein chain.”
BackgroundXu, Y., Liu, D., & Gong, H. (2024) · Nature Computational Science · doi:10.1038/s43588-024-00716-2 · Cited by 41
“However, how to utilize AlphaFold2 prediction in downstream protein functional prediction tasks is controversial 44–46 .”
Single-sequence protein structure prediction by integrating protein language models
Jing, X., Wu, F., Luo, X., & Xu, J. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2308788121 · Cited by 57
“Previous works showed that the MSA - based methods do not work well on predicting mutational effects (4, 5, 16).”
Agajanian, S., Alshahrani, M., Bai, F., Tao, P., & Verkhivker, G. M. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.2c01634 · Cited by 41
Britton, B. M., Yovanno, R. A., Costa, S. F., McCausland, J. W., Lau, A. Y., Xiao, J., & Hensel, Z. (2023) · Nature Communications · doi:10.1038/s41467-023-39921-4 · Cited by 34
“While AF2 makes predictions that are largely consistent with available structural data and can accurately predict protein – protein interfaces 25 – 27 , its predictions can fail to distinguish between different states of the same complex 28 or describe the consequences of point mutations 29 .”
The Statistical Trends of Protein Evolution: A Lesson from AlphaFold Database
Tang, Q.-Y., Ren, W., Wang, J., & Kaneko, K. (2022) · Molecular Biology and Evolution · doi:10.1093/molbev/msac197 · Cited by 31
“Despite the limitations noted in previous research (Bagdonas et al. 2021; Pak et al. 2021; Ruff and Pappu 2021), AlphaFold 2 is acknowledged to provide high-accuracy predictions of protein structures, even for sequences with relatively fewer homologous sequences ( Jumper et al. 2021).”
Extremozymes: Challenges and opportunities on the road to novel enzymes production
Salas-Bruggink, D. I. J., Martín, J. S.-S., Leiva, G., & Blamey, J. M. (2024) · Process Biochemistry · doi:10.1016/j.procbio.2024.04.035 · Cited by 35
Beyond sequence: Structure-based machine learning
Durairaj, J., de Ridder, D., & van Dijk, A. D. J. (2023) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2022.12.039 · Cited by 33
Raisinghani, N., Alshahrani, M., Gupta, G., & Verkhivker, G. M. (2024) · Viruses · doi:10.3390/v16091458 · Cited by 33
“When mutations lead to notable distortions in the native structure, the performance of AF2 predictions may markedly deteriorate [66,67] with only weak or no correlations between the output metrics of AF2 and changes in protein stability or functionality [68].”
BackgroundLeveraging protein structural information to improve variant effect prediction
Gerasimavicius, L., Teichmann, S. A., & Marsh, J. A. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.103023 · Cited by 35
“…evaluation, as has been seen from its inability to produce mutant protein structures that would deviate from the underlying wild-type structure [53,54], it contains the building blocks to capture complex structural relationships from sequence, which can be creatively utilized for a downstream VEP…”
Analysing protein complexes in plant science: insights and limitation with AlphaFold 3
Lin, P.-Y., Huang, S.-C., Chen, K.-L., Huang, Y., Liao, C.-Y., Lin, G.-J., Lee, H., & Chen, P. (2025) · Botanical studies · doi:10.1186/s40529-025-00462-2 · Cited by 41
“However, AF3 predicts only static protein structures under neutral, stable conditions, limiting its ability to evaluate environmental impacts on protein dynamics (Pak et al. 2023; Niazi et al. 2024).”
Computational design of conformation-biasing mutations to alter protein functions
Cavanagh, P. E., Xue, A. G., Dai, S. A., Qiang, A., Matsui, T., & Ting, A. Y. (2026) · Science · doi:10.1126/science.adv7953 · Cited by 41
The impact of AlphaFold2 on experimental structure solution
Edich, M., Briggs, D. C., Kippes, O., Gao, Y., & Thorn, A. (2022) · Faraday Discussions · doi:10.1039/d2fd00072e · Cited by 29
Lubin, J. H., Markosian, C., Balamurugan, D., Ma, M. T., Chen, C.-H., Liu, D., Pasqualini, R., Arap, W., Burley, S. K., & Khare, S. D. (2023) · JCI Insight · doi:10.1172/jci.insight.168296 · Cited by 26
AlphaFold and what is next: bridging functional, systems and structural biology
Szczepski, K., & Jaremko, Ł. (2025) · Expert Review of Proteomics · doi:10.1080/14789450.2025.2456046 · Cited by 29
“…found a weak correlation between local pLDDT score variations and ΔΔG values (Pearson correlation coefficient, PCC = −0.17), and no correlation between the global average pLDDT scores and ΔΔG values, indicating that AF may not effectively predict the impact of mutations on protein stability [48].”
InfluentialA unified evolution-driven deep learning framework for virus variation driver prediction
Nie, Z., Liu, X., Chen, J., Wang, Z., Liu, Y., Si, H.-R., Dong, T., Xu, F., Song, G., Wang, Y., et al. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-024-00966-9 · Cited by 30
Guo, H.-B., Varaljay, V. A., Kedziora, G., Taylor, K., Farajollahi, S., Lombardo, N., Harper, E., Hung, C., Gross, M., Perminov, A., et al. (2023) · Scientific Reports · doi:10.1038/s41598-023-30310-x · Cited by 22
“To answer this question, we performed MD simulations on five different systems: (1) T7RdhA complexed with BVQ, two Fe4S4 clusters (Fe4S4-A and Fe4S4-B) and the PFOA substrate; (2) apo-T7RdhA with no ligand; (3) T7RdhA complexed with BVQ; (4) T7RdhA complexed with two Fe4S4 clusters; and (5) T7RdhA complexed 3 Vol.:(01…”
MethodInfluentialComputational modeling and prediction of deletion mutants
Woods, H., Schiano, D. L., Aguirre, J. I., Ledwitch, K. V., McDonald, E. F., Voehler, M., Meiler, J., & Schoeder, C. T. (2023) · Structure · doi:10.1016/j.str.2023.04.005 · Cited by 22
Directed Evolution of Protoglobin Optimizes the Enzyme Electric Field
Chaturvedi, S. S., Vargas, S., Ajmera, P., & Alexandrova, A. N. (2024) · Journal of the American Chemical Society · doi:10.1021/jacs.4c03914 · Cited by 23
Analysis of AlphaFold and molecular dynamics structure predictions of mutations in serpins
Garrido, P., Carmena‐Bargueño, M., de la Morena‐Barrio, M. E., Bravo‐Pérez, C., de la Morena‐Barrio, B., Cifuentes‐Riquelme, R., Lozano, M. L., Pérez‐Sánchez, H., & Corral, J. (2024) · PLoS ONE · doi:10.1371/journal.pone.0304451 · Cited by 26
“…validated to predict mutational effects [18] and flipping an amino acid in the target protein to model a mutation will not work in Alpha-Fold [19, 20], it would be interesting to challenge the model to solve the conformational consequences of mutations affecting serpins, as these mutations cause…”
InfluentialLarge-scale discovery, analysis and design of protein energy landscapes
Ferrari, Á. J. R., Dixit, S. M., Thibeault, J., Garcia, M., Houliston, S., Ludwig, R. W., Notin, P., Phoumyvong, C. M., Martell, C. M., Jung, M. D., et al. (2026) · Nature · doi:10.1038/s41586-026-10465-z · Cited by 22
Emerging frontiers in protein structure prediction following the AlphaFold revolution
Rennie, M. L., & Oliver, M. R. (2025) · Journal of The Royal Society Interface · doi:10.1098/rsif.2024.0886 · Cited by 40
The druggable genome: Twenty years later
Radoux, C. J., Vianello, F., McGreig, J. E., Desai, N., & Bradley, A. R. (2022) · Frontiers in Bioinformatics · doi:10.3389/fbinf.2022.958378 · Cited by 27
“Additionally, large-scale validation experiments will be required (Pak et al., 2021; Jones and Thornton, 2022).”
BackgroundLimitations of Protein Structure Prediction Algorithms in Therapeutic Protein Development
Niazi, S. K., Mariam, Z., & Paracha, R. Z. (2024) · BioMedInformatics · doi:10.3390/biomedinformatics4010007 · Cited by 22
An integrative approach to protein sequence design through multiobjective optimization
Hong, L., & Kortemme, T. (2024) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1011953 · Cited by 28
Protein Structural Phylogenetics
Puente-Lelièvre, C., Malik, A. J., & Douglas, J. (2025) · Genome Biology and Evolution · doi:10.1093/gbe/evaf139 · Cited by 33
ESM ‐scan—A tool to guide amino acid substitutions
Totaro, M. G., Vide, U., Zausinger, R., Winkler, A., & Oberdorfer, G. (2024) · Protein Science · doi:10.1002/pro.5221 · Cited by 19
“…conformations and evaluating the impact of single amino acid substitutions on protein stability and function with varying levels of accuracy (Bryant, Pozzati and Elofsson 2022, McBride et al. 2022, Cheng et al. 2023, Sala, Hildebrand and Meiler 2023, Pak et al. 2023, Wayment-Steele et al. 2023).”
Recent progress and future challenges in structure-based protein-protein interaction prediction
Yuan, R., Zhang, J., Zhou, J., & Cong, Q. (2025) · Molecular Therapy · doi:10.1016/j.ymthe.2025.04.003 · Cited by 31
AI-Driven Chemical Design: Transforming the Sustainability of the Pharmaceutical Industry
Ruiz-Gonzalez, A. (2025) · Future Pharmacology · doi:10.3390/futurepharmacol5020024 · Cited by 26
“While some studies point out that AlphaFold excels at protein structure prediction, the ability to determine stability changes upon inducing mutations might be limited [49].”
Meller, A., Kelly, D. G., Smith, L. G., & Bowman, G. R. (2024) · Protein Science · doi:10.1002/pro.4902 · Cited by 18
Praljak, N., Lian, X., Ranganathan, R., & Ferguson, A. L. (2023) · ACS Synthetic Biology · doi:10.1021/acssynbio.3c00261 · Cited by 18
AlphaFold2 Update and Perspectives
Tourlet, S., Radjasandirane, R., Diharce, J., & de Brevern, A. G. (2023) · BioMedInformatics · doi:10.3390/biomedinformatics3020025 · Cited by 19
“Similarly, AF2 models are used to establish links with physicochemical parameters and experiments [73,74].”
MethodZibat, A., Zhang, X., Dickmanns, A., Stegmann, K. M., Dobbelstein, A. W., Alachram, H., Soliwoda, R., Salinas, G., Groß, U., Görlich, D., et al. (2023) · iScience · doi:10.1016/j.isci.2023.107786 · Cited by 17
Learning the shape of protein microenvironments with a holographic convolutional neural network
Pun, M. N., Ivanov, A., Bellamy, Q., Montague, Z., LaMont, C., Bradley, P. H., Otwinowski, J., & Nourmohammad, A. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2300838121 · Cited by 23
AI-Predicted Protein Deformation Encodes Energy Landscape Perturbation
McBride, J. M., & Tlusty, T. (2024) · Physical Review Letters · doi:10.1103/physrevlett.133.098401 · Cited by 17
Innovation in cancer pharmacotherapy through integrative consideration of germline and tumor genomes
Tremmel, R., Hübschmann, D., Schaeffeler, E., Pirmann, S., Fröhling, S., & Schwab, M. (2024) · Pharmacological Reviews · doi:10.1124/pharmrev.124.001049 · Cited by 20
Michalewicz, K., Barahona, M., & Bravi, B. (2024) · Structure · doi:10.1016/j.str.2024.10.001 · Cited by 20
Advances in Machine Learning Models for Predicting Enzyme Kinetic Parameters
Malli, A., Vasyutyn, D., & Kim, J. R. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c02428 · Cited by 28
Tremmel, R., Pirmann, S., Zhou, Y., & Lauschke, V. M. (2023) · British Journal of Clinical Pharmacology · doi:10.1111/bcp.15915 · Cited by 24
“106 An effect barely covered by in silico prediction is the influence on protein – protein interactions although an estimated 10.5% of all mis-sense variants impact such interactions.”
InfluentialEvaluation of Myocilin Variant Protein Structures Modeled by AlphaFold2
Ng, T. K., Ji, J., Liu, Q., Yao, Y., Wang, W., Cao, Y., Chen, C., Lin, J., Dong, G., Cen, L., et al. (2023) · Biomolecules · doi:10.3390/biom14010014 · Cited by 15
“Moreover, a study on the use of AlphaFold to predict the impact of single mutations on protein stability and function found very weak or no correlation between AlphaFold output metrics and change in protein stability or fluorescence [52].”
BackgroundPredicting the structural impact of human alternative splicing
Song, Y., Zhang, C., Omenn, G. S., O’Meara, M. J., & Welch, J. D. (2025) · Genome biology · doi:10.1186/s13059-025-03744-x · Cited by 22
“For example, AlphaFold-Multimer and AlphaMissense were developed to address limitations in the prediction of protein complexes and the structural impact of point mutations [30–32].”
SARS-CoV-2 receptor-binding domain deep mutational AlphaFold2 structures
Kilim, O., Mentes, A., Pál, B., Csabai, I., & Gellért, Á. (2023) · Scientific Data · doi:10.1038/s41597-023-02035-z · Cited by 15
Do Deep Learning Models for Co-Folding Learn the Physics of Protein-Ligand Interactions?Preprint
Masters, M., Mahmoud, A., & Lill, M. A. (2024) · bioRxiv · doi:10.1101/2024.06.03.597219 · Cited by 11
“…spawning hundreds of subsequent papers including high-throughput structure prediction [3, 4, 5], ensemble sampling [6, 7, 8], molecular docking [9, 10, 11, 12], stability prediction [13, 14, 15, 16], further method development [17, 18, 19, 20], and a host of other applied studies [21, 22, 23, 24].”
BackgroundPrediction of protein structure and AI
Ohno, S., Manabe, N., & Yamaguchi, Y. (2024) · Journal of Human Genetics · doi:10.1038/s10038-023-01215-4 · Cited by 17
“In fact, some reports indicate that AlphaFold can wrongly predict a mutant protein as folded with a wild-type-like conformation although the mutations are known to cause unfolding [15], thereby making it unsuitable for predicting destabilizing energies or functions [16].”
Gracia, B., Montes, P., Gutierrez‐Barrera, A. M., Arun, B., & Karras, G. I. (2024) · Cell Reports · doi:10.1016/j.celrep.2024.113803 · Cited by 18
Click, Compute, Create: A Review of Web‐based Tools for Enzyme Engineering
Tripp, A., Braun, M., Wieser, F., Oberdorfer, G., & Lechner, H. (2024) · ChemBioChem · doi:10.1002/cbic.202400092 · Cited by 17
Enzymatic CO2 reduction catalyzed by natural and artificial Metalloenzymes
Deng, Y., Wang, J., Ghosh, B., & Lu, Y. (2024) · Journal of Inorganic Biochemistry · doi:10.1016/j.jinorgbio.2024.112669 · Cited by 18
Concepts and Methods for Predicting Viral Evolution
Meijers, M., Ruchnewitz, D., Eberhardt, J., Karmakar, M., Łuksza, M., & Lässig, M. (2025) · Methods in molecular biology · doi:10.1007/978-1-0716-4326-6_14 · Cited by 18
Metadata
- Authors
- Pak, M. A., Markhieva, K. A., Novikova, M. S., Petrov, D. S., Vorobyev, I. S., Maksimova, E. S., Kondrashov, F. A., & Ivankov, D. N.
- Affiliations
- Skolkovo Institute of Science and Technology (RU)
- Peoples' Friendship University of Russia (RU)
- University of New Mexico (US)
- Ural Federal University (RU)
- Institute of Science and Technology Austria (AT)
- Okinawa Institute of Science and Technology Graduate University (JP)
- Corresponding author
- Ivankov, Dmitry N.
- Venue
- PLOS One
- Publisher
- Public Library of Science (PLoS)
- Published
- 2023-03-16
- Volume
- 18
- Issue
- 3
- Pages
- e0282689
- ISSN
- 1932-6203 (ISSN-L)
- PMC
- PMC10019719
- OpenAlex
- W4327566741
- Cited by
- 322
- References
- 32
- Citations by year
- 2026: 51 · 2025: 90 · 2024: 77 · 2023: 66
- FWCI
- 25.19
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- PLoS ONE (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Electronic Archive of Ural Federal University (ELAR UrFU) (submittedVersion, cc-by)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Structural Biology › Enzyme Structure and Function
- Keywords
- AlphaFold, single mutation, protein stability, protein folding, pLDDT, ΔΔG, protein function, green fluorescent protein, fluorescence, protein structure prediction, protein folding problem
- MeSH terms
- Amino Acid Sequence, Mutation, Proteins, Protein Folding, Protein Stability
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.