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
Items 101–200 of 320
Ranking Peptide Binders by Affinity with AlphaFold**
Chang, L., & Perez, A. (2023) · Angewandte Chemie · doi:10.1002/ange.202213362 · Cited by 14
Discovery of novel amino acid production traits by evolution of synthetic co-cultures
Zuchowski, R., Schito, S., Neuheuser, F., Menke, P., Berger, D. J., Hollmann, N., Gujar, S., Sundermeyer, L., Mack, C., Wirtz, A., et al. (2023) · Microbial Cell Factories · doi:10.1186/s12934-023-02078-2 · Cited by 20
Ohnuki, J., Jaunet-Lahary, T., Yamashita, A., & Okazaki, K.-I. (2024) · The Journal of Physical Chemistry Letters · doi:10.1021/acs.jpclett.3c03052 · Cited by 14
Ohnuki, J., Jaunet-Lahary, T., Yamashita, A., & Okazaki, K.-I. (2023) · doi:10.1101/2023.10.26.564285 · Cited by 14
Tarek, M., Yahia, A., Elnakib, M., & ElHefnawi, M. (2021) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2021.104986 · Cited by 16
Dabrowski-Tumanski, P., & Stasiak, A. (2023) · Molecules · doi:10.3390/molecules28227462 · Cited by 12
“This effect was studied before in the case of missense mutations [45,46], but it shows that AlphaFold is of limited use in the case of lasso-like proteins [47,48], where the topology can change when introducing or removing single-cysteine residue.”
BackgroundEstimating the Prevalence of GNE Myopathy Using Population Genetic Databases
Derksen, A., Thompson, R., Shaikh, M. Y., Spendiff, S., Perkins, T. J., & Lochmüller, H. (2024) · Human Mutation · doi:10.1155/2024/7377504 · Cited by 13
“On a broader scale, there are several limitations to the AlphaFold-derived system on which AlphaMissense is in part trained, most notably its weak correlation in predicting the impact of single mutations on protein function and stability [31].”
Ali, R., & Zhang, Y. (2024) · Frontiers of Chemical Science and Engineering · doi:10.1007/s11705-024-2500-7 · Cited by 16
Ramadane‐Morchadi, L., Rotenberg, N., Esteban‐Sánchez, A., Fortuño, C., Gómez-Sanz, A., Varga, M. J., Chamberlin, A. C., Richardson, M. E., Michailidou, K., Pérez‐Segura, P., et al. (2025) · The American Journal of Human Genetics · doi:10.1016/j.ajhg.2024.12.011 · Cited by 15
RaptorX-Single: single-sequence protein structure prediction by integrating protein language modelsPreprint
Jing, X., Wu, F., Luo, X., & Xu, J. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.04.24.538081 · Cited by 12
“Predicted structure changes of single mutations Previous works showed that the MSA-based methods do not work well on predicting mutational effects [4,5,33].”
BackgroundThe Origin and Evolution of Sex Peptide and Sex Peptide Receptor Interactions
Peng, J., Svetec, N., Molina, H., & Zhao, L. (2024) · Molecular Biology and Evolution · doi:10.1093/molbev/msae065 · Cited by 9
An agnostic analysis of the human AlphaFold2 proteome using local protein conformations
de Brevern, A. G. (2022) · Biochimie · doi:10.1016/j.biochi.2022.11.009 · Cited by 12
Hou, Y., Lu, X., Xu, Z., Qu, J., & Huang, J. (2023) · RSC Advances · doi:10.1039/d2ra07472a · Cited by 13
“35 – 37 It is known that one single mutation in a sequence may completely alter the native structure, with additional examples like GA98/GB98 21 and APOE3/APOE4.”
Rosetta Energy Analysis of AlphaFold2 models: Point Mutations and Conformational EnsemblesPreprint
Stein, R. A., & Mchaourab, H. S. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.09.05.556364 · Cited by 13
“A recently published report found that the internal AlphaFold2 quality metric, pLDDT, was not an adequate measure for the energetic changes due to point mutations.(9) Conversely, another study found a correlation between pLDDT and pathogenicity in a set of cancer proteins.”
BackgroundLu, X., Cao, M., Ma, M., Wu, Y., Qu, M., Du, F., Ji, R., Duan, M., Dong, L., Liu, K., et al. (2025) · Food Bioengineering · doi:10.1002/fbe2.70029 · Cited by 14
Mesdaghi, S., Price, R. M., Madine, J., & Rigden, D. J. (2023) · Journal of Structural Biology · doi:10.1016/j.jsb.2023.108010 · Cited by 11
Kulczyk, A. (2023) · Scientific Reports · doi:10.1038/s41598-023-45200-5 · Cited by 12
“In contrast, current applications of AF2 to problems related to protein folded are limited, largely due to the lack of training data sets representing misfolded and partly-folded structures 31,32 .”
BackgroundA substrate-interacting region of Parkin directs ubiquitination of the mitochondrial GTPase Miro1
Koszela, J., Rintala-Dempsey, A., Salzano, G., Pimenta, V., Kamarainen, O., Gabrielsen, M., Parui, A. L., Shaw, G. S., & Walden, H. (2025) · Journal of Cell Biology · doi:10.1083/jcb.202408025 · Cited by 13
Enzyme structure correlates with variant effect predictability
(2024) · doi:10.1016/j.csbj.2024.09.007 · Cited by 13
Statistical Thermodynamics of the Protein Ensemble: Mediating Function and Evolution
Hilser, V. J., Wrabl, J. O., Millard, C. E., Schmitz, A., Brantley, S. J., Pearce, M., Rehfus, J., Russo, M. M., & Voortman‐Sheetz, K. (2025) · Annual Review of Biophysics · doi:10.1146/annurev-biophys-061824-104900 · Cited by 13
HMG-boxes, ribosomopathies and neurodegenerative disease
Moss, T., LeDoux, M. S., & Crane‐Robinson, C. (2023) · Frontiers in Genetics · doi:10.3389/fgene.2023.1225832 · Cited by 10
“However, it should be noted that the AlphaFold algorithm generally underestimates the effects of point variants (Pak et al., 2023).”
BackgroundSearching similar local 3D micro-environments in protein structure databases with MicroMiner
Sieg, J., & Rarey, M. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbad357 · Cited by 9
Analysis of conformational stability of interacting residues in protein binding interfaces
Chauhan, V. M., & Pantazes, R. J. (2022) · Protein Engineering Design and Selection · doi:10.1093/protein/gzad016 · Cited by 12
Kurniawan, J., & Ishida, T. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.3c00750 · Cited by 10
The Origin and Evolution of Sex Peptide and Sex Peptide Receptor Interactions
(2023) · doi:10.1101/2023.11.19.567744 · Cited by 9
Kulminskaya, N., Rodriguez Gamez, C. F., Hofer, P., Cerk, I. K., Dubey, N., Viertlmayr, R., Sagmeister, T., Pavkov-Keller, T., Zechner, R., & Oberer, M. (2024) · Journal of Lipid Research · doi:10.1016/j.jlr.2023.100491 · Cited by 10
Gusakova, M., Ivanov, M. V., Kashtanova, D. A., Taraskina, A. N., Erema, V. V., Mikova, V. M., Loshkarev, R. I., Ignatyeva, O. A., Akinshina, A. I., Игоревич, М. С., et al. (2023) · Frontiers in Medicine · doi:10.3389/fmed.2023.1178939 · Cited by 8
“This method is limited to structure prediction from sequences (15) and is not readily applicable to other protein folding problems.”
MethodKwon, H., Du, Z., & Li, Y. (2024) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2024.134601 · Cited by 11
Raisinghani, N., Alshahrani, M., Gupta, G., & Verkhivker, G. M. (2024) · International Journal of Molecular Sciences · doi:10.3390/ijms251810082 · Cited by 10
“When mutations lead to large changes in the native structure, the performance of AF2 predictions may be compromised [36–38].”
BackgroundRustamov, K. R., Razzokov, J., & Baev, A. Y. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c00580 · Cited by 12
Chakraborty, C., Bhattacharya, M., & Lee, S.-S. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2026.1739303 · Cited by 12
“Predicted structures are also useful for evaluating protein stability and detecting destabilizing mutations (Pak et al., 2023).”
Machine learning approaches for predicting protein-ligand binding sites from sequence data
Vural, O., & Jololian, L. (2025) · Frontiers in Bioinformatics · doi:10.3389/fbinf.2025.1520382 · Cited by 12
Substrate recognition by a peptide-aminoacyl-tRNA ligase
Ramos‐Figueroa, J. S., Liang, H., & van der Donk, W. A. (2025) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2423858122 · Cited by 7
PHEXL222P Mutation Increases Phex Expression in a New ENU Mouse Model for XLH Disease
Hakam, C. E., Parenté, A., Baraige, F., Magnol, L., Forestier, L., Meo, F. D., & Blanquet, V. (2022) · Genes · doi:10.3390/genes13081356 · Cited by 9
“However, it must be stressed that the present structure of PHEXL222P should be considered very carefully since recent studies have pointed out that AlphaFold is unable to accurately predict defective protein folding from single-point mutation [54,55].”
BackgroundLarrea, A., Jebari‐Benslaiman, S., García, U. G., Urteaga, A. S. J., Uribe, K. B., Benito‐Vicente, A., & Martín, C. (2023) · Current Atherosclerosis Reports · doi:10.1007/s11883-023-01154-7 · Cited by 13
“If a pathogenic variant destabilizes the molecule, then it can be expected that the algorithm will predict a more unfolded state [110].”
BackgroundConcepts and methods for predicting viral evolutionPreprint
Meijers, M., Ruchnewitz, D., Eberhardt, J., Karmakar, M., Łuksza, M., & Lässig, M. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.03.19.585703 · Cited by 10
Hold out the genome: A roadmap to solving the cis-regulatory code
(2023) · doi:10.1101/2023.04.20.537701 · Cited by 107
Integrative analysis of pathogenic variants in glucose-6-phosphatase based on an AlphaFold2 model
Sinclair, M., Stein, R. A., Sheehan, J. H., Hawes, E. M., O’Brien, R. M., Tajkhorshid, E., & Claxton, D. P. (2024) · PNAS Nexus · doi:10.1093/pnasnexus/pgae036 · Cited by 8
A review on computational models for predicting protein solubility
Pimtawong, T., Ren, J., Lee, J., Lee, H., & Na, D. (2025) · The Journal of Microbiology · doi:10.71150/jm.2408001 · Cited by 15
“However, owing to recent advances in computational protein structure prediction methods, highly accurate protein structures are now easily obtained in silico (Liu et al., 2022; Pak et al., 2023; Ruff & Pappu, 2021).”
Curse and Blessing of Non‐Proteinogenic Parts in Computational Enzyme Engineering
Korbeld, K. T., & Fürst, M. J. L. J. (2023) · ChemBioChem · doi:10.1002/cbic.202300192 · Cited by 8
Apprehensions and emerging solutions in ML-based protein structure prediction
Dahlström, K. M., & Salminen, T. A. (2024) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2024.102819 · Cited by 10
Corre, M.-H., Rey, B., David, S. C., Torii, S., Chiappe, D., & Kohn, T. (2024) · Communications Biology · doi:10.1038/s42003-024-06627-2 · Cited by 7
Regularly updated benchmark sets for statistically correct evaluations of AlphaFold applications
Dobson, L., Tusnády, G. E., & Tompa, P. (2025) · Briefings in Bioinformatics · doi:10.1093/bib/bbaf104 · Cited by 6
“…AF2 structures to predict and/or assess various structural and physicochemical features of proteins in specific areas, such as protein stability [4], conformational heterogeneity [5], protein function [6], structure, [7] or interfaces [8] of protein complexes, domain definitions [9], or…”
Molla, A., & Abebe, G. M. (2025) · Innovation · doi:10.11648/j.innov.20250603.20 · Cited by 6
AI-based AlphaFold2 significantly expands the structural space of the autophagy pathway
Malhotra, N., Khatri, S., Kumar, A., Arun, A., Daripa, P., Fatihi, S., Sureshkumar, V., Jain, N., & Thukral, L. (2023) · Autophagy · doi:10.1080/15548627.2023.2238578 · Cited by 7
Considering the ethics of large machine learning models in the chemical sciences
Spotte-Smith, E. W. C. (2025) · Machine Learning: Science and Technology · doi:10.1088/2632-2153/adec3c · Cited by 6
Carugo, O., & Djinović‐Carugo, K. (2023) · Frontiers in Molecular Biosciences · doi:10.3389/fmolb.2023.1155629 · Cited by 6
“, 2022), though AlphaFold2 does not seem to be appropriate to forecast the structure of mutated proteins (Buel and Walters, 2022) (Pak et al., 2021).”
BackgroundDeveloping similarity matrices for antibody-protein binding interactions
Islam, S., & Pantazes, R. J. (2023) · PLoS ONE · doi:10.1371/journal.pone.0293606 · Cited by 7
“Although there are important fringe cases where these methods do not yet succeed [3] and they do not work for some closely related problems [4, 5], they can correctly predict the structures of the vast majority of proteins.”
BackgroundMoreland, R. T., Zhang, S., Barreira, S. N., Ryan, J. F., & Baxevanis, A. D. (2024) · PROTEOMICS · doi:10.1002/pmic.202300397 · Cited by 4
Learning the shape of protein micro-environments with a holographic convolutional neural networkPreprint
Pun, M. N., Ivanov, A., Bellamy, Q., Montague, Z., LaMont, C., Bradley, P. H., Otwinowski, J., & Nourmohammad, A. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.10.31.514614 · Cited by 9
“Despite AlphaFold’s remarkable success at predicting protein folding, it still struggles to determine the effect of mutations on the stability and function of a protein [13].”
BackgroundAromatic–aromatic interactions drive fold switch of GA95 and GB95 with three residue difference
Chen, C., Zhang, Z., Duan, M., Wu, Q., Yang, M., Jiang, L., Liu, M., & Li, C. (2025) · Chemical Science · doi:10.1039/d4sc04951a · Cited by 5
Islam, N. N., Coban, M., Fuller, J. M., Weber, C., Chitale, R. A., Jussila, B., Brock, T., Tao, C., & Caulfield, T. R. (2025) · Communications Biology · doi:10.1038/s42003-025-08334-y · Cited by 6
“Studies suggest that both AlphaFold and conservation-only models show low accuracies when tested against functional assay-derived datasets 12,16 , suggesting that these approaches do not incorporate detailed structural dynamics, which limits their ability to predict the nuanced impacts of mutations…”
Vidya, F. N. U., Luo, Y., Wu, H., van der Donk, W. A., McShan, A. C., & Agarwal, V. (2025) · Nature Communications · doi:10.1038/s41467-025-64365-3 · Cited by 7
ZetaDesign: an end-to-end deep learning method for protein sequence design and side-chain packing
Yan, J., Li, S., Zhang, Y., Hao, A., & Zhao, Q. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbad257 · Cited by 6
Analysis of proteins in the light of mutations
Vila, J. A. (2024) · European Biophysics Journal · doi:10.1007/s00249-024-01714-y · Cited by 7
“…2013; Pires et al. 2014; Kulshreshtha et al. 2016; Chen et al. 2020; Panduran-gan and Blundell 2020; Caldararu et al. 2021; Rodrigues et al. 2021; Buel and Walters 2022; Pancotti et al. 2022; Pak et al. 2023; Pucci et al. 2022; Diaz et al. 2023; Benev-enuta et al. 2023; Kurniawan and Ishida 2023).”
Ryan‐Phillips, F., Henehan, L., Ramdas, S., Palace, J. A., Beeson, D., & Dong, Y. Y. (2024) · Biomedicines · doi:10.3390/biomedicines12112549 · Cited by 6
“This is intriguing, because RDS variants would be expected to disrupt structural stability, and pLDDT has been thought to serve as a proxy for residue stability [18,29,32].”
InfluentialIlse, T. E., Zhang, H., Heutinck, A., Liu, C., Eicke, S., Sharma, M., Pfister, B., Santelia, D., & Zeeman, S. C. (2025) · PLANT PHYSIOLOGY · doi:10.1093/plphys/kiaf206 · Cited by 6
Tavares, D., & van der Meer, J. R. (2021) · Applied and Environmental Microbiology · doi:10.1128/aem.02117-21 · Cited by 6
“Therefore, although AlphaFold does not explicitly inform about protein stability (39), these predictions thus”
BackgroundRaballah, E., Wilding, K. M., Anyona, S. B., Munde, E. O., Hurwitz, I., Onyango, C., Ayieko, C., Lambert, C. G., Schneider, K. A., Seidenberg, P., et al. (2022) · Frontiers in Genetics · doi:10.3389/fgene.2022.977810 · Cited by 6
“through changes to the stability of the local structure (Pak et al., 2021) and/or changes in aggregation potential, as is the case with sickle cell disease (Lu et al.”
BackgroundRaisinghani, N., Alshahrani, M., Gupta, G., Tian, H., Xiao, S., Tao, P., & Verkhivker, G. M. (2024) · The Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.4c04985 · Cited by 5
Unmasking AlphaFold: integration of experiments and predictions in multimeric complexesPreprint
Mirabello, C., Wallner, B., Nystedt, B., Azinas, S., & Carroni, M. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.09.20.558579 · Cited by 5
“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 (27).”
BackgroundChuah, A., Li, S., Do, A., Field, M., & Andrews, D. (2022) · Bioinformatics · doi:10.1093/bioinformatics/btac465 · Cited by 5
Calia, C. N., & Paesani, F. (2025) · The Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.5c05027 · Cited by 4
The rise of AlphaFold in drug design
Stecuła, A., Paul, R., Litchfield, K., Dalton, S. E., Low, C. M. R., Reis, C. R., & Congreve, M. (2025) · Progress in medicinal chemistry · doi:10.1016/bs.pmch.2025.09.002 · Cited by 7
Limitations of the refolding pipeline for de novo protein design
Korbeld, K. T., Viliuga, V., & Fürst, M. J. L. J. (2026) · Protein Science · doi:10.1002/pro.70613 · Cited by 5
“These findings extend the widely held view that structure prediction models cannot reliably infer the effects of single mutations on stability or fold (Buel & Walters, 2022; Pak et al., 2023).”
Bacterial Chemoreceptors Transmit Stimulus Signals Through Coupled Entropic Switches
Flack, C. E., & Parkinson, J. S. (2025) · Journal of Molecular Biology · doi:10.1016/j.jmb.2025.169614 · Cited by 4
An evolution-conserved allosteric network in human tubulin governs paclitaxel efficacy
LUO, J., Khoo, C. J., Chen, W., Liu, Z., Li, B., Lau, W. S., Li, X., & Ti, S.-C. (2026) · Nature Chemical Biology · doi:10.1038/s41589-026-02204-2 · Cited by 6
Padilla‐Vaca, F., de la Mora, J., García‐Contreras, R., Ramírez-Prado, J. H., Vicente-Gómez, M., Vargas-Gasca, F., Anaya-Velázquez, F., Páramo-Pérez, I., Rangel‐Serrano, Á., Cuéllar‐Mata, P., et al. (2023) · PeerJ · doi:10.7717/peerj.16309 · Cited by 4
Deep generative models of LDLR protein structure to predict variant pathogenicity
James, J. K., Norland, K., Johar, A., & Kullo, I. J. (2023) · Journal of Lipid Research · doi:10.1016/j.jlr.2023.100455 · Cited by 6
Hawes, E. M., Claxton, D. P., Oeser, J. K., & O’Brien, R. M. (2024) · Bioscience Reports · doi:10.1042/bsr20231851 · Cited by 5
Spotlight on plasticity-related genes: Current insights in health and disease
Brandt, N., Köper, F., Hausmann, J., & Bräuer, A. U. (2024) · Pharmacology & Therapeutics · doi:10.1016/j.pharmthera.2024.108687 · Cited by 4
Identification of the protease inhibitory domain of Trichinella spiralis novel cystatin (TsCstN)
Yuthithum, T., Phuphisut, O., Reamtong, O., Kosoltanapiwat, N., Chaimon, S., Kobpornchai, P., Thawornkuno, C., Malaithong, P., Sawatdichaikul, O., & Adisakwattana, P. (2024) · Parasites Hosts and Diseases · doi:10.3347/phd.24026 · Cited by 4
“Recently, AlphaFold has been extensively used in several structural biology fields, such as drug discovery, protein design, and protein function prediction [25-27].”
BackgroundHassi, K., Weston, T., Rinaldi, G., Ng, J. C.-F., Smahi, A., Twelves, S., Davan-Wetton, C. S. A., Fakhreddine, D., Fraternali, F., & Capon, F. (2023) · Journal of Investigative Dermatology · doi:10.1016/j.jid.2023.06.191 · Cited by 7
Unraveling the impact of ORF3a Q57H mutation on SARS-CoV-2: insights from molecular dynamics
Islam, M. J., Alom, M. S., Hossain, M. S., Ali, M., Akter, S., Islam, S., Ullah, M. O., & Halim, M. A. (2023) · Journal of Biomolecular Structure and Dynamics · doi:10.1080/07391102.2023.2252908 · Cited by 5
“…physical and biological knowledge about protein structure, leveraging multi-sequence alignments, and predicting the highly accurate 3D structure of proteins from their amino acid sequence (Pak et al., 2023; Varadi et al., 2022), and model quality was checked by PROCHECK (Laskowski et al., 1993).”
Vincenzi, M., Mercurio, F. A., Autiero, I., & Leone, M. (2024) · Molecules · doi:10.3390/molecules29051024 · Cited by 5
“However, it needs to be pointed out that, according to previous studies, AF2 is unable to precisely predict the structural outcomes of missense mutations by employing, as input, the sequence of a mutated protein as there will be a bias towards wild-type or homologous sequences [60,61].”
BackgroundThe impact of genetically controlled splicing on exon inclusion and protein structure
Einson, J. E., Minaeva, M., Rafi, F., & Lappalainen, T. (2024) · PLoS ONE · doi:10.1371/journal.pone.0291960 · Cited by 4
Singh, K., Nainwal, N., & Chitme, H. R. (2024) · Annales Pharmaceutiques Françaises · doi:10.1016/j.pharma.2024.08.001 · Cited by 7
Predicted binding interface between coronavirus nsp3 and nsp4Preprint
Hensel, Z. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.03.05.483145 · Cited by 5
“Importantly, AlphaFold2 models are not trained to directly predict the impact of mutations and may not perform well in this task (Pak et al. 2021).”
BackgroundThe impact of AlphaFold on experimental structure solutionPreprint
Edich, M., Briggs, D. C., Kippes, O., Gao, Y., & Thorn, A. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.04.07.487522 · Cited by 4
Britton, B. M., Yovanno, R. A., Costa, S. F., et al. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.11.27.518129 · Cited by 4
Predicting MmpR-based bedaquiline resistance using sequence- and structure-based featuresPreprint
Rivière, E., Safarpour, M., Meysman, P., Cuypers, B., Meehan, C. J., Laukens, K., & Rie, A. V. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.12.19.521152 · Cited by 3
Najafi, N. N., Karbassian, R., Hajihassani, H., & Irani, M. A. (2025) · Journal of Molecular Modeling · doi:10.1007/s00894-025-06392-x · Cited by 4
“However, they found no correlation between Alpha-Fold’s predictions and experimental outcomes[73].”
Cormier, K. E., Kantor, B. L., Rajan, K. C., Blanco-Fernandez, X., & Galazo, M. J. (2025) · Frontiers in Neuroscience · doi:10.3389/fnins.2025.1605762 · Cited by 3
Computational Bioprospecting of Enzymes
Xiang, R., Martinez-Sugrañes, M., Muñoz‐Tafalla, R., Floor, M., & Guallar, V. (2025) · Wiley Interdisciplinary Reviews Computational Molecular Science · doi:10.1002/wcms.70037 · Cited by 5
“While AF2's predictive metrics, such as pLDDT, can highlight regions of high structural confidence, they do not necessarily guarantee solubility, stability, or catalytic efficiency [30, 31].”
Babić, D., Abualia, R., Fiedler, L., Qi, L., Tellier, F., Smoljan, A., Bertrand-Rakusova, H., Valošek, P., Han, H., Benková, E., et al. (2025) · The Plant Journal · doi:10.1111/tpj.70396 · Cited by 4
Integrative Neoepitope Discovery in Glioblastoma via HLA Class I Profiling and AlphaFold2-Multimer
Francés, R., Bonifacio-Mundaca, J., Casafont, Í., Desterke, C., & Mata-Garrido, J. (2025) · Biomedicines · doi:10.3390/biomedicines13112715 · Cited by 4
“Clinical and genomic data for GBM-TCGA patients were retrieved from the cBioPortal for Cancer Genomics [26].”
Synergizing genome editing and artificial intelligence for predictive crop design
Gao, Z., Zhu, J., & Xie, C. (2026) · Molecular Plant · doi:10.1016/j.molp.2026.04.007 · Cited by 4
GWAS reveal SUBER GENE1-mediated suberization via type one phosphatases
Han, J., Lefebvre‐Legendre, L., Yu, J., Capitão, M. B., Beaulieu, C., Gully, K., Shukla, V., Wu, Y., Boland, A., Nawrath, C., et al. (2026) · Nature Plants · doi:10.1038/s41477-026-02292-x · Cited by 3
“Using AlphaFold, which is known to have a limited capacity to accurately predict the structural consequences of point mutations 35,36 , the predicted SBG1_SM-TOPP2 and SBG1_DM-TOPP2 complexes showed only a modest reduction in prediction confidence relative to the wild-type complex (Extended Data…”
AlphaFold for a medicinal chemist: tool or toy?
Ivanenkov, Y. A., Evteev, S. A., Малышев, А. В., Terentiev, V. A., Bezrukov, D. S., Ereshchenko, A., Korzhenevskaya, A. A., Zagribelnyy, B. A., Шегай, П. В., & Каприн, А. Д. (2024) · Russian Chemical Reviews · doi:10.59761/rcr5107 · Cited by 3
Zheng, W. (2024) · PLoS ONE · doi:10.1371/journal.pone.0302504 · Cited by 3
Machmouchi, D., Courageot, M.-P., Ogire, E., Redecke, L., Kohl, A., Desprès, P., & Roche, M. (2024) · PLoS neglected tropical diseases · doi:10.1371/journal.pntd.0012146 · Cited by 4
Rinne, A., & Buenemann, M. (2024) · Biomolecules · doi:10.3390/biom14121652 · Cited by 4
Köller, Z., Németh, B. Z., Kiss, B., Nagy, Z. A., Schlosser, G., Magyar, C., Demcsák, A., Sahin‐Tóth, M., & Pál, G. (2025) · Protein Science · doi:10.1002/pro.70146 · Cited by 4
“Based on the literature, some studies support (McBride et al. 2023) while others question (Laurents 2022; Pak et al. 2023) this capacity of the software.”
Kapitan, V. Y., & Choi, M. C. H. (2025) · BMC Bioinformatics · doi:10.1186/s12859-025-06185-2 · Cited by 4
“Despite their success, machine learning methods have limitations, particularly in their ability to account for the basic physical and chemical properties of the systems they model [45–47].”
A New GlyT2 Variant Associated with Hyperekplexia
Sarmiento-Jiménez, J., Felipe, R., Núñez, E., Ferrando-Muñoz, A., Benito-Muñoz, C., Gago, F., Vázquez, J., Camafeita, E., Clement, E. M., Wilson, B. T., et al. (2025) · International Journal of Molecular Sciences · doi:10.3390/ijms26146753 · Cited by 4
“…α atoms) and between the wild-type and the pathogenic G449E variant (0.18–0.25 Å over 461–481 C α atoms) are in line with the current lack of sensitivity of this revolutionary methodology to single amino acid replacements that are known to have a large impact on protein stability and function [60].”
InfluentialSacchi, S., Comani, V. B., Arisi, I., Marchesani, F., Rabattoni, V., Bei, O. D., Motta, Z., Peracchi, A., Bruno, S., Pollegioni, L., et al. (2025) · FEBS Journal · doi:10.1111/febs.70169 · Cited by 4
“These differences may be absent at the 3D structural level, or it is possible that the advancements in AF3 are still insufficient to overcome the limitations previously identified in ALPHAFOLD regarding the prediction of structural effects induced by single substitutions [21].”
AlphaFold2 captures conformational transitions in the voltage-gated channel superfamily
Tao, E., & Corry, B. (2025) · Biophysical Journal · doi:10.1016/j.bpj.2025.08.033 · Cited by 7
Strategies to identify and edit improvements in synthetic genome segments episomally
Rudolph, A., Nyerges, A., Chiappino-Pepe, A., Landon, M., Baas-Thomas, M., & Church, G. (2023) · Nucleic Acids Research · doi:10.1093/nar/gkad692 · Cited by 3
“Although this enormous undertaking represents a significant leap forward in computational biology, recent work has indicated the need for cautious optimism regarding the use of AlphaFold to predict the impact of individual mutations on a protein, encouraging researchers to bear in mind that protein function and protei…”
BackgroundMethodGhalamfarsa, F., Savardashtaki, A., Irajie, C., Emami, A., Nezafat, N., & Ghasemi, Y. (2023) · Current Proteomics · doi:10.2174/0115701646244648231014153217 · Cited by 3
Influence of Model Structures on Predictors of Protein Stability Changes from Single-Point Mutations
Rollo, C., Pancotti, C., Birolo, G., Rossi, I., Sanavia, T., & Fariselli, P. (2023) · Genes · doi:10.3390/genes14122228 · Cited by 4
“Despite the significant strides made by Alphafold2 [33–35], the difficulty in accurately calculating free energies with atomistic simulations remains a well-known challenge in molecular modeling and drug design [36,37].”
Background
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.