The impact of AI ‐based modeling on the accuracy of protein assembly prediction: Insights from CASP15
Open accessPublished version of a preprint
Abstract
In CASP15, 87 predictors submitted around 11 000 models on 41 assembly targets. The community demonstrated exceptional performance in overall fold and interface contact predictions, achieving an impressive success rate of 90% (compared to 31% in CASP14). This remarkable accomplishment is largely due to the incorporation of DeepMind's AF2‐Multimer approach into custom‐built prediction pipelines. To evaluate the added value of participating methods, we compared the community models to the baseline AF2‐Multimer predictor. In over 1/3 of cases, the community models were superior to the baseline predictor. The main reasons for this improved performance were the use of custom‐built multiple sequence alignments, optimized AF2‐Multimer sampling, and the manual assembly of AF2‐Multimer‐built subcomplexes. The best three groups, in order, are Zheng, Venclovas, and Wallner. Zheng and Venclovas reached a 73.2% success rate over all (41) cases, while Wallner attained 69.4% success rate over 36 cases. Nonetheless, challenges remain in predicting structures with weak evolutionary signals, such as nanobody–antigen, antibody–antigen, and viral complexes. Expectedly, modeling large complexes also remains challenging due to their high memory compute demands. In addition to the assembly category, we assessed the accuracy of modeling interdomain interfaces in the tertiary structure prediction targets. Models on seven targets featuring 17 unique interfaces were analyzed. Best predictors achieved a 76.5% success rate, with the UM‐TBM group being the leader. In the interdomain category, we observed that the predictors faced challenges, as in the case of the assembly category, when the evolutionary signal for a given domain pair was weak or the structure was large. Overall, CASP15 witnessed unprecedented improvement in interface modeling, reflecting the AI revolution seen in CASP14.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Highly 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 · see citing works
ColabFold - Making protein folding accessible to all
Mirdita, M., Schütze, K., Moriwaki, Y., Heo, L., Ovchinnikov, S., & Steinegger, M. (2021) · doi:10.1101/2021.08.15.456425 · Cited by 9,391 · see citing works
Varadi, M., Anyango, S., Deshpande, M., Nair, S., Natassia, C., Yordanova, G., Yuan, D., Stroe, O., Wood, G., Laydon, A., et al. (2022) · Nucleic Acids Research · doi:10.1093/nar/gkab1061 · Cited by 7,850 · see citing works
seaborn: statistical data visualization
Waskom, M. (2021) · Journal of Open Source Software · doi:10.21105/joss.03021 · Cited by 7,310 · see citing works
Protein complex prediction with AlphaFold-Multimer
Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A., Green, T., Žídek, A., Bates, R., Blackwell, S., Yim, J., et al. (2021) · doi:10.1101/2021.10.04.463034 · Cited by 3,808 · see citing works
Potter, S. C., Luciani, A., Eddy, S. R., Park, Y., Lopez, R., & Finn, R. D. (2018) · Nucleic Acids Research · doi:10.1093/nar/gky448 · Cited by 2,607 · see citing works
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 · see citing works
Improved prediction of protein-protein interactions using AlphaFold2
Bryant, P., Pozzati, G., & Elofsson, A. (2021) · doi:10.1101/2021.09.15.460468 · Cited by 1,025 · see citing works
CAPRI: A Critical Assessment of PRedicted Interactions
Janin, J., Henrick, K., Moult, J., Eyck, L. T., Sternberg, M. J. E., Vajda, S., Vakser, I., & Wodak, S. J. (2003) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.10381 · Cited by 703 · see citing works
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 · see citing works
Benchmarking AlphaFold for protein complex modeling reveals accuracy determinants
Yin, R., Feng, B. Y., Varshney, A., & Pierce, B. G. (2021) · doi:10.1101/2021.10.23.465575 · Cited by 335 · see citing works
‘The entire protein universe’: AI predicts shape of nearly every known protein
Callaway, E. (2022) · Nature · doi:10.1038/d41586-022-02083-2 · Cited by 113 · cited 1× · see citing works
“27,28 Since the release of AF2, scientists have sought to incorporate this framework into their modeling pipelines.”
BackgroundDeep, A., Gu, Y., Gao, Y.-Q., Ego, K. M., Herzik, M. A., Zhou, H., & Corbett, K. D. (2022) · Molecular Cell · doi:10.1016/j.molcel.2022.09.008 · Cited by 106 · see citing works
Can AlphaFold2 predict protein-peptide complex structures accurately?
Ko, J., & Lee, J. (2021) · doi:10.1101/2021.07.27.453972 · Cited by 83 · see citing works
Structure of an endogenous mycobacterial MCE lipid transporter
Chen, J., Fruhauf, A., Fan, C., Ponce, J., Ueberheide, B., Bhabha, G., & Ekiert, D. C. (2023) · Nature · doi:10.1038/s41586-023-06366-0 · Cited by 69 · see citing works
New prediction categories in <scp>CASP15</scp>
Kryshtafovych, A., Antczak, M., Szachniuk, M., Zok, T., Kretsch, R. C., Rangan, R., Pham, P., Das, R., Robin, X., Studer, G., et al. (2023) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.26515 · Cited by 54 · cited 1× · see citing works
“25,85,86 Figure 4A also suggests that the size of the complex is another factor influencing modeling accuracy, as larger targets (1800 residues or more) were modeled slightly worse than the smaller ones (average ICS of 0.68 vs. 0.75 for the latter).”
BackgroundInfluentialSelf-citationLimits and potential of combined folding and docking
Pozzati, G., Zhu, W., Bassot, C., Lamb, J., Kundrotas, P., & Elofsson, A. (2022) · Bioinformatics · doi:10.1093/bioinformatics/btab760 · Cited by 25 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Kryshtafovych, A., Schwede, T., Topf, M., Fidelis, K., & Moult, J. (2023) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.26617 · Cited by 148 · see citing works
“Thus, this appears to be a clear example of single sequence being insufficient to produce a good model, a deficiency also seen in DeepMind's benchmarking 27 and in the previous CASP.”
BackgroundMethodSelf-citationAssessment of Nucleic Acid Structure Prediction in <scp>CASP16</scp>
Kretsch, R. C., Hummer, A. M., He, S., Yuan, R., Zhang, J., Karagianes, T., Cong, Q., Kryshtafovych, A., & Das, R. (2026) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.70072 · Cited by 51 · see citing works
Self-citationA fully automated benchmarking suite to compare macromolecular complexes
Studer, G., Robin, X., Bienert, S., Durairaj, J., Škrinjar, P., Tauriello, G., Waterhouse, A. M., & Schwede, T. (2026) · Nature Methods · doi:10.1038/s41592-025-02973-z · Cited by 10 · see citing works
Xia, Y., Pu, Y., Wang, S., Zhuang, J., Liu, D., Hou, M., & Zhang, G. (2025) · Journal of Molecular Biology · doi:10.1016/j.jmb.2025.169128 · Cited by 2 · see citing works
- Authors
- Ozden, B., Kryshtafovych, A., & Karaca, E.
- Affiliations
- Dokuz Eylül University (TR)
- University of California, Davis (US)
- Corresponding author
- Karaca, Ezgi
- Venue
- Proteins: Structure, Function, and Bioinformatics
- Publisher
- Wiley
- Published
- 2023-10-20
- Volume
- 91
- Issue
- 12
- Pages
- 1636-1657
- ISSN
- 0887-3585 (ISSN-L)
- 1097-0134
- PMC
- PMC10873090
- OpenAlex
- W4387798089
- Cited by
- 51
- References
- 17
- Citations by year
- 2026: 9 · 2025: 20 · 2024: 17 · 2023: 5
- FWCI
- 4.54
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Proteins Structure Function and Bioinformatics (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- eScholarship (California Digital Library) (submittedVersion, cc-by)
- PubMed Central (submittedVersion, cc-by-nc-nd)
- Funders
- European Molecular Biology Organization
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Physical Sciences › Materials Science › Materials Chemistry › Enzyme Structure and Function
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Keywords
- CASP15, protein complex structure prediction, interdomain interface, multiple sequence alignment, immune complexes, protein tertiary structure prediction
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.
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
