Uni-Fold: An Open-Source Platform for Developing Protein Folding Models beyond AlphaFold
Open accessThe DOI matches its Crossref record; 8 of the 8 references it cites were found in a registry or an index.
Abstract
Recent breakthroughs on protein structure prediction, namely AlphaFold, have led to unprecedented new possibilities in related areas. However, the lack of training utilities in its current open-source code hinders the community from further developing or adapting the model. Here we present Uni-Fold as a thoroughly open-source platform for developing protein folding models beyond AlphaFold. We reimplemented AlphaFold and AlphaFold-Multimer in the PyTorch framework, and reproduced their from-scratch training processes with equivalent or better accuracy. Based on various optimizations, Uni-Fold achieves about 2.2 times training acceleration compared with AlphaFold under similar hardware configuration. On a benchmark of recently released multimeric protein structures, Uni-Fold outperforms AlphaFold-Multimer by approximately 2% on the TM-Score. Uni-Fold is currently the only open-source repository that supports both training and inference of multimeric protein models. The source code, model parameters, test data, and web server of Uni-Fold are publicly available 3 .
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
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
UniProt: the universal protein knowledgebase in 2021
The UniProt Consortium, Bateman, A., Martin, M.-J., Orchard, S., Magrane, M., Agivetova, R., Ahmad, S., Alpi, E., Bowler-Barnett, E. H., Britto, R., et al. (2021) · Nucleic Acids Research · doi:10.1093/nar/gkaa1100 · Cited by 6,372 · cited 1×
“For multimers, we additionally used JackHMMER with UniProt [9] to search for sequences with species annotations.”
MethodAccurate prediction of protein structures and interactions using a 3-track network
Baek, M., DiMaio, F., Anishchenko, I., Dauparas, J., Ovchinnikov, S., Lee, G. R., Wang, J., Cong, Q., Kinch, L. N., Schaeffer, R. D., et al. (2021) · doi:10.1101/2021.06.14.448402 · Cited by 5,549
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 · cited 4×
“Adapted from AlphaFold, AlphaFold-Multimer [2] was later developed by the same team via training AlphaFold on multimeric protein structures.”
BackgroundMethodResultInfluentialHHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment
Remmert, M., Biegert, A., Hauser, A., & Söding, J. (2012) · Nature Methods · doi:10.1038/nmeth.1818 · Cited by 2,288
UniRef clusters: a comprehensive and scalable alternative for improving sequence similarity searches
Suzek, B. E., Wang, Y., Huang, H., McGarvey, P. B., Wu, C. H., & the UniProt Consortium (2015) · Bioinformatics · doi:10.1093/bioinformatics/btu739 · Cited by 1,928
Rasley, J., Rajbhandari, S., Ruwase, O., & He, Y. (2020) · Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3394486.3406703 · Cited by 823 · cited 1×
“Similar to previous works on Transformer acceleration [14, 15, 16], we fused the softmax and layer normalization operators.”
MethodMGnify: the microbiome analysis resource in 2020
Mitchell, A. L., Almeida, A., Beracochea, M., Boland, M., Burgin, J., Cochrane, G., Crusoe, M. R., Kale, V., Potter, S. C., Richardson, L. J., et al. (2019) · Nucleic Acids Research · doi:10.1093/nar/gkz1035 · Cited by 562
Research works printed but not linked: 14
- 10.1038/d41586-022-00997-5 · DOI · Paper page
- 2010, Hidden markov model speed heuristic and iterative HMM search procedure, BMC Informatics, 11, 1
- 10.1093/nar/gkw1081 · DOI · Paper page
- Tian-ming Zhou , Sheng Wang , and Jinbo Xu . Deep learning reveals many more inter-protein residue-residue contacts than direct coupling analysis. bioRxiv, 2018.
- 2016, Gaussian error linear units (gelus), arXiv
- 2017, Mixed precision training, arXiv
- 2020, Revisiting bfloat16 training, arXiv
- 2019, Megatron-lm: Training multi-billion parameter language models using model parallelism, arXiv
- 2022, Metro: Efficient denoising pretraining of large scale autoencoding language models with model generated signals, arXiv preprint
- 10.1007/978-1-4939-7000-1_26 · DOI · Paper page
- 2022, PSP: million-level protein sequence dataset for protein structure prediction, arXiv
- 2022, Helixfold: An efficient implementation of alphafold2 using paddlepaddle, arXiv
- 2022, Fastfold: Reducing alphafold training time from 11 days to 67 hours, arXiv
- 10.1126/science.abj8754 · doi:10.1126/science.abj8754 · Paper page
Cited by
Ahdritz, G., Bouatta, N., Floristean, C., Kadyan, S., Xia, Q., Gerecke, W., O’Donnell, T. J., Berenberg, D., Fisk, I., Zanichelli, N., et al. (2024) · Nature Methods · doi:10.1038/s41592-024-02272-z · Cited by 439
“OpenFold, unlike AlphaFold2, was trained using per-batch as opposed to per-sample gradient clipping (first noted by the Uni-Fold team 42 ).”
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
Yin, R., & Pierce, B. G. (2024) · Protein Science · doi:10.1002/pro.4865 · Cited by 145
Modelling protein complexes with crosslinking mass spectrometry and deep learning
(2024) · doi:10.1038/s41467-024-51771-2 · Cited by 94
Challenges in bridging the gap between protein structure prediction and functional interpretation
Varadi, M., Tsenkov, M., & Velankar, S. (2025) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.26614 · Cited by 36
Metadata
- Authors
- Li, Z., Liu, X., Chen, W., Shen, F., Bi, H., Ke, G., & Zhang, L.
- Venue
- bioRxiv (Cold Spring Harbor Laboratory)
- Publisher
- openRxiv
- Published
- 2022-08-06
- OpenAlex
- W4290005562
- Cited by
- 59
- References
- 8
- Type
- posted content
- Language
- en
- Open access
- green
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
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.