Student or educator? Sign up with your school email and get 6 months of Pro free.

The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index; 18 of the 18 sources its claims cite were checked for support.

AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences

Open access

Abstract

The AlphaFold Database Protein Structure Database (AlphaFold DB, https://alphafold.ebi.ac.uk) has significantly impacted structural biology by amassing over 214 million predicted protein structures, expanding from the initial 300k structures released in 2021. Enabled by the groundbreaking AlphaFold2 artificial intelligence (AI) system, the predictions archived in AlphaFold DB have been integrated into primary data resources such as PDB, UniProt, Ensembl, InterPro and MobiDB. Our manuscript details subsequent enhancements in data archiving, covering successive releases encompassing model organisms, global health proteomes, Swiss-Prot integration, and a host of curated protein datasets. We detail the data access mechanisms of AlphaFold DB, from direct file access via FTP to advanced queries using Google Cloud Public Datasets and the programmatic access endpoints of the database. We also discuss the improvements and services added since its initial release, including enhancements to the Predicted Aligned Error viewer, customisation options for the 3D viewer, and improvements in the search engine of AlphaFold DB.

Overview

CoChat Summary

TL;DR

AlphaFold DB has grown from ~300k to over 214 million predicted protein structures as of September 2023, with expanded access, search, and visualization features. This 2024 update describes data archiving, access mechanisms, integration with core resources, and new clustering/BLAST/PAE viewer capabilities.

Key findings

  • AlphaFold DB expanded from approximately 300k structures in 2021 to over 214 million predicted protein structures as of September 2023; the initial July 2021 release housed over 360,000 structures for 20 model organism proteomes.
  • Predictions are integrated into resources including PDB, UniProt, Ensembl, InterPro and MobiDB, and are accessible via FTP, Google Cloud Public Datasets (~23 TiB), API, and the website; PAE data is unavailable from EMBL-EBI public FTP.
  • Structure similarity clustering used MMseqs2 to cluster 214 million UniProtKB sequences into 52 million clusters, then Foldseek secondary clustering yielded 18.8 million clusters, finalizing 2.30 million robust clusters each with at least a pair of structures.
  • New features include BLAST sequence similarity search, improved search filters for reviewed/unreviewed and reference proteome status, enhanced PAE viewer–Mol* interactivity, and a November 2022 update fixing a numerical bug that affected ~4% of total structure predictions.

Methods

  • Descriptive database update/report with no primary experimental design; data generation is carried out by Google DeepMind, with predictions stored in PDB, mmCIF (modelCIF), and binaryCIF formats plus JSON metadata and confidence metrics pLDDT and PAE.
  • Archiving uses successive releases from UniProt reference proteomes (e.g., 2021_02, 2021_04), Swiss-Prot, WHO global health proteomes, and MANE select; as of September 2023, EMBL-EBI FTP hosts TAR files for proteomes of 48 organisms.
  • Access and analysis infrastructure includes FTP, Google Cloud Public Datasets, an API keyed on UniProt accessions, Apache-Solr search, BLAST on Google Cloud Platform, and Foldseek/MMseqs2 clustering.

Limitations

  • Stated: sequences are not covered if they are less than 16 amino acids; >2700 aa for Swiss-Prot/proteome sequences or >1280 aa for other UniProt sequences; contain non-standard amino acids; are not in the UniProt ‘one sequence per gene’ FASTA file; or are viral proteins; these limitations are under discussion.
  • Stated: some sequences may be outdated compared to UniProt due to less frequent AlphaFold DB releases; PAE data is unavailable from EMBL-EBI public FTP; full Google Cloud dataset download requires a Google account and can take roughly 2.5 days at 1 Gbps.
  • Evident from the text: all archived structures are predicted models rather than experimental structures, and users must interpret pLDDT/PAE cautiously (e.g., pLDDT <50 likely disordered, 50–70 lower confidence); the database also currently lacks isoform and multimeric predicted structures, which are listed as future work.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/nar/gkad1011

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Graphical Abstract
Graphical Abstract CC BY · PubMed Central
Figure 1. The expansion of AlphaFold DB. The AlphaFold Protein Structure Database increased in size through consecutive releases. As of September 2023, it archives over 214 million predicted protein structures.
Figure 1. The expansion of AlphaFold DB. The AlphaFold Protein Structure Database increased in size through consecutive releases. As of September 2023, it archives over 214 million predicted protein structures.CC BY · PubMed Central
Figure 2. Improve search results UI. The improved search UI includes more filtering options based on the underlying sequence of the protein structures and easier access to the most popular organisms.
Figure 2. Improve search results UI. The improved search UI includes more filtering options based on the underlying sequence of the protein structures and easier access to the most popular organisms.CC BY · PubMed Central
Figure 3. Sequence similarity search results. We added support for performing sequence similarity searches in AlphaFold DB. The search results page displays a list of predicted structures with sequences similar to the user's.
Figure 3. Sequence similarity search results. We added support for performing sequence similarity searches in AlphaFold DB. The search results page displays a list of predicted structures with sequences similar to the user's.CC BY · PubMed Central
Figure 4. Structure similarity cluster members. Using data from AFDB Clusters, we display lists of AlphaFold predictions structurally similar to a protein of interest.
Figure 4. Structure similarity cluster members. Using data from AFDB Clusters, we display lists of AlphaFold predictions structurally similar to a protein of interest.CC BY · PubMed Central
Figure 5. Improved support for highlighting non-consecutive regions. The new version of the interactive PAE viewer makes it easier to distinguish between highlighted non-consecutive regions when assessing their relative positions' confidence, as shown for AlphaFold DB accession https://alphafold.ebi.ac.uk/entry/Q7RTU9.
Figure 5. Improved support for highlighting non-consecutive regions. The new version of the interactive PAE viewer makes it easier to distinguish between highlighted non-consecutive regions when assessing their relative positions' confidence, as shown for AlphaFold DB accession https://alphafold.ebi.ac.uk/entry/Q7RTU9.CC BY · PubMed Central
Figure 6. Improved customisation in Mol*. Enhanced customisation options in Mol* allow users to perform popular actions such as measuring distances or changing the rendering style.
Figure 6. Improved customisation in Mol*. Enhanced customisation options in Mol* allow users to perform popular actions such as measuring distances or changing the rendering style.CC BY · PubMed Central

Verification

Claims the paper makes

11 Valid · 1 Not valid · 0 Not checked · 28 Own results and interpretations

  • Probably not a substantial claim. 30
  • It may state two separate claims as one. 15
  • The quote only partly states this claim. 1

Create a free account to see these claims.

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?

Citations

Cites

Show all 29 references

Cited by

See all 2,175

Metadata

Authors
Varadi, M., Bertoni, D., Magana, P., Paramval, U., Pidruchna, I., Radhakrishnan, M., Tsenkov, M., Nair, S., Mirdita, M., Yeo, J., Kovalevskiy, O., Tunyasuvunakool, K., Laydon, A., Žídek, A., Tomlinson, H., Hariharan, D., Abrahamson, J., Green, T., Jumper, J., Birney, E., Steinegger, M., Hassabis, D., & Velankar, S.
Affiliations
  • European Molecular Biology Laboratory, European Bioinformatics Institute , Hinxton, UK
  • European Bioinformatics Institute (GB)
  • Seoul National University (KR)
  • Google DeepMind (United Kingdom) (GB)
Corresponding author
  • Hassabis, Demis
  • Velankar, Sameer
Venue
Nucleic Acids Research
Published
2023-11-02
Volume
52
Issue
D1
Pages
D368-D375
ISSN
  • 0305-1048 (ISSN-L)
  • 1362-4962
OpenAlex
W4388464011
Cited by
2,052
References
29
Citations by year
2026: 724 · 2025: 1015 · 2024: 307 · 2023: 5
FWCI
181.98
Type
journal article
Language
en
Open access
Locations
  • Nucleic Acids Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Seoul National University Open Repository (Seoul National University) (submittedVersion)
Funders
  • Google DeepMind
  • National Research Foundation of Korea (2019R1A6A1A10073437, 2020M3A9G7103933, 2021R1C1C102065, 2021M3A9I4021220, RS-2023-00250470)
  • Samsung DS Research Fund
  • Seoul National University
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Structural Biology › Enzyme Structure and Function
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
Keywords
predicted protein structures, AlphaFold2, protein structure prediction, protein sequence coverage, data archiving, database access, search engines
MeSH terms
Amino Acid Sequence, Artificial Intelligence, Proteins, Protein Structure, Secondary, Proteome, Databases, Protein, Search Engine

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 (gpt-6-sol).

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences. https://app.cochat.ai/papers/10.1093/nar/gkad1011
The paper
Varadi, M., Bertoni, D., Magana, P., Paramval, U., Pidruchna, I., Radhakrishnan, M., Tsenkov, M., Nair, S., Mirdita, M., Yeo, J., Kovalevskiy, O., Tunyasuvunakool, K., Laydon, A., Žídek, A., Tomlinson, H., Hariharan, D., Abrahamson, J., Green, T., Jumper, J., . . . Velankar, S. (2024). AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences. Nucleic Acids Research. https://doi.org/10.1093/nar/gkad1011

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.