> ## Documentation Index > Fetch the complete documentation index at: https://proto.evodesign.org/docs/llms.txt > Use this file to discover all available pages before exploring further. # UniProt > UniProt (the Universal Protein Resource) is the reference database of protein sequences and their functional annotation, maintained by the UniProt consortium. The `uniprot-fetch` tool retrieves a UniProtKB entry over the UniProt REST API, either directly by accession or through a ranked gene- or protein-name and organism search, returning the sequence, gene names, review status, linked PDB structures, and the full JSON record. It runs on CPU and requires only network access.
UniProt
SIB EMBL-EBI PIR
**License:** UniProt has a CC-BY-4.0 license and may require explicit attribution when utilized. Please refer to [the license](https://www.uniprot.org/help/license) for full terms.

Proto is not affiliated with SIB, EMBL-EBI, and PIR. This toolkit is open source and builds on the implementations produced by these organizations. Product names, logos, and trademarks are the property of their respective owners.

/
/
View repo
uniprot.org
Visit website
UniProt: the Universal Protein Knowledgebase in 2025
The UniProt Consortium
Nucleic Acids Research (2025)
Read paper
```bibtex theme={null} @article{theuniprotconsortium2025, title={UniProt: the Universal Protein Knowledgebase in 2025}, author={The UniProt Consortium}, journal={Nucleic Acids Research}, volume={53}, number={D1}, pages={D609--D617}, year={2025}, publisher={Oxford University Press}, doi={10.1093/nar/gkae1010} } ```
Copy citation
evo-design/proto-tools/proto\_tools/tools/database\_retrieval/uniprot
View source
Open Notebook
Open notebook
UniProt Fetch
Toolkit contributors
| Function | Description | | | --------------------- | ------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | `run_uniprot_fetch()` | Fetch protein entries from UniProt by accession or search by name and organism | Docs Source | ## Background UniProt ([The UniProt Consortium, 2025](https://doi.org/10.1093/nar/gkae1010)) is the central, freely accessible resource for protein sequence and functional annotation, maintained by [SIB](https://www.sib.swiss/), [EMBL-EBI](https://www.ebi.ac.uk/), and [PIR](https://proteininformationresource.org/). Its core database, UniProtKB, has two sections: [Swiss-Prot](https://en.wikipedia.org/wiki/UniProt#UniProtKB/Swiss-Prot), whose entries are manually reviewed and curated from the literature, and [TrEMBL](https://en.wikipedia.org/wiki/UniProt#UniProtKB/TrEMBL), whose entries are automatically annotated. As of release 2026\_01 (January 2026), Swiss-Prot contains 574,627 reviewed entries, alongside hundreds of millions of unreviewed TrEMBL entries; current counts are published on the [UniProt statistics page](https://www.uniprot.org/uniprotkb/statistics). Internally, the tool calls the UniProt REST API at `rest.uniprot.org`. Given an accession it fetches that UniProtKB entry directly; given a protein or gene name and an organism it runs a UniProt search and selects one entry deterministically, preferring an exact gene-name match, then optionally entries with linked PDB structures, then reviewed Swiss-Prot status. It extracts the sequence, length, review status, gene symbols, and PDB cross-references, and also returns the complete entry JSON; the `fields` option narrows the API response. Results reflect the live database at query time rather than a fixed release snapshot. Records and their provenance come directly from UniProt's official REST API, maintained by the UniProt consortium. ### Learning Resources * [UniProt help and documentation](https://www.uniprot.org/help) (UniProt) - official documentation covering accessions, query syntax, return fields, and the REST API. * [Exploring protein sequence and functional information](https://www.ebi.ac.uk/training/online/courses/uniprot-exploring-protein-sequence-and-functional-info/) (EMBL-EBI Training) - a guided introduction to UniProt's data and how to query it. ## Tools
### UniProt Fetch (`uniprot-fetch`) Retrieves a single UniProtKB entry, either by accession or by a ranked name-and-organism search, and returns its sequence, length, gene names, review status, PDB cross-references, source URL, and the full entry JSON. #### API Reference
Source UniProt accession for direct entry lookup. Gene or protein name for search-based lookup. Organism name for disambiguation during search. When searching, prefer entries that have linked PDB structures. Maximum number of search results to evaluate when ranking.
Source UniProt's `fields=` query parameter — if set, restrict the API response to these fields. `None` (default) returns the full entry (\~880 KB for human TP53); a targeted selection can shrink it \~1000x. Caveat: typed Output fields are only populated when the corresponding API field is included, so callers using `accession` / `sequence` / `entry_type` / `gene_names` / `pdb_crossrefs` must include `"accession"` / `"sequence"` / `"reviewed"` / `"gene_names"` / `"xref_pdb"`. Search-mode ranking reads `reviewed` / `gene_names` / `xref_pdb`. Full list: Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). `True` is coerced to `1` and `False` to `0`. Device to run the tool on. Maximum execution time in seconds. `None` waits indefinitely. Random seed. When set, tools run reproducibly up to small GPU float noise (see `BaseToolOutput.approx_equal`), and the seed participates in cache keys. When None, cacheable seed-sensitive tools skip cache until seeded.
Source Primary UniProt accession. Protein sequence string. Sequence length. Review status (e.g. 'UniProtKB reviewed (Swiss-Prot)' for curated entries). Extracted gene name symbols. PDB structure IDs linked to this protein entry. UniProt entry URL. Complete UniProt JSON record for advanced programmatic access.
#### Applications Use this to pull a reference protein sequence and its annotation into a pipeline: fetch a target by accession before sequence design or optimization, resolve a gene symbol plus organism to a canonical reviewed entry, or discover which experimental structures are linked to a protein before structure-based work. The returned PDB identifiers feed directly into the [PDB](https://bio-pro.mintlify.app/tools/database-retrieval/pdb) and [AlphaFold DB](https://bio-pro.mintlify.app/tools/database-retrieval/alphafold-db) tools. #### Usage Tips * **Provide either `uniprot_id` or both `target_name` and `organism`.** An accession does a direct lookup; a name requires the organism to disambiguate, and the search returns the single best-ranked entry, not a list. * **`prefer_pdb_crossref` only affects search ranking.** It biases the name-and-organism search toward entries with linked PDB structures; it has no effect on a direct accession lookup and never filters out entries that lack structures. * **`fields` narrows the response but can blank typed outputs.** Restricting `fields` shrinks large entries substantially, but the typed outputs are only populated when their source fields are kept, so include `accession`, `sequence`, `reviewed`, `gene_names`, and `xref_pdb` if you read those. * **Results track the live database.** The same call can return updated annotation as UniProt releases change; it is not pinned to a fixed release.
## Toolkit Notes These apply to every UniProt tool in this toolkit (`uniprot-fetch`). * **Requires network access.** The tool calls the live UniProt REST API; it does not run offline and keeps no local copy of the database. * **Subject to UniProt rate limits.** Large or rapid batches may be throttled by the UniProt API; space out high-volume requests. * **Runs on CPU.** There is no model and no GPU; latency is dominated by the network round-trip. **Example notebook:** See the [full working example](https://github.com/evo-design/proto-tools/blob/47e34afa5ea240a3b406e323dc38aa5dc85f223e/proto_tools/tools/database_retrieval/uniprot/examples/example.ipynb) for a copy-paste-ready walkthrough. ## Infrastructure Guides The following guides cover how to run tools efficiently and at scale. Keep a tool's model warm across calls instead of reloading it every invocation. How GPUs are allocated to tools and how to target specific devices. Fan a batch of inputs out across multiple GPUs.