---
title: UniProt Search on the platform
description: >-
  Search the UniProt protein database for sequences, annotations, GO terms, and
  cross-references.
tags:
  - uniprot
---

UniProt Search lets you look up protein sequences and annotations — function, disease associations, GO terms, domains, and cross-references to PDB, ChEMBL, and AlphaFold. Reach for it to characterize a therapeutic target, retrieve a sequence for structure prediction, or navigate between databases by accession. It often follows a literature search: names extracted by [PubMed](/docs/pubmed) or [BioBERT](/docs/biobert) become the terms you look up here. See the [Databases overview](/docs/databases) for where this fits.

## Try it

**Try this prompt** — [open in a guest chat](/chat/new?guest=true&prompt=Find%20all%20human%20kinases%20in%20UniProt%20with%20GO%20term%20'protein%20kinase%20activity').

> Find all human kinases in UniProt with GO term 'protein kinase activity'

_Runs a UniProt annotation search across function and GO-term metadata._

**Try this prompt** — [open in a guest chat](/chat/new?guest=true&prompt=Get%20the%20sequence%20for%20P04637%20from%20UniProt).

> Get the sequence for P04637 from UniProt

## What UniProt Search does

UniProt Search queries the **Universal Protein Resource (UniProt)**, the most comprehensive protein knowledge base with over **250 million protein entries**. UniProt is a collaboration between the **European Bioinformatics Institute (EMBL-EBI)**, the **Swiss Institute of Bioinformatics (SIB)**, and the **Protein Information Resource (PIR)**. On the platform, UniProt Search provides direct API access to the UniProt REST API, returning protein records with functional annotations, sequences, and cross-references in 3-8 seconds.

UniProt consists of two main sections:
- **UniProt/SwissProt**: Manually reviewed, high-quality annotations (~570,000 entries). Entries are curated by expert biologists with experimental evidence.
- **UniProt/TrEMBL**: Computationally annotated entries (~250 million entries). Automatically generated from genome sequencing projects.

## Citations

> The UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025. *Nucleic Acids Research*. 2025;53(D1):D609-D617. doi: [10.1093/nar/gkae1010](https://doi.org/10.1093/nar/gkae1010)

## Key capabilities

- **Protein lookup by accession**: Retrieve a specific protein entry using its UniProt accession ID (e.g., P00533 for human EGFR)
- **Gene-based search**: Find proteins by gene symbol (e.g., TP53, BRCA1, EGFR)
- **Free-text search**: Search by protein name, function description, disease association, or any keyword
- **Organism filtering**: Restrict results to a specific species (all organisms unless you name one)
- **Sequence retrieval**: Get the full amino acid sequence for direct chaining into ESMFold or BLAST
- **Feature annotation**: Access domain annotations, post-translational modification sites, active sites, and binding sites
- **Cross-reference mapping**: Map between UniProt, PDB, ChEMBL, AlphaFold, Pfam, InterPro, and Gene IDs

## How it works technically

UniProt Search on the platform wraps the [UniProt REST API](https://rest.uniprot.org) with two access patterns:

### Direct accession lookup

When you provide a UniProt accession ID, the tool fetches the specific entry from `https://rest.uniprot.org/uniprotkb/{accession}`. This is the fastest path and returns the most complete data.

### Search queries

When you provide a gene name or free-text query, the tool constructs a search against `https://rest.uniprot.org/uniprotkb/search` with Lucene-style query parameters. The search supports:

- Gene name queries: `(gene:EGFR) AND (organism_name:"Homo sapiens")`
- Free-text queries: `(protein kinase inhibitor) AND (organism_name:"Homo sapiens")`
- Combined queries with organism filtering

### Response processing

The tool extracts and flattens key fields from UniProt's nested JSON response:

1. **Identification**: Accession, entry name, entry type (SwissProt vs. TrEMBL)
2. **Protein description**: Recommended name, alternative names
3. **Gene information**: Gene name, synonyms
4. **Organism**: Scientific name, common name, taxonomy ID
5. **Function**: Functional annotation, catalytic activity, pathway involvement
6. **Disease associations**: Known disease links with evidence codes
7. **GO terms**: Molecular function, biological process, cellular component annotations
8. **Features**: Domains, active sites, binding sites, PTMs, signal peptides
9. **Sequence**: Full amino acid sequence and length
10. **Cross-references**: PDB structures, ChEMBL targets, AlphaFold models, Pfam domains

## When to use UniProt Search

### Primary use cases

**Target identification and characterization**
- Look up a therapeutic target to understand its function, localization, and disease associations
- Retrieve the protein sequence for structure prediction with ESMFold
- Identify known PDB structures for the target via cross-references
- Assess druggability through domain and active site annotations

**Protein sequence retrieval for downstream analysis**
- Fetch sequences for ESMFold structure prediction
- Get sequences for BLAST similarity search
- Retrieve sequences for ESM-3 embedding generation
- Source sequences for molecular dynamics or docking workflows

**Cross-database navigation**
- Map a UniProt accession to PDB structure IDs
- Find ChEMBL target IDs for bioactivity queries
- Locate AlphaFold predicted structures
- Identify Pfam domain families and InterPro annotations

**Target passport data sourcing**
- UniProt annotations feed into Target Passport generation
- Function descriptions, disease associations, and druggability markers inform target assessment
- GO terms provide pathway and localization context

### When to use alternatives

| Need | Use instead |
|------|-------------|
| Protein domain analysis only | InterPro Search |
| Gene expression data | HPA Tissue Expression, GTEx |
| Drug-gene interactions | DGIdb Interactions |
| Bioactivity data | ChEMBL Activity Search |
| Protein structure (experimental) | PDB Search |
| Protein structure (predicted) | ESMFold, AlphaFold Fetch |
| Nucleotide sequences | Entrez Search (nucleotide database) |

## Integration with the platform ecosystem

### Upstream data sources

UniProt Search accepts input from:
- **Free-text chat queries**: Protein names, gene symbols, or accession IDs
- **BioBERT results**: Protein and gene names extracted from literature
- **PubMed Search**: Target names identified in publications
- **Entrez Gene**: Gene IDs linked to UniProt accessions

### Downstream applications

UniProt results feed into:
- **ESMFold**: Protein sequences for AI-based structure prediction
- **BLAST**: Sequences for similarity searching across databases
- **ESM-3**: Protein sequences for embedding generation
- **AlphaFold Fetch**: UniProt accessions for pre-computed structure retrieval
- **DiffDock**: Protein sequences or PDB cross-references for docking
- **P2Rank**: Structure-based binding site prediction
- **Target Passport**: Comprehensive target annotation aggregation

### Workflow chaining examples

**Target-to-structure pipeline:**
1. UniProt Search for "human EGFR"
2. ESMFold for structure prediction (or PDB Search via cross-reference)
3. P2Rank for binding pocket prediction
4. DiffDock for ligand docking

**Literature-to-target pipeline:**
1. PubMed Search for disease-related publications
2. BioBERT entity extraction for protein names
3. UniProt Search for each extracted protein
4. Target Passport generation

## Scope and limitations

### Coverage

- **SwissProt**: ~570,000 manually curated entries with high-quality annotations
- **TrEMBL**: ~250 million computationally annotated entries
- **Organisms**: All sequenced organisms, with best coverage for model organisms
- **Update frequency**: Monthly major releases, with continuous updates between releases

### Limitations

- **TrEMBL quality**: Automatically annotated entries may have incomplete or incorrect functional annotations. SwissProt entries are more reliable.
- **Organism bias**: Best annotation coverage for human, mouse, rat, and well-studied model organisms. Annotations for non-model organisms may be sparse.
- **Isoform coverage**: UniProt represents canonical sequences by default. Alternative isoforms exist but may not be returned unless specifically requested.
- **Result limit**: The platform returns up to 25 entries per search to keep response times fast. The full UniProt database may contain many more matches.
- **Sequence length**: Some very large proteins (>5000 residues) may cause downstream issues with structure prediction tools.

### Performance characteristics

- **Typical runtime**: 3-8 seconds for accession lookup, 5-15 seconds for search queries
- **Maximum results**: 25 entries per query (5 by default; ask for more or fewer)
- **Licensing**: UniProt data is available under CC BY 4.0 (attribution required)

## References

- UniProt: [https://www.uniprot.org/](https://www.uniprot.org/)
- UniProt REST API documentation: [https://www.uniprot.org/help/api](https://www.uniprot.org/help/api)
- The UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025. *Nucleic Acids Res.* 2025;53(D1):D609-D617. doi: [10.1093/nar/gkae1010](https://doi.org/10.1093/nar/gkae1010)

## Chat Usage

### Example Prompts

- "Search UniProt for human EGFR"
- "Get the sequence for P04637 from UniProt"
- "Find all human kinases in UniProt with GO term 'protein kinase activity'"

### Workflow Chaining

- **Feeds into**: ESMFold (structure prediction), BLAST (similarity search), ESM-3 (embeddings), Target Passport
- **Accepts from**: BioBERT (extracted protein names), literature search results

## FAQ

### How is UniProt Search different from BioTite?

  UniProt Search queries the protein annotation database for sequences and metadata. BioTite performs structural bioinformatics analysis. Use UniProt to find a protein, then chain into structure prediction or analysis tools.

### Can fetched sequences be used in other tools?

  Yes. Sequences from UniProt chain directly into ESMFold (structure prediction), BLAST (similarity search), or ESM-3 (embeddings).

## Read next