# How to Retrieve Peptide Sequences from UniProt: A Guide

By the end of this guide you will be able to retrieve peptide sequence from UniProt in four ways: look up a single protein and copy its sequence, search the whole knowledgebase for a short peptide you already have, batch-retrieve hundreds of sequences from a list of identifiers, and pull sequences programmatically with the UniProt REST API or the UniprotR R package. You will also know how to slice out the exact peptide region you need, verify the result, and avoid the mistakes that most often corrupt a downstream analysis.

To follow along you need a web browser and, for the scripted parts, either `curl` on Linux or macOS, or Python or R installed locally. Every command in this guide runs as written on a standard terminal. No accounts or API keys are required because UniProt is open and free under a CC BY 4.0 license [1].

## What UniProt Is and Where Peptides Live Inside It

The Universal Protein Resource (UniProt) is the primary public knowledgebase for [protein sequence](/blog/guides/protein-sequence) and functional annotation. At the 2021 release it held roughly 190 million sequences, and its website serves hundreds of thousands of unique visitors per month [1]. Before you retrieve anything, you need to understand the four connected datasets, because "peptide sequence" can mean different things in each of them.

**UniProtKB** (the UniProt Knowledgebase) is the annotated core. It contains two sections. Swiss-Prot entries are manually reviewed by expert curators, carry detailed functional annotation, and have stable, well-checked sequences. TrEMBL entries are computationally annotated and unreviewed. When you retrieve a peptide sequence for a publication, the reviewed Swiss-Prot section is almost always the safer choice.

**UniParc** is a sequence-only archive. It stores every distinct protein sequence ever loaded, without annotation, and is useful when you need to confirm whether a given sequence string exists anywhere.

**UniRef** clusters sequences at 100%, 90%, and 50% identity thresholds to reduce redundancy. UniRef100 representative sequences can also be searched with the peptide matching service [2].

**Proteomes** are the complete sets of proteins attributed to a genome or plasmid. If your goal is a background database for mass spectrometry searches or metaproteomics, download the proteome rather than searching the entire knowledgebase.

A "peptide" in this context is any amino acid string, typically short, that is either a region inside a larger protein (a mature chain, a [signal peptide](/knowledge/molecular-biology/signal-peptide), an active peptide, an epitope) or a standalone entry in its own right. UniProt annotates many of these regions on the full protein entry. Mature peptide annotations are a standard way curators mark the biologically active portion of a precursor, and this matters whenever you intend to order a synthetic peptide or model a structure.

## Choosing the Right Retrieval Method

![Decision path asking what you have and what you need, leading to the website, Peptide Search, ID Mapping, REST API or a proteome download.](/article-images/illustrations/retrieve-peptide-sequence-from-uniprot-retrieval-method-choice.webp)

The UniProt website provides four main tools: BLAST for similarity searching, Align for [multiple sequence alignment](/blog/guides/multiple-sequence-alignment), Peptide Search for finding proteins that contain a given short peptide, and Retrieve/ID Mapping for converting a list of identifiers into UniProtKB entries [3]. The table below helps you pick the right one for your task.

| Your task | Best method | Where it runs |
|--|--|--|
| Get the sequence of one known protein or its peptide region | Web entry page or a single REST call | Browser or `curl` |
| Find which proteins contain a peptide you measured or synthesized | Peptide Search tool | Browser |
| Retrieve 2 to 20,000 sequences from a list of accessions or gene names | Retrieve/ID Mapping | Browser or REST job API |
| Pull sequences inside a script or pipeline | REST API with `curl`, Python, or R | Terminal |
| Download every protein of an organism for database searches | Proteome download or streaming API | Browser or `curl` |
| Retrieve and visualize sequences plus annotation in one R session | UniprotR package | R console |

## Method 1: Retrieving a Single Sequence on the UniProt Website

This is the fastest way to get one sequence and the best starting point if you are learning how to use UniProt.

1. Go to `https://www.uniprot.org`.
2. Type your identifier or name into the search box. Try `P01308` (human insulin) or a protein name such as `insulin human`.
3. The results page lists matching entries. Check the "Reviewed" column. A gold star with "reviewed" means Swiss-Prot, a grey check means TrEMBL. For any downstream analysis where accuracy matters, prefer the reviewed entry.
4. Open the entry. The sequence appears in the "Sequence" section, split into annotated regions such as signal peptide, chain, and peptides.
5. Click "Download" and choose FASTA, or simply select and copy the sequence text from the Sequence viewer.

The FASTA you get looks like this (abbreviated):

```
>sp|P01308|INS_HUMAN Insulin OS=Homo sapiens OX=9606 GN=INS PE=1 SV=1
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKT
RREAEDLQVGQVELGGGPGAGSLQPLALEGSLQKRGIVEQCCTSICSLYQLEN
YCN
```

The header line carries four useful fields. `sp` means reviewed Swiss-Prot, `P01308` is the accession, `INS_HUMAN` is the entry name, and `SV=1` is the sequence version. Note that the full insulin precursor includes the signal peptide, the B chain, the C peptide, and the A chain. If you need only the mature insulin peptide, you must slice the sequence, which Method 6 covers.

## Method 2: Peptide Search, Finding Which Proteins Contain Your Peptide

Peptide Search answers the reverse question. You have a peptide, often from mass spectrometry, phage display, or an epitope study, and you want every UniProt protein that contains it [3].

1. From the UniProt website, open the Peptide Search tool under the "Tools" menu, or go to `https://peptidesearch.uniprot.org`.
2. Paste your peptide using single-letter amino acid code, for example `SIINFEKL` (a well-known influenza-derived epitope).
3. Restrict the taxonomy if you can. Searching all of UniProtKB returns matches from every organism. Restricting to, say, Mammalia or a single species keeps the results usable.
4. Run the search. Results are grouped by taxonomy, so you can browse matches by organism or taxonomic group, and each match shows the matched sequence region with a link into the corresponding protein entry [2].
5. Click any protein link to land directly on the entry with the matching region highlighted.

The underlying service is a fast Lucene-based index of UniProtKB that is rebuilt with every UniProt release, roughly every four weeks [2]. Two options deserve attention. First, the tool can treat the isobaric residues leucine and isoleucine as equivalent, which matters because mass spectrometry cannot distinguish them, so a measured peptide may map to proteins containing either residue at that position [2]. Second, you can optionally search UniRef100 representative sequences instead of all UniProtKB entries [2].

For very large matching jobs, for example scanning thousands of T cell epitopes against dozens of proteomes, a dedicated tool such as PEPMatch preprocesses proteomes with a k-mer index and runs roughly 50-fold faster than BLAST while maintaining full recall [4]. PEPMatch was built specifically for immunology use cases such as neoepitope and viral epitope matching [4], and it is a sensible choice when the built-in UniProt tool becomes the bottleneck.

## Method 3: Batch Retrieval and ID Mapping

When you have a list of accessions, gene names, or identifiers from another database, use the Retrieve/ID Mapping tool [3].

1. On the UniProt website, open "Retrieve/ID Mapping" under the Tools menu.
2. Paste your list, one identifier per line, or upload a file. Example list:

```
P01308
P02768
P55084
```

3. Choose the source format. If your list is already UniProt accessions, leave it as UniProtKB. If your identifiers come from Ensembl, RefSeq, PDB, or dozens of other sources, select the matching source database, and UniProt will convert them [3].
4. Choose the target, usually UniProtKB, and click Map.
5. On the results page, select "Download" and pick FASTA, TSV, or Excel. FASTA gives you the sequences, TSV gives you a table of annotations you can open in a spreadsheet.

The same job runs through the REST API. The ID mapping service is asynchronous, so you submit a job, poll for completion, then download the results:

```bash
curl -X POST -H "Accept: text/plain" \
  -d "ids=P01308 P02768 P55084" \
  -d "from=UniProtKB_AC-ID" \
  -d "to=UniProtKB" \
  https://rest.uniprot.org/idmapping/run
```

The response is a job ID. Poll it, then fetch the mapped accessions and follow with a streaming search for FASTA output. For small day-to-day work the website is simpler. For repeated pipelines, script the whole thing once and reuse it.

This identifier mapping step is not cosmetic. A survey of protein identifications deposited in the PRIDE proteomics repository found that protein identifiers are an unstable reference system across databases, with roughly 20% of submitted IPI entries deleted within two years, while virtually all UniProtKB entries remained either active or traceably replaced [5]. Building your workflows on UniProt accessions, and resolving external IDs to UniProt early, protects your data against exactly this kind of decay.

## Method 4: The REST API for Scripts and Pipelines

Every UniProtKB query you can run on the website also runs as a URL. This is the core of how to retrieve peptide sequences from UniProt reproducibly.

Retrieve one entry as FASTA:

```bash
curl https://rest.uniprot.org/uniprotkb/P01308.fasta
```

Search by query and get FASTA for all hits:

```bash
curl "https://rest.uniprot.org/uniprotkb/search?query=gene:INS+AND+organism_id:9606&format=fasta"
```

Get a tabular annotation table instead of sequences:

```bash
curl "https://rest.uniprot.org/uniprotkb/search?query=gene:INS+AND+organism_id:9606&format=tsv&fields=accession,id,length,sequence"
```

The key parameters are the ones you will tune most often. The table below summarizes them.

| Parameter | Values you will use | Effect |
|--|--|--|
| query | `accession:P01308`, `gene:INS`, `organism_id:9606`, `reviewed:true` | Full-text and fielded search. Combine with `AND`, `OR`, `NOT` |
| format | `fasta`, `tsv`, `json`, `gff`, `txt` | Output format. `fasta` for sequences, `tsv` for spreadsheets |
| fields | `accession,length,sequence,protein_name,...` | Which columns appear in `tsv` output |
| size | `500`, `1000`, up to `500` per page | Number of results per page, use the `Link` header for pagination |
| compressed | `true` | Gzip the response, recommended for large downloads |

Two useful query fields for peptide-focused work are `ft_peptide` and `ft_propeptide`, which target curated peptide and propeptide annotations, and `reviewed:true`, which limits results to Swiss-Prot. For example, to list reviewed human entries annotated with a mature peptide region:

```bash
curl "https://rest.uniprot.org/uniprotkb/search?query=reviewed:true+AND+organism_id:9606+AND+ft_peptide*&format=tsv&fields=accession,protein_name,ft_peptide&size=100"
```

In the `tsv` output, peptide annotations arrive with position ranges like `PEPTIDE 57..90`, which are exactly the coordinates you need for slicing in the next method.

## Method 5: Bulk Downloads for Organism-Scale Work

If you need every protein of an organism, for example as a search database for mass spectrometry, do not page through search results. Use the streaming endpoint with a proteome identifier.

```bash
curl "https://rest.uniprot.org/uniprotkb/stream?query=proteome:UP000005640&format=fasta" -o human_proteome.fasta
```

`UP000005640` is the human proteome. Every proteome page on the UniProt website shows its own identifier together with a ready-made download link. Add `&compressed=true` for large proteomes.

Database choice at this scale has real consequences. Large search databases, those above one million sequences as commonly arise in metaproteomics and proteogenomics, force strict filtering that trades sensitivity for control of false positives. A two-step strategy, where a first pass over the large database produces a smaller subset database for a second, target-decoy search, roughly doubled the number of high confidence peptide sequence matches in both metaproteomic and proteogenomic settings compared with a single one-step search [6]. The practical lesson for retrieval is simple: download the smallest, most relevant database that covers your question, and only widen it when the biology demands it.

## Method 6: Retrieval in R with UniprotR

If you work in R, the UniprotR package wraps the retrieval step and adds clustering and publication-quality visualization. The package retrieves protein sequence and functional information from UniProtKB and returns results as R data frames, CSV files, or graphical output, covering gene ontology, chromosomal location, physicochemical properties, and sequence-based visualizations [7].

```r
install.packages("UniprotR")
library(UniprotR)

ids <- c("P01308", "P02768", "P55084")

# Fetch the sequences
seqs <- GetProteinFASTA(ids)

# Pull the key annotation into a data frame
info <- GetProteinNames(ids)
```

The exact function names vary by package version, so check `?UniprotR` after installing. The value of the package is what happens after retrieval. It connects to external resources such as SWISS-MODEL and STRING, and it supports clustering proteins by primary gene name or chromosomal location, which is convenient when your peptide work feeds a comparative or proteomics analysis [7]. For peptide-focused pipelines, the typical pattern is to retrieve full protein sequences in R, slice the peptide regions with string functions, then export a FASTA of exactly the peptides you need.

## Cutting the Exact Peptide Out of a Protein Sequence

![Worked calculation showing the insulin precursor is 110 residues and that a Python slice from 89 to 110 returns residues 90 to 110.](/article-images/illustrations/retrieve-peptide-sequence-from-uniprot-slice-precursor-peptide.webp)

Most of the time your "peptide" is a defined interval of a larger protein. UniProt gives you the coordinates on the entry page in the sequence annotations, for example `Signal peptide 1..24` or `Peptide 90..121`. Slicing is a three-line exercise once you have the FASTA.

```python
from urllib.request import urlopen

accession = "P01308"
fasta = urlopen(f"https://rest.uniprot.org/uniprotkb/{accession}.fasta").read().decode()
sequence = "".join(fasta.splitlines()[1:])
print(len(sequence))          # 110, the precursor length
print(sequence[89:110])       # residues 90..110 in UniProt numbering
```

Three numbering rules prevent most slicing errors:

1. UniProt positions are 1-based, so residue 57 is `sequence` in Python.
2. A Python slice `sequence[a-1:b]` returns residues a through b inclusive in UniProt numbering.
3. Coordinates always refer to the canonical sequence displayed on that entry, unless the annotation explicitly says otherwise.

This matters because the same protein is numbered differently in different resources. Structural databases let authors number residues however they like, and some include the N-terminal signal peptide or the initiator methionine while others do not, which makes cross-resource comparison error-prone. The PDBrenum project addresses this by renumbering Protein Data Bank files according to their corresponding UniProt sequences, using the SIFTS mapping between structures and UniProt entries [8]. If you retrieved a peptide sequence from UniProt and want to align it with a crystal structure, converting the structure to UniProt numbering through PDBrenum eliminates an entire class of off-by-a-few-residues errors [8].

A worked example shows why. A bioinformatics study of secretory production of the Echinococcus granulosus antigen AgB8/1 used UniProt both to retrieve the AgB8/1 sequence and to cross-check candidate signal peptide sequences, then ran SignalP 4.1 to locate cleavage sites [9]. The retrieved precursor sequence and the predicted cleavage site together determine what the mature, secreted protein actually looks like. If you skip the cleavage-site step and order a synthetic peptide spanning the predicted signal peptide, you have synthesized a segment that no natural secreted protein ever displays.

For peptide vaccines and epitope work the same logic applies. A reverse vaccinology study of leishmaniasis retrieved twelve experimentally validated antigenic protein sequences from UniProt before running antigenicity prediction and epitope mapping [10]. A machine learning study of peptide toxin function likewise retrieved its peptide toxin dataset from UniProt before building positive-unlabeled learning models for activities such as hemolysis, neurotoxicity, and cytolytic function [11]. In both cases the retrieval step, including correct reviewed status and correct region boundaries, defined the quality ceiling of everything downstream.

## Verifying You Got the Right Sequence

Check four things before a sequence enters your pipeline.

**Length and header.** The FASTA header states the length implicitly through the sequence itself. Compare it with the "Length" field on the entry page. A mismatch means you grabbed an isoform or a partial sequence.

**Reviewed status.** The `sp` versus `tr` prefix in the FASTA header tells you Swiss-Prot versus TrEMBL. Cite the accession and sequence version (`SV=`) in your methods so the exact sequence is recoverable years later.

**Version stability.** UniProt accessions are stable identifiers. This is one of the strongest arguments for anchoring proteomics and peptide data on UniProtKB, since other database identifiers can vanish while UniProtKB entries remain active or are formally replaced by successor accessions [5].

**Independent confirmation.** If the sequence matters, run it through the UniProt BLAST tool against UniProtKB [3]. An exact self-hit with 100% identity confirms retrieval. For peptides destined for synthesis or structural work, compare against UniParc to confirm the sequence string exists, and consider a second source. Curated peptide archives such as PepBank, which aggregates peptide sequences from text mining and from UniProt among other public sources [12], provide an independent cross-check for peptides with biological activity annotations.

## Common Mistakes and Limitations

Even experienced analysts make the same handful of errors when they retrieve peptide sequence from UniProt. Each of these has cost someone a reagent order or a week of reanalysis.

**Grabbing the precursor when you wanted the mature peptide.** Many bioactive peptides are annotated as regions of precursors. The FASTA you download by default is the precursor. Always read the peptide annotation range and slice accordingly. Signal peptides and propeptides must be removed, and the cleavage site should be confirmed with a dedicated predictor rather than assumed [9].

**Confusing canonical and isoform sequences.** An entry can encode several isoforms. The Peptide Match service searches UniProtKB with isoforms included [2], so a peptide may match an isoform that is absent from the canonical FASTA you downloaded. If your peptide does not match the canonical sequence, check the isoform annotations before concluding the match failed.

**Assuming leucine and isoleucine are distinguishable.** They are isobaric, so mass spectrometry cannot tell them apart. The UniProt Peptide Match service offers an explicit option to treat I and L as equivalent for exactly this reason [2]. If you run your own matching scripts, decide the L/I policy once, document it, and apply it consistently.

**Searching the entire knowledgebase when a proteome would do.** A whole-UniProtKB peptide search returns hits from thousands of organisms. Restrict taxonomy, or download the relevant proteome and match locally. For very large matching jobs, indexed tools built for the purpose, such as PEPMatch with its k-mer preprocessing, outperform general-purpose aligners by a large factor without sacrificing recall [4].

**Building on obsolete identifiers.** Identifiers from some databases are deleted at an appreciable rate, and different cross-referencing algorithms produce markedly different remapping results [5]. Keep the UniProt accession, not a third-party ID, as your canonical key.

**Expecting modified peptides to look like plain FASTA.** UniProt FASTA is a plain one-letter-code sequence. Biologically real peptides often carry modifications that FASTA cannot express, such as the pyroglutamate residue found in the bradykinin-potentiating venom peptide Batroxin I, whose de novo sequence was reported with `<E` marking pyroglutamic acid at the N terminus [13]. Synthetic and modified peptides studied in binding or epitope work, including the modified-peptide records curated in protein-peptide binding affinity databases [14] and the synthetic epitope sub-libraries used in urease flap-region antibody studies [15], may never appear as retrievable FASTA entries. For those, the peptide sequence lives in the literature and in specialized peptide databases, and UniProt gives you the parent protein context instead.

**Ignoring release cycles.** UniProt updates on a roughly four-week release schedule, and the peptide match index is refreshed with each release [2]. A result reproduced six months later can differ because entries were added, merged, or reannotated. Record the release version with your results.

**Overlooking false positives in huge searches.** When the searched database grows past roughly a million sequences, the number of peptide-to-spectrum matches that survive strict filtering drops, and unfiltered matches accumulate false positives [6]. A smaller, curated search database, for example a reviewed proteome subset, is usually the better retrieval decision.

One limitation applies to the whole enterprise. UniProt tells you what sequences exist and what curators have annotated about them. It does not replace experimental validation of any individual peptide's activity, and any biological conclusion drawn from retrieved sequences needs laboratory confirmation.

## Applying Retrieved Sequences in Real Workflows

Retrieval is step one of recognizable end-to-end pipelines, and seeing where it sits helps you scope it correctly.

In immunoinformatics, retrieved antigen sequences feed epitope prediction servers, and the highest-scoring predicted epitopes are assembled into multi-epitope vaccine constructs with linkers and adjuvants, followed by structure prediction and validation [10]. The retrieval step determines which antigens ever get considered, so casting a wide, well-filtered UniProt query at the start pays off across the entire design.

In toxin and bioactive peptide discovery, UniProt serves as the source of labeled training and validation data. Peptide toxin function prediction models were trained on sequences retrieved from UniProt and validated against external datasets of three-finger toxins and hemotoxic peptides [11]. Reviews of food-derived bioactive peptide research similarly describe workflows in which in silico identification and characterization of peptides from protein sequences, retrieved from sequence databases, precede prediction, validation, and optimization [16]. In each case, the retrieved sequences are the substrate for everything that follows, which is why disciplined retrieval with version tracking is worth the small overhead.

In peptidomics, endogenous peptides identified by mass spectrometry are matched back to parent proteins. A peptidomic study of rubber tree latex serum peptides identified forty-three peptide sequences by tandem mass spectrometry, and thirty-four of them matched proteins associated with plant defense and stress response, which then guided functional testing against Phytophthora pathogens [17]. The matching step is precisely the peptide-versus-database problem covered in Method 2, and choosing the right database scope, the latex proteome rather than all of UniProtKB, is what makes those matches interpretable.

Natural product researchers take a related approach at the genetic level, predicting nonribosomal peptide biosynthetic pathways directly from DNA sequence data in order to enumerate the peptides a microbe can produce before any of them are isolated [18]. Retrieving the corresponding protein and domain sequences from UniProt and related resources supports that annotation work.

## Frequently Asked Questions

### How do I retrieve a peptide sequence from UniProt step by step?

Search the accession or protein name at uniprot.org, open the reviewed entry, find the peptide region annotation such as signal peptide or mature peptide in the Sequence section, note the position range, then download the FASTA and slice the range. For scripted work, `curl https://rest.uniprot.org/uniprotkb/P01308.fasta` returns the same FASTA in one line.

### How do I find which proteins contain a short peptide I have?

Use the UniProt Peptide Search tool, paste your peptide in one-letter code, restrict the taxonomy to your organism of interest, and run the search. Results are grouped by organism and show the matched region within each protein, with an option to treat leucine and isoleucine as equivalent [4].

### What is the difference between reviewed and unreviewed UniProt entries?

Reviewed entries belong to Swiss-Prot and are manually curated by experts with detailed functional annotation, while unreviewed entries belong to TrEMBL and carry computational annotation. Prefer reviewed entries when retrieving sequences for analysis or publication [1].

### Can I download all protein sequences for an organism at once?

Yes. Find the organism's proteome on the UniProt website, note its proteome identifier such as UP000005640 for human, and download with the streaming endpoint using `query=proteome:UP000005640&format=fasta`. Use the compressed option for large proteomes.

### How do I get only the mature peptide rather than the full precursor?

Read the peptide annotation coordinates on the entry page or in the TSV output, then slice the FASTA sequence at those positions using 1-based numbering. Confirm the maturation boundaries with a cleavage-site prediction tool when the peptide is secreted or processed [9].

### Why does my peptide search return proteins even when leucine and isoleucine differ?

Because leucine and isoleucine are isobaric and mass spectrometry cannot distinguish them, the UniProt Peptide Match service can treat them as equivalent, and this option is on by default in some search modes. Set the option explicitly in your searches so the L/I policy is deliberate and documented [2].

### How do I map old database identifiers to current UniProt accessions?

Use the Retrieve/ID Mapping tool, choose the source database your identifiers come from, and map them to UniProtKB. Doing this early protects your analysis, since identifiers in some databases are deleted at a substantial rate while UniProtKB entries remain active or are formally replaced [9].

### Is UniProt data free to use in publications and scripts?

Yes. UniProt resources are freely available under a CC BY 4.0 license at uniprot.org, with no account or API key required for retrieval, and the knowledgebase is updated on a regular release cycle that you should record in your methods [5].

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How do I retrieve a peptide sequence from UniProt step by step?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Search the accession or protein name at uniprot.org, open the reviewed entry, find the peptide region annotation in the Sequence section, note the position range, then download the FASTA and slice the range. For scripted work, a single REST call returns the same FASTA in one line."
      }
    },
    {
      "@type": "Question",
      "name": "How do I find which proteins contain a short peptide I have?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Use the UniProt Peptide Search tool, paste your peptide in one-letter code, restrict the taxonomy to your organism of interest, and run the search. Results are grouped by organism and show the matched region within each protein, with an option to treat leucine and isoleucine as equivalent."
      }
    },
    {
      "@type": "Question",
      "name": "What is the difference between reviewed and unreviewed UniProt entries?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Reviewed entries belong to Swiss-Prot and are manually curated with detailed functional annotation, while unreviewed entries belong to TrEMBL and carry computational annotation. Prefer reviewed entries when retrieving sequences for analysis or publication."
      }
    },
    {
      "@type": "Question",
      "name": "Can I download all protein sequences for an organism at once?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. Find the organism's proteome on the UniProt website, note its proteome identifier such as UP000005640 for human, and download with the streaming endpoint using a proteome query and FASTA format. Use the compressed option for large proteomes."
      }
    },
    {
      "@type": "Question",
      "name": "How do I get only the mature peptide rather than the full precursor?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Read the peptide annotation coordinates on the entry page or in the TSV output, then slice the FASTA sequence at those positions using 1-based numbering. Confirm the maturation boundaries with a cleavage-site prediction tool when the peptide is secreted or processed."
      }
    },
    {
      "@type": "Question",
      "name": "Why does my peptide search return proteins even when leucine and isoleucine differ?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Leucine and isoleucine are isobaric and mass spectrometry cannot distinguish them, so the UniProt Peptide Match service can treat them as equivalent. Set the option explicitly in your searches so the policy is deliberate and documented."
      }
    },
    {
      "@type": "Question",
      "name": "How do I map old database identifiers to current UniProt accessions?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Use the Retrieve/ID Mapping tool, choose the source database your identifiers come from, and map them to UniProtKB. Doing this early protects your analysis, since identifiers in some databases are deleted at a substantial rate while UniProtKB entries remain active or are formally replaced."
      }
    },
    {
      "@type": "Question",
      "name": "Is UniProt data free to use in publications and scripts?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. UniProt resources are freely available under a CC BY 4.0 license at uniprot.org, with no account or API key required for retrieval, and the knowledgebase is updated on a regular release cycle that you should record in your methods."
      }
    }
  ]
}
</script>

## Sources

1. [UniProt: the universal protein knowledgebase in 2021](https://pubmed.ncbi.nlm.nih.gov/33237286/)
2. [A fast Peptide Match service for UniProt Knowledgebase.](https://pubmed.ncbi.nlm.nih.gov/23958731/)
3. [UniProt Tools: BLAST, Align, Peptide Search, and ID Mapping.](https://pubmed.ncbi.nlm.nih.gov/36943033/)
4. [PEPMatch: a tool to identify short peptide sequence matches in large sets of proteins](https://pubmed.ncbi.nlm.nih.gov/38110863/)
5. [Published and perished? The influence of the searched protein database on the long-term storage of proteomics data.](https://pubmed.ncbi.nlm.nih.gov/21700957/)
6. [A two-step database search method improves sensitivity in peptide sequence matches for metaproteomics and proteogenomics studies](https://pubmed.ncbi.nlm.nih.gov/23412978/)
7. [UniprotR: Retrieving and visualizing protein sequence and functional information from Universal Protein Resource (UniProt knowledgebase).](https://pubmed.ncbi.nlm.nih.gov/31843688/)
8. [PDBrenum: A webserver and program providing Protein Data Bank files renumbered according to their UniProt sequences](https://pubmed.ncbi.nlm.nih.gov/34228733/)
9. [Analyzing Signal Peptides for Secretory Production of Recombinant Diagnostic Antigen B8/1 from Echinococcus granulosus: An In silico Approach.](https://pubmed.ncbi.nlm.nih.gov/32582787/)
10. [In silico designing of a novel polyvalent multi-subunit peptide vaccine leveraging cross-immunity against human visceral and cutaneous leishmaniasis: an immunoinformatics-based approach.](https://pubmed.ncbi.nlm.nih.gov/36928431/)
11. [Function Prediction of Peptide Toxins with Sequence-Based Multi-Tasking PU Learning Method.](https://pubmed.ncbi.nlm.nih.gov/36422985/)
12. [PepBank--a database of peptides based on sequence text mining and public peptide data sources.](https://pubmed.ncbi.nlm.nih.gov/17678535/)
13. [Batroxin I: A Novel Bradykinin-Potentiating Peptide with Cytotoxic Activity Isolated from Bothrops atrox Snake Venom](https://doi.org/10.1007/s10989-023-10493-7)
14. [ProPepBA: a binding affinity database for canonical and chemically modified protein-peptide interactions.](https://pubmed.ncbi.nlm.nih.gov/42952447/)
15. [Synthesis of Bacterial Urease Flap Region Peptide Equivalents and Detection of Rheumatoid Arthritis Antibodies Using Two Methods](https://doi.org/10.1007/s10989-019-09815-5)
16. [Advancement and prospects of bioinformatics analysis for studying bioactive peptides from food-derived protein: Sequence, structure, and functions](https://doi.org/10.1016/J.TRAC.2018.04.005)
17. [Peptidomic Analysis and Antimicrobial Activity of Serum Peptide from Hevea brasiliensis clone BPM24.](https://pubmed.ncbi.nlm.nih.gov/37005518/)
18. [Chapter 8. Methods for in silico prediction of microbial polyketide and nonribosomal peptide biosynthetic pathways from DNA sequence data.](https://pubmed.ncbi.nlm.nih.gov/19374984/)