Ncbi Protein

NCBI Protein is the National Center for Biotechnology Information’s central database for protein sequence records, integrating data from RefSeq, GenBank translations, SwissProt, and other sources. This guide is intended for bench scientists, bioinformatics students, and anyone who needs to retrieve, interpret, or evaluate protein sequences for their research. The NCBI Bookshelf provides authoritative background on the database’s structure and curation policies.

The practical framework below will help you search effectively, choose the right entry type, avoid common pitfalls, and understand the limits of what the database can teach you. The EMBL EBI Training materials cover complementary approaches for analyzing protein sequences and are a recommended companion.

At a Glance

Core Concepts

A protein record in NCBI Protein includes the sequence itself, a definition line, organism source, gene name, and links to coding regions in nucleotide databases. Each entry is assigned a stable accession number and a version number (e.g., NP_000123.2). The Galaxy Training Network offers tutorials on how to parse these records programmatically.

Key record types

• RefSeq proteins (NP_, XP_, WP_): Curated or automatically annotated sequences derived from RefSeq genomes. NP_ entries are manually reviewed, XP_ are predicted.
• SwissProt (typically with accession format Oxxxxx, Pxxxxx, etc.): Expertly reviewed by UniProt. These records carry rich functional annotation.
• GenPept translations: Derived from GenBank nucleotide records. No curation beyond the submitter’s annotation.

Accessing records

You can retrieve a single protein by its accession number directly in a browser, e.g., https://www.ncbi.nlm.nih.gov/protein/NP_000123. Use the “Send to” menu to download sequences in FASTA or other formats. The Bioconductor project’s Bioconductor packages (like rentrez) allow batch retrieval from R.

Decision Points

When choosing a protein record, ask:

1. Do I need a reviewed or unreviewed entry?
   Reviewed (SwissProt or RefSeq NP) are more reliable for functional annotation. Unreviewed (XP_, GenPept) may contain errors.

2. Do I need the canonical sequence or an isoform?
   Many genes produce multiple splice variants. NCBI Protein lists isoforms with distinct accession numbers. Specify the isoform in your search.

3. Which taxon?
   The same protein in different species will have different records. Use the organism filter (txid) in Entrez.

4. Do I need structural data?
   Cross references to the Molecular Modeling Database (MMDB) or PDB indicate that an experimental structure is available. If not, computational models may be the only option.

These decision criteria are summarized in tutorials from the EMBL EBI Training on protein data resources.

Practical Workflow

Follow this step by step procedure to retrieve and evaluate a protein sequence.

Step 1: Formulate your query

Use the Entrez search bar at the NCBI Protein homepage. Combine terms with Boolean operators. Example:
("ERBB2" AND "human" AND "reviewed")
This yields high quality entries like those described in the clinical literature on trastuzumab therapy Trastuzumab Therapy and ERBB2 Genotype.

Step 2: Filter results

Refine by source database, organism, sequence length, or molecular weight. Apply the “Reviewed” filter to limit to SwissProt or RefSeq NP records.

Step 3: Select an entry

Click an accession to view the full record. Check the “Source” line for the organism and the “Gene” line for the gene symbol. Read the “Comment” or “Function” sections if present.

Step 4: Download the sequence

Click “Send to” and choose “FASTA” for most bioinformatics tools. For multiple sequences, use the “Clipboard” feature first.

Step 5: Verify the record

• Compare the definition line with known literature.
• Look at the “PubMed” links: do they cite the expected study?
• For unreviewed entries, run a BLAST search to ensure the sequence matches known homologs.

The Galaxy Training Network provides a complete workflow for using BLAST against the NCBI Protein database.

Step 6: Integrate with other data

If you will analyze the protein further, consider using tools from Bioconductor for domain annotation, alignment, or phylogenetics. For raw sequencing origin, consult the NCBI Sequence Read Archive.

Common Mistakes

1. Ignoring the curation status
   Using an unreviewed XP_ entry for a functional assay when a reviewed NP_ entry exists can lead to wrong conclusions.

2. Confusing protein name with gene symbol
   A gene symbol (e.g., ERBB2) is distinct from the protein name (e.g., receptor tyrosine kinase erbB 2). Search terms should be precise.

3. Overlooking isoforms
   Searching only by gene name may return the canonical transcript. Alternative splicing produces different protein isoforms that may have different functions.

4. Assuming experimental validation
   Many sequences in NCBI Protein are predicted from genome annotations. Check cross references for evidence from mass spectrometry or crystallography.

5. Misinterpreting accession version changes
   If you download a FASTA file and later reference the same accession, the version number (the digit after the decimal) may have changed, indicating a sequence update. Always cite both accession and version.

For more on avoiding these errors, the EMBL EBI Training highlights best practices in sequence retrieval.

Limits and Uncertainty

No protein database is complete or error free. NCBI Protein has several inherent limitations:

• Incomplete annotation: Many entries lack functional descriptions, especially for predicted proteins from environmental samples. The literature mining pipeline described in SigMine and OPathDb shows how automated approaches can fill some gaps, but they are not perfect.
• Missing structural data: Most proteins have no experimentally determined structure. Computational models exist, but their accuracy varies. For a rare example where structural evidence was obtained, see Structural evidence of thioacetate reduction by fluorescent protein mRuby.
• Taxonomic bias: Model organisms (human, mouse, E. coli) are well represented, but sequences from less studied species may be sparse or contain errors.
• Redundancy: The same protein may appear under multiple accessions from different source databases. Choose the most curated entry.
• Version instability: Accessions can be withdrawn or merged, so always record the exact version number and retrieval date.
• Predicted vs. experimental: A large fraction of proteins come from computational prediction pipelines. Without experimental validation, the sequence or its features may be incorrect. Comparative genomics approaches, like those used in Adaptations in electron transport chain complexes of clinical Rhodococcus equi, can provide indirect validation but not proof.

Keep these limits in mind when designing experiments or drawing conclusions from database searches.

Frequently Asked Questions

1. How do I download multiple protein sequences at once?
   Use the NCBI “Clipboard” feature: search for proteins, check the boxes for records you want (up to 500), then click “Send to” and choose “FASTA” and “Clipboard”. Alternatively, use the Entrez Programming Utilities (E Utils).

2. What is the difference between a RefSeq NP_ entry and a SwissProt entry in NCBI Protein?
   NP_ entries are part of NCBI’s own curated reference set, they often include links to genomic coordinates. SwissProt entries are reviewed by UniProt and usually contain more functional annotation. Both are reliable, but SwissProt often has deeper literature integration.

3. Can I use NCBI Protein to find proteins related to human disease?
   Yes, many disease associated genes are represented. Cross references to PubMed and OMIM (Online Mendelian Inheritance in Man) help connect sequence data to clinical phenotypes. For example, entries for ERBB2 are used in studies on pertuzumab therapy Pertuzumab Therapy and ERBB2 Genotype.

4. How do I know if a protein entry has been experimentally validated?
   Look for the presence of cross references to the Protein Data Bank (PDB), to mass spectrometry studies in the PRIDE database, or to PubMed identifiers that describe experimental characterization. Entries with a “REVIEWED” label are preferred, but even they may rely on indirect evidence.

References and Further Reading

• NCBI Bookshelf: The Protein Database , detailed chapters on database organization and content.
• EMBL EBI Training: Protein sequence resources , practical online courses for retrieving and analyzing protein data.
• Galaxy Training Network: NCBI BLAST tutorial , step by step workflow for sequence similarity searches.
• Bioconductor: rentrez package vignette , programmatic access to NCBI databases from R.
• NCBI Sequence Read Archive: Understanding protein origins , how raw reads lead to predicted protein sequences.
• SigMine and OPathDb: a literature mining pipeline , example of automated protein function prediction.
• Structural evidence of thioacetate reduction by mRuby , case study combining protein structure with sequence data.
• Sympotthastia annularis sp. nov. from Oriental China , example of protein sequences used in taxonomic keys.
• Pertuzumab Therapy and ERBB2 Genotype , clinical application of protein database entries.
• Adaptations in electron transport chain complexes of Rhodococcus equi , comparative genomics using NCBI Protein records.

Related Articles

• Protein Synthesis
• Incomplete Dominance Gene
• Cell Membrane Function Biology
• Dna Structure
• Protein Structure