01Identify the database before interpreting the code

An unexplained string of letters and digits is not enough to identify the kind of record. Look for the database name or linked page. Sequence records, literature identifiers, catalogue numbers and chemical registry numbers serve different purposes, even when a table places them beside one another.

UniProt's user manual describes accessions as stable identifiers for its entries and recommends citing the primary accession. Entry names are a separate field and can change. This distinction helps a reader find the intended record despite changes in descriptive naming.UniProt Knowledgebase User Manual (opens in a new tab)

02A stable accession does not freeze every annotation

UniProt separately tracks sequence and entry versions. A sequence-version change concerns the sequence data; an entry-version change can reflect other updated information. The manual also explains how merged or split records can retain secondary accessions.UniProt Knowledgebase User Manual (opens in a new tab)

For a reproducible note, save the accession, the version information relevant to your task and the date you accessed the record. If your conclusion depends on the sequence, preserve the sequence you used. A later reader can then distinguish a transcription error from a subsequent database update.

03Locate the peptide within the record

If the accession identifies a precursor or parent protein, state the relationship of the peptide to it. Include the relevant coordinates and written sequence, together with documented modifications. The parent accession alone does not distinguish every possible fragment derived from that sequence.

Suppose a project note selects residues 11–19 from a parent record. The accession tells another reader where to look; the coordinates identify the region; the sequence checks the extraction. These three fields answer different questions and work best together.

If a source uses mature-chain numbering while another uses precursor numbering, reconcile the reference before comparing coordinates. Do not silently align numbers that refer to different starting points. The fragment guide covers that boundary problem in more detail.

04Read annotation evidence separately

UniProt distinguishes manually reviewed records from computationally annotated records awaiting full review. That status concerns the database's annotation process. It does not establish that an external supplier's physical sample matches the record.UniProt Knowledgebase User Manual (opens in a new tab)

Likewise, an annotation can point to published evidence without making every sentence in the record a new experiment. Follow the relevant evidence references when a particular functional claim matters. A database is an organised route into the evidence, not a reason to omit its context.

Keep your record lookup and sample assessment in separate notes. You may be confident about the intended reference sequence while still needing analytical evidence for a delivered preparation. Combining the two into “identity verified” would conceal that difference.

05Create a useful accession citation

• ✓Name the database and primary accession.
• ✓Record the organism and relevant molecular description.
• ✓Preserve sequence or entry version information needed for the question.
• ✓State fragment coordinates and modifications separately.
• ✓Keep the access date and link to the record.

This small amount of context makes a database reference far more useful than a code copied into a specification without explanation. It lets the next reader find the record and understand precisely which part of it your research used.