Glycosylation site finder

How to find N-linked glycosylation sites

Paste a protein sequence or upload FASTA, text, CSV, or PDB data, then run the analysis. The glycosylation site finder lists every N-X-S/T triplet where X is not proline, with the asparagine position and nearby sequence. These are potential sites, not evidence that a glycan is attached.

The bundled example includes overlapping sequons and a proline-containing triplet:

>sequon-demo
NNSTNPSNVT

Both overlapping sites at positions 1 and 2 count. The NPS triplet at position 5 does not count because its middle residue is proline.

Several proteins in one run

Use FASTA format to keep separate proteins separate. A wrapped plain sequence is treated as one protein.

>first
NVT
>second
NPS

A zero count is a successful analysis. For example, the 110-residue human insulin precursor, UniProt accession P01308, contains no matching triplets:

>sp|P01308|INS_HUMAN insulin
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKT
RREAEDLQVGQVELGGGPGAGSLQPLALEGSLQKRGIVEQCCTSICSLYQLENYCN

Its result is P01308, 110 amino acids, 0 sites, density 0.00, and None found in both site columns.

Input

Standard amino-acid letters, U, O, and ambiguity codes B, Z, J, and X are retained. Ambiguous residues produce a warning because they can create uncertain matches or hide sites. Whitespace and invisible characters from pasted documents are normalized. Protein names belong in FASTA headers, not on unmarked sequence lines.

Alignment gaps (- or .), stop markers (*), punctuation, and embedded numbers are rejected. Removing them would change positions or create false triplets. GenBank, FASTQ, and document files must first be converted to a supported sequence format.

Settings

There are no adjustable analysis settings. Every run uses the N-X-S/T rule, scans all positions including overlapping triplets, and captures up to five residues before and after each triplet.

Results

The Results tab contains one spreadsheet row per protein, in input order. Use the spreadsheet's copy or download action to export the table.

Density counts matching asparagines per 100 sequence positions. It does not count all three residues of each triplet and is not the percentage of occupied glycosylation sites. There are no validated low, moderate, or high glycosylation categories in this tool.

What does a sequon tell us?

N-linked glycosylation attaches a glycan to an asparagine side chain. The common sequence motif is asparagine, any amino acid except proline, then serine or threonine. Bause's proline-peptide experiments establish the importance of the middle residue. A sequence scan can identify these triplets, but cannot determine whether a cell actually glycosylates them.

The tool checks the three-residue consensus independently. PROSITE PS00001 instead uses a four-residue pattern that also excludes proline immediately after the triplet. Therefore NVTP and a terminal NVT count here and can give different results from that PROSITE pattern.

Glycosylation depends on cellular localization, folding, enzyme access, and sequence context. Site density alone cannot establish glycosylation, protein stability, or glycan occupancy. Experimental evidence, such as glycopeptide mass spectrometry, is needed to confirm attached glycans.

Limitations

• Only the common N-X-S/T consensus is scanned. O-linked glycosylation, C-mannosylation, and uncommon N-linked motifs such as N-X-C are outside its scope.
• There is no probability score, glycan composition prediction, structural accessibility model, or organism-specific model.
• Ambiguous residues stay in their original positions. For example, NXS is reported as a possible match even though X could be proline. Unknown N/S/T positions can also conceal matches.
• PDB coordinates often omit residues. Numbering gaps are represented by X, but the full sequence and its numbering cannot always be recovered from coordinates. Use full-length FASTA for complete-protein site counts and positions.

Which tool should I use?

FAQ

Does a matching sequon prove N-glycosylation?

No. It identifies a sequence-compatible candidate. A protein can contain several sequons without any attached glycans.

Are the positions zero-based or one-based?

Positions are one-based and refer to the asparagine at the start of the triplet. For PDB input they are relative to the extracted chain sequence, including inserted X positions, rather than the original residue labels.

Why was an alignment or stop-marked sequence rejected?

Deleting a gap or stop marker can join residues that were not adjacent and create a false site. Submit an ungapped protein sequence without stop markers whose positions you can interpret consistently.