Glycoengineering

Analyses protein glycosylation: canonical N-X-S/T sequon scanning, S/T-rich regions, GlyTouCan lookup and preparing NetNGlyc, NetOGlyc and GlycoSHIELD runs, keeping motif, prediction and occupancy apart. Use when working on a glycoprotein.

Databases & SQLPack: K-Dense scientific skillslinkglycoengineering
Author
Source
K-Dense-AI/scientific-agent-skills/skills/glycoengineering @ ec59a29
License
Unknown in the skill, MIT for the repository

About this skill

What it does

The skill is about reading protein glycosylation evidence without mixing up its kinds. Its central rule: a sequence motif, a predictor score, experimentally supported occupancy and a glycan structure are four different results, and none establishes the others. The procedure around it: record the accession, isoform, exact sequence, expression host, topology and construct boundaries first, and keep a mapping from submitted coordinates to the mature protein, PDB residues and antibody numbering — a residue number in a FASTA file is not one in a structure. Motif scanning uses the canonical N-X-S/T pattern with X not proline and keeps overlapping hits; S/T density is a descriptive feature, not a probability; any planned edit is followed by a re-scan of the whole product, since motifs can be lost and created. The closing sections bound the evidence: a composition does not resolve linkages or branches, and relative signal between detected forms is not occupancy.

Two Python scripts ship with it. glycoengineering_tools.py normalises a sequence — rejecting FASTA headers, gaps and unknown residues instead of silently shifting coordinates — finds sequons, edits one site and lists S/T-rich windows, with 1-based positions and asserts in its examples. glytoucan_lookup.py reads a GlyTouCan accession's WURCS sequence from the public SPARQL endpoint, checking the accession's format before putting it into the query. references/glycan_databases.md covers GlyTouCan, GlyConnect (its REST API written up as unverified, with the failing endpoints named) and notation; references/glycoshield.md pins the GlycoSHIELD commit and documents a CLI that silently ignores several options, with the direct Python API as the workaround.

When to use it, and when not

Use it for candidate site analysis on a glycoprotein, for comparing antibody glycoforms, for structural shielding questions, or when reading site-specific glycoproteomics output.

It covers eukaryotic secretory-pathway N-glycosylation and mucin-type O-GalNAc only; other O-linked modifications need their own predictors. The DTU predictors are web forms, not an API, so the agent prepares and interprets a submission rather than automating it.

How to run it

We do not copy this skill: it declares its licence as Unknown, which does not match the repository's MIT. Install it from K-Dense's repository with the button to the source: gh skill install K-Dense-AI/scientific-agent-skills glycoengineering for one skill, npx skills add K-Dense-AI/scientific-agent-skills for the set, or the whole repository as an Agent Plugins package for Cursor and Codex, where the skill may show under the scientific-agent-skills prefix. The local helpers need only Python 3.10+; the GlyTouCan lookup needs requests and network access; GlycoSHIELD is a separate source install with its conformer libraries, plus GROMACS for the SASA step. No API keys anywhere.

  • Claude Code: /glycoengineering, or ask for the skill by name.
  • Codex: $glycoengineering.
  • Cursor: type / in the Agent chat and pick glycoengineering.

Example: /glycoengineering Scan this sequence for canonical N-glycosylation sequons, list the overlaps, and say which of them the topology puts outside the cell.

How it differs from similar skills

Nothing else in our catalog deals with glycobiology. Phylogenetics workflows also takes protein sequences, but aligns homologous ones to build a tree and does not look for glycosylation sites. The other two K-Dense pages work on trees and graphs: ETE Toolkit for phylogenetic trees and PrimeKG knowledge graph.

Install

This skill is not copied to the catalog. Install it from the author's repository.

Open the source

Connect through AnyModel API

This skill runs on whatever model your agent uses. Point Claude Code, Codex or Cursor at AnyModel to run it on Claude, GPT, Gemini and more with one key — pay per token, no subscription.

More from the pack K-Dense scientific skills