Binding Site Detection (bioSkills)

A Claude Code skill that detects putative ligand-binding pockets de novo on a structure with no bound ligand, and ranks them by druggability or ligandability score.


Type	Claude Skill
Supplier	GPTomics bioSkills (community OSS, MIT)
Availability	GA — part of the bioSkills collection
Pricing	Free / OSS (MIT) — fpocket, P2Rank and mdpocket are separately installed OSS; CASTp and DoGSiteScorer are free academic web services
Capabilities	Read/Write — Claude runs the skill’s workflow locally (Bash/Python) and can call the two web servers, not as an MCP tool
Verified	works · 2026-08-03
Security	cleared · 2026-08-03 — GPTomics/bioSkills MIT confirmed, provenance matches, no advisories

How to install

bioSkills is not an npm package — skills are plain markdown/code read directly by the agent. Clone the repo, then either run the installer for the whole category or copy the single skill directory.

Claude Code — clone and install via the bundled script:
```
git clone https://github.com/GPTomics/bioSkills
cd bioSkills
./install-claude.sh --categories "structural-biology"
```
The installer copies matching skills into ~/.claude/skills/ (default target). Use ./install-claude.sh --list to preview the skills first.
Claude Code / other agents — copy just this one skill:
```
cp -r bioSkills/structural-biology/binding-site-detection ~/.claude/skills/
```
(run from inside your clone — the previous step left you in bioSkills/; otherwise replace bioSkills/ with the absolute path of your clone). Install fpocket and, if you want ML ranking, P2Rank when prompted on first use.

What it does

Enumerates surface concavities, then scores them — and is explicit that the two steps answer different questions:

Enumeration — geometric cavity detection with fpocket (Voronoi alpha-spheres) or CASTp (analytic surface topology, area and volume).
Ranking — machine-learned ligandability with P2Rank (surface-point clustering) or DoGSiteScorer (difference-of-Gaussians features plus an SVM druggability model, via the ProteinsPlus server).
Interpretation discipline — a geometric cavity is a hypothesis, not automatically a functional or druggable site: it may be a crystal-additive cleft or a non-functional groove. Druggability scores were trained on holo sets, so they systematically under-detect apo, shallow and cryptic pockets.
Cryptic and transient pockets — mdpocket tracks pocket occurrence and persistence across an MD trajectory or conformational ensemble, which is the route to sites that are closed in the deposited structure.
Predicted models — warns that pocket-lining side-chain rotamers are among the least reliable atoms in an AlphaFold/ESMFold model, so detection on a predicted structure needs extra scepticism.
Components — fpocket 4.1+ (primary), P2Rank 2.4+, CASTp and DoGSiteScorer (web), mdpocket, Biopython 1.83+, NumPy 1.26+.

Primary use cases: apo-structure pocket discovery for a new target, choosing a docking box, cryptic-pocket hunting over an MD ensemble.

Notes

Distributed as a SKILL.md (plus reference material) in the bioSkills collection — Claude executes the workflow locally rather than as an MCP server. The upstream skill front-matter name is bio-structural-biology-binding-site-detection; if invoked as a namespaced plugin command it resolves under the bioSkills plugin, not as a bare /binding-site-detection. tool_type upstream is mixed: fpocket, P2Rank and mdpocket run locally, while CASTp and DoGSiteScorer are submitted to external academic web servers — check your data-sharing policy before sending an unpublished structure to either. Sits upstream of the catalogued docking entries AutoDock Vina, smina and DiffDock, and downstream of Structure Preparation; the mdpocket route needs a trajectory from GROMACS MCP Server or OpenMM MCP Server. Upstream directory: structural-biology/binding-site-detection.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.