MSI Detection (bioSkills)

A Claude Code skill that calls microsatellite instability status from sequencing data — paired, tumour-only or cfDNA — with the panel-specific thresholds each caller actually requires.


Type	Claude Skill
Supplier	GPTomics bioSkills (community OSS, MIT)
Availability	GA — part of the bioSkills collection
Pricing	Free / OSS (MIT). MSIsensor, MSIsensor-pro, MSIsensor-ct, MANTIS and mSINGS are separately installed command-line tools under their own licences
Capabilities	Read/Write — Claude drives the CLI tools locally on your BAMs; it is not an MCP tool
Verified	works · 2026-08-17
Security	caution · 2026-08-17 — GPTomics/bioSkills is now archived upstream; MIT and skill dir confirmed unchanged

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 "clinical-databases"
```
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/clinical-databases/msi-detection ~/.claude/skills/
```
(run from the directory holding your clone — if you are still in bioSkills/ from the previous step, use cp -r clinical-databases/msi-detection ~/.claude/skills/, or replace bioSkills/ with the absolute path of your clone). Install the primary caller:
```
conda install -c bioconda msisensor-pro
```
(bioconda is the documented distribution route for MSIsensor-pro; MANTIS and mSINGS install from their own repositories.)

What it does

Chooses the caller that matches the data you actually have, then applies its calibrated cutoff:

MSIsensor — paired tumour–normal, the reference case.
MSIsensor-pro — tumour-only, using a population baseline built from a cohort of normals.
MSIsensor-ct — cfDNA and liquid biopsy, panel-aware.
MANTIS — step-wise difference algorithm (> 0.4 instability threshold).
mSINGS — background-panel approach for targeted panels.

Reported thresholds and gates:

Parameter	Value
MSI-H	≥ 20–30% unstable loci (panel-calibrated)
Bethesda 5-locus panel	≥ 2/5 unstable = 40%
Minimum informative loci	≥ 50 for NGS-derived site sets
Tumour purity	≥ 20%
ctDNA fraction (MSIsensor-ct)	≥ 3%
MANTIS step-wise difference	> 0.4

Output is the raw unstable-locus count, the percentage unstable, and an MSI-H / MSI-L / MSS classification, with concordance checks against MMR immunohistochemistry.

Primary use cases: checkpoint-inhibitor eligibility (FDA pembrolizumab pan-tumour MSI-H, 2017; KEYNOTE-177), Lynch syndrome screening, dMMR confirmation, liquid-biopsy MSI.

Notes

Research use, not a diagnostic result — clinical MSI status comes from a validated assay plus MMR IHC, and the skill’s own workflow treats IHC concordance as a required cross-check rather than an optional one.

Two failure modes it is explicit about: a tumour-only call without a properly matched population baseline is unreliable, and low tumour purity or low ctDNA fraction produces false-MSS results rather than a flagged failure. Distinguishing MSI-H from a POLE-exonuclease hypermutator matters clinically because the mutational signatures overlap.

Read alongside Tumor Mutational Burden — MSI-H and TMB-H are correlated but separately reported ICI biomarkers (Sha 2020, Salem 2018) — and Somatic Signatures for MMR-deficiency versus POLE etiology.

Distributed as a SKILL.md (plus reference material) in the bioSkills collection. Upstream skill front-matter name is bio-clinical-databases-msi-detection; upstream directory clinical-databases/msi-detection. The skill is description-activated — there is no bare /msi-detection slash command.

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.