Tumor Mutational Burden (bioSkills)

A Claude Code skill that computes tumour mutational burden the way assay-harmonization work says it must be computed — per-assay calibration and explicit filtering — instead of dividing a raw variant count by a nominal panel size.

   
Type Claude Skill
Supplier GPTomics bioSkills (community OSS, MIT)
Availability GA — part of the bioSkills collection
Pricing Free / OSS (MIT). cyvcf2, pandas and numpy are separately installed; VEP/snpEff, LOHHLA and DASH carry their own licences
Capabilities Read/Write — Claude runs the skill’s Python workflow locally on your VCFs; it is not an MCP tool

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/tumor-mutational-burden ~/.claude/skills/

    (run from the directory holding your clone — if you are still in bioSkills/ from the previous step, use cp -r clinical-databases/tumor-mutational-burden ~/.claude/skills/, or replace bioSkills/ with the absolute path of your clone). Install the Python dependencies:

    pip install "cyvcf2>=0.30" "pandas>=2.2" "numpy>=1.26"

    Consequence annotation requires VEP ≥ 111 or snpEff ≥ 5.2 to have been run on the VCF beforehand.

What it does

  • Counts and normalizes — parses an annotated somatic VCF with cyvcf2, counts nonsynonymous variants (missense, frameshift, stop-gained, splice-site), and divides by the assay’s actual scored megabases rather than its nominal panel size.
  • Filters that change the answer — VAF floors (≥ 5% with a paired normal, ≥ 10% tumour-only), depth ≥ 100, and germline removal by gnomAD population frequency (AF ≤ 0.5%). Tumour-only calling without germline subtraction is the classic source of inflated TMB.
  • Cross-assay harmonization — Friends of Cancer Research equations and per-assay calibration, so the FDA’s 10 mut/Mb cutoff maps to ~7.8 on TSO500 and ~8.4 on Oncomine TML instead of being applied identically everywhere.
  • Tiering and context — hypermutator tiers (POLE/POLD1, MMR-deficient), tumour-type-specific cutoffs per McGrail 2021, and ESMO 2024 / FDA pan-tumour pembrolizumab (2020) reporting expectations.
  • Blood TMB — tissue vs bTMB comparison, with the ctDNA-fraction caveats that make low-shedding tumours unreliable.
  • Immunogenicity integration — pairs TMB with HLA loss-of-heterozygosity (LOHHLA, DASH) and neoantigen quality (Luksza 2017 fitness), since a high count with lost HLA presentation is not the same thing.

Primary use cases: checkpoint-inhibitor eligibility assessment, tissue-vs-blood TMB comparison, auditing a TMB-H call against the assay it came from.

Notes

Research use, not a diagnostic result. A TMB value that drives a treatment decision must come from a validated clinical assay with its own cutoff; this skill’s value is making explicit the filtering and calibration choices that a report usually hides.

The single most common error it guards against is comparing TMB across assays as if the number were assay-independent — panel size, scored region, synonymous-inclusion convention (FoundationOne includes them) and germline handling all shift the value by enough to cross the 10 mut/Mb line.

Read alongside MSI Detection — MSI-H and TMB-H are separate but correlated immunotherapy biomarkers, and POLE-exonuclease hypermutators can mimic one another — and Somatic Signatures for the mutational-process etiology behind a high count. Downstream neoantigen work is covered by Neoantigen Prediction and Immunotherapy Response Prediction.

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