Notebooks

Worked examples as marimo notebooks — plain Python files with @app.cell decorators, so they diff and review like source rather than like JSON, and they run three ways:

pip install 'vdjtools[examples]'
marimo edit examples/signature_features.py    # interactive
marimo run  examples/signature_features.py    # read-only app
python      examples/signature_features.py    # plain script; prints, no UI

Each one bootstraps its own data from Hugging Face (isalgo/airr_benchmark) and caches it, so a fresh pip install is enough — no local paths and no pre-staged files. Drop a copy under ./data_dump/ (gitignored) and it is used instead of downloading.

Signatures

Signature features

The vsig half block by block: what each column measures, how it is transformed, and which ones survive a shallow library.

Signature pipeline (mirpy)

A folder of AIRR TSVs to one table joined with your metadata, in one command --- both halves at once. The end-to-end recipe.

Scaling a cohort

Coverage-standardised diversity across samples of wildly different depth, and why the standardisation is what makes them comparable.

Analysis

Aging

78 donors aged 0-103, three ways: streamed cohort statistics, coverage-standardised iNEXT diversity, and repertoire divergence by MDS.

Pre-processing

The three filtering axes, and what each one does to a real cohort.

Overlap and similarity

Pairwise repertoire overlap, the metrics that behave under depth variation, and the ones that do not.

Model workshop

Build, check and compare V(D)J generation models; score sequences; read the training log.