bolerodata

Dataset, model-zoo and metadata registry for Bolero.

A uniform Python API over the single-cell ATAC/RNA datasets, model checkpoints, QTL collections and differential-accessibility records behind Bolero. It maps short keys to artifacts and returns them as pandas / AnnData objects and ready-to-use file paths. Large artifacts are downloaded on demand from HuggingFace and cached locally.

Documentation: https://liuhlab.github.io/bolerodata/

Installation

pip install bolerodata            # registry + auto-download of shared data
pip install 'bolerodata[diff]'    # + differential-analysis stack

import bolerodata
from bolerodata import DATASETS, MODELS, metadata, GTEx

metadata.MODEL_ZOO.head()                    # browse the model zoo (offline)
DATASETS["HumanBrainDev"].cell_metadata      # downloads from HuggingFace on first use
model = MODELS["<model_key>"]                # checkpoint downloads lazily

Building / running a model predictor, resolving a Genome, or using diff_analysis additionally needs the companion bolero package (GPU stack); bolerodata imports it lazily. See the installation guide.

Data is fetched from arcinstitute/bolero-models (checkpoints) and arcinstitute/bolero-data into $BOLERODATA_HOME (default: the platform cache dir). See Data setup & caching.

Developing

The repo uses pixi (lean — no GPU stack):

git clone https://github.com/liuhlab/bolerodata.git
cd bolerodata
pixi install            # runtime env; or: pixi install -e dev / -e diff / -e docs

Maintainers publishing data to HuggingFace: see huggingface_digest.md and the sharing guide.

License

MIT — see LICENSE.