Bundles#

A BDP bundle is a pre-assembled collection of related data sources for a specific use case. Instead of adding each source individually, you add one bundle and get a versioned, reproducible collection of everything that workflow needs.

Bundles use the same identifier format as individual data sources:

The difference is in the type: bundles have source_type = bundle internally. When you pull a bundle, BDP resolves all its constituent sources, pins their exact versions, and records them in bdp.lock.

Examples#

Rather than adding dozens of individual entries:

A bundle collapses this into a single line:

The bundle declares its own dependencies internally. BDP resolves them, version-pins them, and writes the full dependency tree into bdp.lock.

What Makes a Bundle#

A bundle is a data source where:

• The source_type is bundle
• Its members are declared as version-pinned dependencies
• The bundle itself has a BDP version (@1.0, @2.0, ...) that tracks the collection as a whole — independently of the versions of the individual sources it contains

This means a bundle version is a stable snapshot: bdp:homo-sapiens-proteome@1.0 will always resolve to the same set of protein sequences, regardless of when you pull it.

Bundle Versioning#

Bundles follow the same versioning rules as all data sources — explicit version required, no @latest. The version of a bundle reflects the state of the collection at a point in time, not the versions of its individual members (though those are pinned inside).

Available Bundles#

Bundles are currently being defined as the platform matures. The set of available bundles depends on which use cases the community needs most.

If you have a specific workflow in mind — a species proteome, a disease-specific ontology bundle, a variant + phenotype combination for a particular research area — open an issue or chat on Matrix. We build bundles based on demand.