DAPPER

LinkML models for dataset attribution, provenance, evidence retrieval, and related biomedical knowledge products. DAPPER is the working repository name; the acronym and public model branding are still being finalized.

Repository layout

The root schema/ directory is the model boundary. Keep schema modules, examples, converters, and model-specific documentation relative to that directory so the repository can grow without mixing model concerns with repository tooling.

schema/
  dapper.yaml               # current root model
  trusty-identifiers.md     # Trusty URI and nanopublication design notes
  identity/                 # computed content identifiers (DAPPER-ID-1)
  examples/                 # LinkML instance and graph examples
  converter/                # Source-data to model converters

Identifiers

Every object carries a computed content address, dapper:{ClassName}.{digest} — a GA4GH sha512t24u digest over the fields that constitute what the object is. Timestamps, signatures and mirror observations are excluded, so re-running a pipeline or re-signing a nanopublication does not change an identifier; changing the analysis does.

Identifiers are never written by hand. Leave id out of your source data and let the tool mint it. To turn a pile of gene-set runs into one identified collection:

uv run schema/identity/mint.py /path/to/your/genesets -o collection.yaml
uv run schema/identity/dapper_identity.py verify collection.yaml

schema/identity/README.md has the walkthrough, the full DAPPER-ID-1 profile, and why this is deliberately not a Trusty URI.

Additional model types should be added as separate YAML modules under schema/, with imports expressed relative to schema/. Shared vocabulary, base classes, and reusable enums should be factored into their own modules once there is a concrete second consumer.

Current migration

schema/dapper.yaml is the DAPPER model, migrated from the NIH Dataset Attribution and Provenance Profile. Its NIH-DAPP vocabulary is intentionally retained where it names the underlying standard. The schema already covers citation, funding, PROV lineage, file identity, controlled-access terms, workflow provenance, nanopublications, hypotheses, and agentic replay.

Validate from the repository root with:

uv run --with linkml linkml-validate \
  -s schema/dapper.yaml -C Dataset schema/examples/example_dataset.yaml

Contribution workflow

Make initial schema changes in this checkout, validate examples and converters, then commit and push to broadinstitute/dapper only after the migration shape and acronym are agreed.