NMDC source-file selection
selection.json is the saved selection report for the original
14 NMDC GFF examples described in the acquisition history.
It records the smallest file found for each sampled type, with its URL, checksum,
workflow, and number of records sampled. It is not a complete DataObject census;
the separate DataObject analysis has that scope.
Run just nmdc-sample from the repository root for a new live sampling.
It writes local/nmdc-selection/selection.json and reports missing types or request
errors to stderr. Review the result before replacing this dated report or changing
the corpus index. A later sample may select different files and does not recreate
the original acquisition. Upstream source bytes remain under corpus/sources/.
NMDC's data-use terms apply to the source metadata; the indexed source files carry their individual provenance and license records in the corpus index.