SEAD categorical component lineage¶
This account follows categorical source items into their exact staged parent observations. A linked category item, the component's nullable fields and the parent display text are separate source facts. The compact report keeps all three, together with their source class and type identities.
The governed acquisition is sead-full-evidence-39bfff6a-ce80714e. Its build is
sha256:ce80714e4c9e9974b24913e5da50f49854670ed642879c1ca5076499e1d56725.
The producer binds 35 source files: admission and acquisition manifests, the
materialization manifest and observation header, six raw relational tables,
and all 25 observation partitions. It checks every parent analysis observation;
it does not repeat provider acquisition or original materialization.
Source populations and interpretation¶
| Population | Rows |
|---|---|
| All staged observations scanned | 177,763 |
| All parent analysis values reconciled | 68,705 |
| Categorical components and distinct categorical parents | 4,800 |
Null component value and is_variant fields |
4,800 |
| Parent display equals the linked item name | 1,979 |
| Parent display differs from the linked item name | 2,821 |
| Categorical parents with analysis-entity chronology context only | 25 |
| Categorical parents without analysis-entity chronology | 4,775 |
| Eligible events admitted | 0 |
All linked items share the category type declared by their parent value class.
Null component fields do not erase that linked item identity. Conversely, the
item name does not replace the parent display string. For example, source item
-lw has parent displays -LW, -lw and -lw?; their distinction survives.
The producer records 15 exact class/item/value/variant/display groups.
The source classes describe 4,695 waney-edge components, 80 earlywood/latewood
components and 25 components concerning SNP capture, damage treatment, library
preparation and molecular sex. Identical item names such as NO can occur under
different source types. These counts describe analysis components; they are
not counts of physical specimens, libraries, sequencing assays or people.
No category is translated into an ecological presence/absence observation.
No unit is exposed by the linked source type for these 4,800 components. The
25 chronology-context links provide neither an eligible chronology claim nor
an ecological event. All 68,705 parent observations preserve refused event
eligibility and source_classification_not_accepted.
Reproduce¶
From the repository root, with an absent artifact destination:
PYTHONDONTWRITEBYTECODE=1 uv run --frozen python -m \
bijux_pollenomics_dev.source_completion.sead.categorical_values \
--repository "$PWD" \
--output "$PWD/artifacts/sead-categorical-lineage/report.json"
Compare the output bytes with the governed report. The producer refuses changed source hashes, missing or repeated parent keys, altered categorical components, item/class type mismatches, changed units or chronology support, promoted event eligibility, unsafe destinations and existing outputs. It checks all consumed source hashes again before publication; it does not lock external source files against concurrent writes. Source continuity supplies no new rights grant, biological interpretation, specimen identity or independent scientific decision.