Repository navigation
Normalize the curated Thiothrix lineage without altering primary-source evidence #1978
Description
Activity
Addressed in #1975 at
52e905def8367ffccf9179af7337e6f80710dd4d.Batch128 corrects the corpus hierarchy regression found in CI: only the active Thiothrix phylum name is normalized to Pseudomonadota. Original Proteobacteria source quotations, all caches, genus identity, nodes and arrows are unchanged. #1978 is addressed pending protected-main merge. #1979 is addressed by the verified archival tag
review-provenance/20261009-batch128, which retains the reviewed source through squash merge and branch cleanup.246 focused tests passed; 17 external-crosswalk-dependent cases skipped. The unchanged whole-corpus hierarchy test and four mutation probes pass. Schema, strict, terms, references, snippets, history, inventory, network and deterministic rendering checks passed within recorded warning/cache limits; postcommit docs-current passed. Full local suite did not complete: pytest-cov and UMAP were missing, then locked isolated dependency installation failed building llvmlite. These failures are retained, not counted as passes. Exact-head and merge-group CI remain required.
Before review | After review | Structured after YAML | Validation | Provenance proof
This is a scoped self-adversarial review, not independent approval or whole-record scientific certification. Prior batches126-127 remain partial identity repairs: #1841 and #1860 stay open, both records stay needs_research, and no new topology or Edison submission was needed. The root PR's existing protected queue request is separate from this non-native child. No topic-parent merge, bypass or branch deletion is claimed.
Keeping this issue open until the repair and verification receipt reach protected main.
Finding
P2 / major: PR #1975 at 57b17a7 fails the corpus hierarchy regression. Batch127 copied the study's historical p__Proteobacteria spelling into the active curated GTDB lineage, placing c__Gammaproteobacteria under a different parent path from the rest of the corpus. This is an integration defect in the new repair, not evidence that Thiothrix belongs to a different biological phylum.
CI: https://github.com/CultureBotAI/CommunityMech/actions/runs/37845443207 (1 failed, 4667 passed, 87 skipped, 8 deselected). The focused batch127 suite omitted tests/test_gtdb_lineage_tree.py.
Repair And Acceptance
Normalize only the active lineage's phylum name to Pseudomonadota with explicit provenance. Preserve the original Table S4 row, all three verbatim evidence snippets, cache bytes, genus1030 assignment, unknown GTDB release, no invented vote counts, existing graph and unresolved Beggiatoa/CPR boundaries. GTDB's R08-RS214 announcement documents the phylum-name update: https://forum.gtdb.ecogenomic.org/t/announcing-gtdb-r08-rs214/456 . The resource paper discusses Proteobacteria/Pseudomonadota: https://doi.org/10.1093/nar/gkaf1040 . This does not certify a new release-specific placement for the study genome.
Reproduce the failure, add focused control/mutation tests, run the full corpus hierarchy regression and relevant validation, append curation history and an immutable successor ledger, regenerate the page, and record an exact-head adversarial self-review. Keep #1841 open for its separate unresolved scientific scope. Do not weaken the hierarchy validator or rewrite historical ledgers/source quotes. Close this repair issue only after protected-main merge.
Deduplication: searched all 1206 open/closed issue titles and bodies; five matching bodies read. Scoped reports/history/tests/.github search included ignored and hidden files. No existing issue covers this new batch127 CI defect; #1841/#1838 track different scientific findings. Comments and historical PRs were not exhaustively searched.