Repository navigation
Enterococcus is grounded to a non-type GTDB clade: g__Enterococcus_B is the faecium clade #373
Description
Found reviewing PR #372. This is a live defect in the KB, and my PR framed it the wrong way round.
kb/communities/Electrostimulated_Mixotrophic_VFA_Producing_Enrichment_Consortium.yaml grounds NCBITaxon:1350 (Enterococcus) to GTDB:g__Enterococcus_B at 0.596.
GTDB's naming rule is that the lineage containing the nomenclatural type keeps the unsuffixed name; all others take alphabetic suffixes. The type species of Enterococcus is E. faecalis, and in the crosswalk faecalis (9,797 genomes) maps to unsuffixed g__Enterococcus while faecium (15,170) maps to g__Enterococcus_B.
So the majority-of-genomes rule picks _B purely because E. faecium is more sequenced — not because it is the right home for the NCBI genus. The correct grounding is g__Enterococcus, or no grounding at all.
Under the alternative deepest denominator this taxon resolves to AMBIGUOUS, which is the honest answer. PR #372's flip table labels that "loses grounding", i.e. it presents the correct outcome as the alternative's failure. That framing is wrong and is corrected in the PR body.
Fix is a curation call: set it to g__Enterococcus, or withhold and record why.
Activity
Address the review of #372: complete the analysis, and disclose where…
Fold in the named-species filter, and report what it actually does (#375
Address the #394 review: tests that skipped in CI and broke on merge
Stop grounding an exact 50/50 tie, and add a way to withdraw one (#382)…
Ground taxa at domain rank (#393) (#402)
Record what a GTDB sweep did, in the record rather than only in git (#…
realmarcin commentedon Oct 6, 2026
Reopening after the batch37 source-based graph review: on published batch36 head a691e1d, Electrostimulated_Mixotrophic_VFA_Producing_Enrichment_Consortium.yaml still grounds NCBITaxon:1350 Enterococcus to GTDB:g__Enterococcus_B (0.598, 17011/28462 genomes; mapping built 2026-07-25). The exact live grounding defect is already described here; no duplicate issue created. The graph-only changes will preserve the taxonomy block and explicitly avoid certifying it. Current primary study resolves only the genus and cannot justify the faecium clade from sequence counts.
realmarcin commentedon Oct 6, 2026
Graph-only corrections published in #1437. This separate follow-up remains OPEN; non-graph raw blocks preserved, not certified. No taxonomic grounding or reactor-protocol repair is claimed by this graph change.