Lipoate Biosynthetic Process — Obsoletion & Merge into Protein Lipoylation

Bottom line: Lipoate is not made as a free pool: the octanoyl group is
attached to a lipoyl domain first and then sulfurated in place, so GO's
lipoate biosynthetic process (GO:0009107) and protein lipoylation
(GO:0009249) described the same reactions and were used inconsistently. GO
merged the first into the second on 2026-08-22, broadening the definition of
GO:0009249 to cover both assembly and attachment. We recorded the 12
experimental annotations and 11 InterPro2GO/UniRule mappings the merge
touches, then migrated the five reviews in this repo that carried GO:0009107
(POPTR LIP1, LIP1P-1, LIP1P-2; BACSU lipA; PSEPK lipA). All nine affected
existing_annotations rows are now MODIFY → GO:0009249, and none of the four
core_functions blocks that listed GO:0009107 still does (PR #2784). What
remains is an upstream comment on the orphaned GO:0016992 part_of GO:0009107
edge, a GOA re-fetch once GOA catches up, and optional reviews of the
B. subtilis GcvH-relay genes (lipM, lipL, gcvH). The Priority paragraph
below predates the merge landing.

Overview

A GO obsoletion proposal will obsolete GO:0009107 lipoate biosynthetic process
and merge it into GO:0009249 protein lipoylation, with the latter's
definition broadened to cover both cofactor assembly and its attachment to
lipoyl-carrier proteins.

The upstream rationale is that the two terms are used inconsistently: enzymes
that perform the biosynthetic chemistry (octanoyl transfer, sulfur insertion)
are frequently annotated to protein lipoylation because the experimental
readout in the primary literature is almost always the lipoylation status of
carrier proteins, not free lipoate. Lipoate is essentially never made as a
free pool — the octanoyl group is installed on the lipoyl domain first and
sulfurated in situ — so "biosynthesis" and "protein lipoylation" describe the
same set of reactions from two angles. Consolidating them follows the precedent
set by protein glycosylation.

This project tracks the impact of that merge on AI Gene Review. Unlike most
obsoletion projects in this repo, several affected genes are already reviewed
here (see Impact on this repo), so this is a concrete
re-review queue rather than a pure documentation exercise.

Upstream tickets

• Annotation tracker: geneontology/go-annotation#6505 (opened 2026-08-14)
• Ontology ticket: geneontology/go-ontology#32418 (opened 2026-08-06, open)
• Affected annotations spreadsheet: Google Sheet
• Impacted groups (per upstream issue): EcoliWiki 2, FlyBase 2, MGI 1 (done),
  MTBBASE 1, SGD 1, TIGR 1, UniProt 6

Obsoletion plan (per upstream)

Update 2026-08-30: the obsoletion has landed. Live QuickGO
(/ontology/go/terms/GO:0009107/complete, checked 2026-08-30 during the POPTR
re-review) returns isObsolete: true with replaced_by GO:0009249, the comment
"The reason for obsoletion is that the term usage has been inconsistent", and
ontology edits timestamped 2026-08-22. The paragraph below records the pre-obsoletion
state for history.

Term status verified via OLS on 2026-08-15 — both terms were then live:

• GO:0009107 (lipoate biosynthetic process) — "The chemical reactions and
  pathways resulting in the formation of lipoate, 1,2-dithiolane-3-pentanoate,
  the anion derived from lipoic acid." 12 experimental annotations;
  23,924 annotations total (QuickGO, exact-term, all evidence, 2026-08-15)
  — the electronic bulk comes from the InterPro2GO/UniRule mappings below.
• GO:0009249 (protein lipoylation) — current definition "The lipoylation of
  peptidyl-lysine to form peptidyl-N6-lipoyl-L-lysine." 76,753 annotations
  total (QuickGO, exact-term, 2026-08-15). Proposed new definition (edwong57,
  go-ontology#32418): "The chemical reactions and pathways resulting in the
  formation of lipoate ... and its attachment to lipoyl-carrier proteins."

Ontology-structure note

GO:0009107 currently has exactly one asserted child, and it is a part_of
link from a molecular function: GO:0016992 lipoate synthase activity
(QuickGO children endpoint, 2026-08-15). When the merge lands, that part_of
edge must be re-pointed at GO:0009249, or lipoate synthase loses its only
BP anchor. This is worth flagging on the ontology ticket — it is not mentioned
in either upstream issue.

A related caveat raised by Antonialock on go-ontology#32418: GCSH-type proteins
are not merely assembly scaffolds but are themselves lipoyl-dependent enzymes
whose lipoyl group acts as a swinging arm, so assembly and transfer are
genuinely intertwined. The merge is consistent with that view.

Affected experimental annotations

Verified via the QuickGO annotation API on 2026-08-15 (exact term, experimental
evidence codes only) — 12 annotations, matching the "12 EXP" count in
go-ontology#32418.

Notes on this list:

• The upstream group tally lists TIGR 1, but no TIGR-assigned annotation
  appears in the QuickGO snapshot above; conversely the human LIPT2 annotation
  (#1) is assigned by FlyBase. Both are bookkeeping discrepancies to confirm
  against the upstream spreadsheet rather than substantive issues.
• Five of the twelve are B. subtilis (lipA, gcvH, lipL ×2, lipM),
  reflecting the Cronan lab's dissection of the GcvH-relay route
  (PMID:21338420, PMID:21338421). That route is exactly the case the merge is
  meant to clarify: LipM octanoylates GcvH, LipL transfers the octanoyl
  group to other lipoyl domains, and LipA sulfurates — no free lipoate is ever made,
  so "biosynthetic process" is a misnomer for every step.
• The two EcoliWiki and one MGI annotation use acts_upstream_of_or_within,
  which will also want revisiting to involved_in when the terms are merged.

Mappings flagged for redirection

All five InterPro entries verified via the InterPro REST API on 2026-08-15
(names match upstream exactly; protein counts show why the electronic impact is
large):

UniRule mappings listed upstream (not independently verified here):
UR000080080, UR000112906, UR000159987, UR000375959, UR000376419,
UR000376665.

All eleven mappings redirect cleanly to GO:0009249 protein lipoylation — every
one of these families is a lipoate installation enzyme, which is precisely
what the broadened protein lipoylation definition will cover. There is no
case here (unlike the ent-kaurene obsoletion) where the mapping would be better
served by an MF term instead, because the corresponding MFs
(GO:0016992 lipoate synthase activity, octanoyltransferase activities) are
already separately mapped.

Impact on this repo

Five gene reviews carry GO:0009107 and ten carry GO:0009249. Because
existing_annotations[].term.id values are GOA-sourced and deliberately not
hard-validated (see CLAUDE.md), the obsoletion does not break validation there —
but core_functions term ids are strictly validated. Four reviews used
GO:0009107 inside core_functions.directly_involved_in; as of 2026-08-30,
zero do — the migration below was applied in the POPTR knowledge-base
re-review (PR #2784), exactly as prescribed by this tracker.

Reviews containing GO:0009107

The merge was a clean deletion in every one of the four core_functions
blocks: each already listed GO:0009249 protein lipoylation alongside
GO:0009107, so the fix was to drop the GO:0009107 entry rather than to
re-point it — applied 2026-08-30; GO:0009249 is now the sole
directly_involved_in BP in all four. Example of the pre-migration state
(genes/POPTR/LIP1/LIP1-ai-review.yaml):

  directly_involved_in:
  - id: GO:0009107          # <- removed 2026-08-30
    label: lipoate biosynthetic process
  - id: GO:0009249          # <- already present; now the sole BP
    label: protein lipoylation

The same paired-redundancy pattern holds in the GOA records themselves — every
in-repo gene with a GO:0009107 IEA also has a GO:0009249 IEA from the same
UniRule — which is direct evidence for the upstream claim that the two terms
are used interchangeably.

Reviews containing GO:0009249 (unaffected, but in scope for re-check)

POPTR/LIP1, POPTR/LIP1P-1, POPTR/LIP1P-2, BACSU/lipA, PSEPK/lipA,
PSEPK/lipB, PSEPK/gcvH1, PSEPK/gcvH2, METEA/gcvH, human/GCSH. These
gain scope (not lose it) when the definition broadens — human/GCSH already
argues in its review text that GO:0009249 is the better description of its
role, which the broadened definition makes unambiguously correct.

Related in-repo work

modules/endogenous_protein_lipoylation.yaml is grounded on GO:0009249 as its
source term and models the direct LipB–LipA route plus the Bacillus GcvH-relay
and human variants. The broadened definition strengthens that module's
framing; no change is required, but the module is the natural place to record
the merge. See also
PSEPK ppu00785 endogenous protein lipoylation batch.

Scope

• Organisms: bacteria-dominant (B. subtilis 5 EXP annotations, E. coli 2,
  M. tuberculosis 1), plus human, mouse, D. melanogaster, and
  S. cerevisiae. In-repo affected reviews are POPTR (3), BACSU (1), PSEPK (1).
• GO branch: BP only. No MF term is obsoleted — GO:0016992 lipoate synthase activity and the octanoyltransferase MFs are unaffected (though the
  part_of edge noted above must be re-pointed).
• Type of fix: structural / curation hygiene. The biology is
  uncontroversial and no annotation needs to be rebutted on biological grounds;
  the change removes a redundant BP framing.

Candidate genes for initial review

Listed in priority order.

1. BACSU/lipA (O32129) — highest priority. Already reviewed here and
   carries one of the twelve upstream experimental annotations (IGI,
   PMID:19820084). The review's core_functions block needs the redundant
   GO:0009107 entry dropped.
2. POPTR/LIP1, LIP1P-1, LIP1P-2 (B9H5L9 and paralogues) — already reviewed;
   mechanical core_functions edits plus a note on the two IBA/IEA
   existing_annotations. The IBA descends from MGI:1934604 (mouse Lias),
   which is upstream item #12, so these move together.
3. PSEPK/lipA (Q88DM5) — already reviewed; existing_annotations only, no
   core_functions change needed.
4. B. subtilis lipM (P54511), lipL (P39648), gcvH (O32174) — not yet in the
   repo. Four of the twelve upstream experimental annotations sit on these three
   proteins, and together they define the GcvH-relay route that motivates the
   merge. Reviewing them would give this repo the clearest worked example of why
   lipoate biosynthetic process was the wrong framing.
5. human LIPT2 (A6NK58) — not yet in the repo. Human octanoyltransferase;
   PMID:28757203 is a disease-gene paper, and human LIPT1/LIPT2/LIAS are a
   coherent trio for a future human lipoylation review set.
6. E. coli lipA (P60716) / lipB (P60720) — not yet in the repo. The
   canonical two-step pathway; both annotations use
   acts_upstream_of_or_within and would benefit from a relation review at the
   same time.

Proposed approach

1. ~~Wait for the merge to land before editing gene reviews.~~ Done —
   the obsoletion landed upstream on 2026-08-22 (verified via live QuickGO on
   2026-08-30), so the wait ended.
2. Comment upstream on the GO:0016992 part_of GO:0009107 edge, which is
   the one structural detail neither issue mentions. Still open — GO:0016992
   remains live and is still the core_functions MF in POPTR/LIP1 and
   LIP1P-1, unaffected by the BP obsoletion.
3. ~~When the merge lands, drop the redundant GO:0009107 entry from the
   four core_functions blocks and re-run just validate.~~ Done
   2026-08-30 (PR #2784): all four core_functions entries dropped, the
   existing_annotations rows switched to MODIFY → GO:0009249, and the
   cache/ontologies/go.tsv row refreshed to the obsolete state; all five
   reviews validate with 0 errors and 0 warnings. Remaining sub-step:
   re-fetch GOA for the five affected genes once GOA itself catches up with the
   merge, so existing_annotations picks up the replacement term.
4. Optionally extend coverage to the B. subtilis GcvH-relay trio
   (lipM, lipL, gcvH), which is the most instructive untouched cluster on the
   upstream list. Still open.

Priority

Medium. Higher than most obsoletion projects in this repo because five existing
reviews are directly affected and four contain author-supplied core_functions
ids that must change — but not urgent, since the ontology ticket is still open
and only MGI has marked its annotations done upstream.

Status

• 2026-08-30 — Obsoletion confirmed landed; in-repo migration applied
  (PR #2784, POPTR knowledge-base re-review). Live QuickGO returns
  isObsolete: true and replaced_by GO:0009249 for GO:0009107, with the
  comment "The reason for obsoletion is that the term usage has been
  inconsistent" and ontology edits timestamped 2026-08-22. All five affected
  reviews migrated: the redundant GO:0009107 entries dropped from the four
  core_functions blocks (POPTR/LIP1, LIP1P-1, LIP1P-2, BACSU/lipA), and every
  existing_annotations row carrying the term (including the BACSU/lipA IGI,
  upstream experimental item #2) switched to MODIFY → GO:0009249 with reasons
  recording the obsoletion; the cache/ontologies/go.tsv row refreshed to
  obsolete lipoate biosynthetic process / True. Remaining: the upstream
  GO:0016992 part_of edge comment (approach step 2), a GOA re-fetch for the
  five genes once GOA catches up, and the optional GcvH-relay trio reviews
  (step 4).
• 2026-08-15 — Project file created. Tracking go-annotation#6505 (opened
  2026-08-14) and go-ontology#32418 (opened 2026-08-06, open). Obsoletion not
  yet applied. GO:0009107 and GO:0009249 both confirmed live in OLS; the 12
  experimental annotations and the GO:0016992 part_of child confirmed via
  QuickGO; all five InterPro entries confirmed via the InterPro REST API.
  Upstream, only MGI has marked its annotation done.

Slides

• Slides (Marp source: LIPOATE_BIOSYNTHESIS_OBSOLETION-slides.md) — AI generated