Mevalonate Pathway Term Cleanup — Obsoletion & Replacement

IN_PROGRESS OBSOLETION

Species: human, rat, yeast

Genes: HMGCS1 HMGCR MVK PMVK MVD FDPS IDI1 Hmgcs2 ERG19

Mevalonate Pathway Term Cleanup — Obsoletion & Replacement

Bottom line: GO had two overlapping process terms for the route from
acetyl-CoA through mevalonate to isoprenoid precursors: GO:1902767 isoprenoid
biosynthetic process via mevalonate and GO:0010142 farnesyl diphosphate
biosynthetic process, mevalonate pathway. Both are now obsolete in OLS
(checked 2026-09-26), with annotations to be redirected to GO:0019287
isopentenyl diphosphate biosynthetic process, mevalonate pathway (the steps
up to IPP) or GO:0045337 farnesyl diphosphate biosynthetic process (IPP to
FPP). We tracked the upstream lists (2 and 21 experimental annotations) and
argued that each row needs a per-enzyme choice between the two replacements,
not a relabel. The repo has moved past the Impact and Candidate sections
below: the human mevalonate enzymes HMGCS1, HMGCR, MVK, PMVK, MVD, IDI1 and
FDPS are now reviewed (PRs #1998, #2153) with mevalonate_pathway and
isoprenoid_diphosphate_biosynthesis modules, and MVK, PMVK and MVD accept
GO:0019287 while FDPS accepts GO:0045337. Two reviews carried
obsolete-term rows, both resolved in #3232 (per-row outcome in Status
below): yeast ERG19's GO:0010142 RCA row is now REMOVE, since its
replacement GO:0019287 is already accepted, and rat Hmgcs2's GO:0010142 IBA
and IEA rows stay UNDECIDED, with the obsoletion recorded in reason.

Overview

A GO ontology cleanup proposes to obsolete two overlapping "mevalonate pathway"
terms in favour of better-scoped existing terms. The two terms describe the
same biology already covered by GO:0019287 isopentenyl diphosphate biosynthetic process, mevalonate pathway and/or GO:0045337 farnesyl diphosphate biosynthetic process, so existing annotations need to be redirected (with evidence review)
rather than carried on stale terms.

This project tracks both obsoletions as a single piece of work because they
share an upstream ontology ticket (geneontology/go-ontology#32082) and the
replacement target choice is the same for either source term.

Upstream tickets

Obsoletion plan (per upstream)

Obsoleted term ID Replacement(s)
isoprenoid biosynthetic process via mevalonate GO:1902767 consider GO:0019287 isopentenyl diphosphate biosynthetic process, mevalonate pathway or GO:0045337 farnesyl diphosphate biosynthetic process
farnesyl diphosphate biosynthetic process, mevalonate pathway GO:0010142 consider GO:0019287 isopentenyl diphosphate biosynthetic process, mevalonate pathway or GO:0045337 farnesyl diphosphate biosynthetic process

Verify all term IDs and labels in OLS before acting; the obsoletion is still
under discussion upstream at the time of this file.

The replacement is not a single 1:1 swap — GO:1902767 and GO:0010142 both
describe a multi-step pathway that runs through mevalonate to IPP and then on
to FPP. After obsoletion, each annotation needs to land on whichever of the
two replacement terms (or both) accurately describes the actual reaction(s)
that the experiment supports. That is a scientific judgement, not a mechanical
relabel.

Affected annotations — upstream counts

From the ontology issue (geneontology/go-ontology#32082):

Term Direct EXP annotations Genes listed in upstream
GO:1902767 (this issue) 2 erg9, yajO
GO:0010142 (companion #6439) 21 fps1, dps1, spo9, ERG12, ERG8, hcs1, Idi1, Hmgcr, Acat2, Hmgcs1, Fdps, Mvd, ACAT2, Pmvk, Mvk, hmgr

Per #6440's body, the GO:1902767 list is now down to "1 EcoCyc" annotation
(the PomBase entry was already fixed). The exact remaining EcoCyc gene/PMID
is in the linked spreadsheet, which is gated behind Google sign-in and
should be retrieved via QuickGO when work starts.

For GO:0010142 (companion #6439), the full set should be re-pulled from
QuickGO at review time — many of the 21 listed genes will have been
reassigned in the interim.

Impact on this repo

Scanning genes/**/*-goa.tsv for either GO ID returned a single hit:

None of the other listed genes (erg9, yajO, fps1, dps1, spo9, ERG12, ERG8,
hcs1, Idi1, Hmgcr, Fdps, Mvd, ACAT2, Pmvk, Mvk, hmgr, Hmgcs1, Acat2) have an
*-ai-review.yaml in this repo at the time of this writing.

Scope

Candidate genes for initial review

Listed in priority order. Each should be set up with
just fetch-gene <organism> <gene> before review begins.

  1. HMGCS1 (human, Q01581) — Cytosolic HMG-CoA synthase, canonical entry
    to the mevalonate pathway in the cholesterol/isoprenoid branch. A clean
    reference for what GO:0019287 should look like and a strong candidate
    for the GO:0019287 replacement for any current GO:0010142 annotation.
  2. MVK / MVD / PMVK (human) — The three "pure mevalonate-to-IPP"
    enzymes. Strong candidates for GO:0019287 replacement; useful as a
    batch because the same logic applies to all three.
  3. FDPS (human, P14324) — Farnesyl diphosphate synthase. Catalyses the
    IPP → GPP → FPP steps, so any GO:0010142 annotation here actually
    belongs on GO:0045337 farnesyl diphosphate biosynthetic process rather
    than the obsoleting term.
  4. HMGCR (human, P04035) — HMG-CoA reductase. The rate-limiting step
    of the mevalonate route; likely annotated to GO:0010142 but really
    covers only the upper (mevalonate-formation) half of the pathway. Worth
    a careful review because HMGCR is also heavily studied for
    statin/cholesterol biology and its non-pathway annotations should be
    left untouched.
  5. HMGCS2 (rat, P22791) — already in repo. Two existing IBA/IEA
    GO:0010142 annotations are already flagged for removal in
    Hmgcs2-notes.md because of paralog conflation with the cytosolic
    HMGCS1. The review file can be updated to record the obsoletion
    rationale in the review.reason for those two annotations.
  6. ERG9 / ERG12 / ERG8 (yeast) — Yeast mevalonate-pathway enzymes,
    upstream-listed for both terms. Lower priority because the yeast
    annotations are already curator-maintained.
  7. yajO (E. coli) — Listed under GO:1902767 in the ontology issue.
    The yajO assignment to a mevalonate pathway is dubious (E. coli uses
    the MEP/DXP pathway, not the mevalonate pathway). Worth a quick check
    to confirm whether this is the remaining "1 EcoCyc" annotation flagged
    in #6440.

Proposed approach

  1. The obsoletion has landed (GO:0010142 is obsolete in the GO release
    2026-07-26). GOA-sourced term ids are never rewritten; record the
    obsoletion in action/reason and, where a replacement applies,
    proposed_replacement_terms. Where the correct replacement is already
    annotated to the gene, prefer REMOVE over a duplicate MODIFY
    (as done for yeast ERG19 in PR #3232).
  2. Group the review by replacement target, not by the obsoleting term.
    Genes whose experimental evidence supports the upper half of the
    pathway (mevalonate → IPP) want GO:0019287; genes covering the lower
    half (IPP → FPP) want GO:0045337; some may want both. Trying to
    review "every annotation to GO:0010142" without that split conflates
    two different replacements.
  3. Reuse Hmgcs2 as the worked example. The existing
    genes/rat/Hmgcs2/Hmgcs2-ai-review.yaml already records the
    paralog-conflation logic that motivates the obsoletion. Updating its
    review.reason fields with explicit reference to this project is a
    useful first concrete step.
  4. Watch out for the bacterial annotation. E. coli does not use the
    mevalonate pathway, so a GO:1902767 annotation on yajO (or any other
    E. coli protein) is almost certainly wrong rather than just misnamed.
    This is a removal candidate, not a remap candidate.

Priority

Low–medium — the per-annotation review work here is small (the upstream
lists are short and several entries have already been fixed at source).
The main repo-level effect is updating the rat HMGCS2 review to cite the
obsoletion rationale; the rest is queueing work for when human
HMGCS1/MVK/MVD/PMVK/FDPS/HMGCR enter the review cycle.

Status

Slides