Amino Acid Activation Terms — Obsoletion of 43 Substrate-Specific tRNA Aminoacylation BPs

SCOPING OBSOLETION

Species: human, PSEPK, POPTR, DANRE, METTP, DROME

Amino Acid Activation Terms — Obsoletion of 43 Substrate-Specific tRNA Aminoacylation BPs

Bottom line: GO is obsoleting 43 biological-process terms that name the
amino acid charged onto a tRNA (alanyl-tRNA aminoacylation and so on, plus
their mitochondrial twins) and merging them into two parents, GO:0006418 and
GO:0070127, because amino-acid identity is already captured by the 1:1
<aa>-tRNA ligase activity MF terms. We checked every affected term in
QuickGO, counted its annotations (six mitochondrial terms have none), and
enumerated the reviews in this repo that use them. The obsoletion matters here
because it reverses a judgment our reviews make: they mark GO:0006418 as
over-annotated because the amino-acid-specific child "already captures" the
process. The impact tables below were written on 2026-08-29, before the PSEPK
aminoacyl-tRNA batch (#2899) merged; the repo now has 33 reviews touching the
obsoleted terms, 26 of them in strictly validated core_functions, and 16
reviews that mark GO:0006418 over-annotated or modify it away (15 PSEPK
synthetases plus human AARS1). No review has been edited for the obsoletion
yet, and the module text records that QuickGO's 2026-09-22 snapshot already
obsoletes GO:0006421, GO:0006425 and GO:0070681.

It also surfaced an upstream problem worth reporting: zebrafish gtpbp3, a
tRNA-modifying GTPase rather than a synthetase, is the only gene annotated to
two of the mitochondrial terms, and a bulk migration would carry those six
over-annotations into GO:0070127.

Overview

A GO obsoletion proposal will obsolete 43 biological-process terms that name a
specific amino acid being charged onto tRNA — the 20 cytosolic
<aa>yl-tRNA aminoacylation terms, their 20 mitochondrial counterparts, two
redundant compartment-qualified variants, and the transamidation route term.
Every one of them is replaced by a compartment-level parent:

The rationale is that amino-acid specificity is a molecular-function
distinction, not a process one. Each obsoleted BP has an exact 1:1 MF
counterpart that already exists (GO:0006419 alanyl-tRNA aminoacylation ↔
GO:0004813 alanine-tRNA ligase activity, and so on for all 20 amino acids), so
no new terms are required. Upstream states explicitly that the mitochondrial
terms map to the same MF as their cytosolic counterparts, because compartment
is likewise not a molecular-function distinction: a synthetase charging tRNA in
both compartments takes one MF annotation, not two.

This project tracks the impact on AI Gene Review. Fifteen gene reviews in this
repo are affected, eleven of them through author-supplied core_functions term
ids that are strictly validated and will need to change — so this is a concrete
re-review queue, not a documentation exercise. More importantly, four reviews
contain reasoning that the obsoletion will invert (see
The four inverted judgments).

Upstream tickets

Two structural changes have already landed

The surrounding hierarchy was repaired ahead of the obsoletion, in two PRs merged
against go-ontology#15375:

Release lag caveat: as of 2026-08-29 neither OLS nor QuickGO reflects these
merges. Both still return GO:0043038 as non-obsolete, and both still label
GO:0043039 as "tRNA aminoacylation" (OLS does already carry tRNA charging
among its synonyms). Do not treat the live lookup services as evidence that the
merges have not happened; check the ontology repo instead.

Count discrepancy worth confirming upstream

The ontology ticket's proposal says "Obsoleted (42)"; the annotation ticket
lists 43 terms. The difference is GO:0070681 glutaminyl-tRNAGln biosynthesis via transamidation, which is the one entry that breaks the 1:1 pattern — it is
the indirect transamidation route (misacylated Glu-tRNA(Gln) formed by a
non-discriminating GluRS, then amidated by GatCAB), and its MF counterpart is
GO:0050567 glutaminyl-tRNA synthase (glutamine-hydrolyzing) activity, not a
glutamine-tRNA ligase. Whether it is genuinely in the same obsoletion batch
should be confirmed before any in-repo edits, because five of this repo's
fifteen affected reviews hang on GO:0070681 alone, and it is also the source
term for a concepts node in modules/bacterial_aminoacyl_trna_charging.yaml.

Obsoletion plan (per upstream)

All 43 terms confirmed live (non-obsolete) in QuickGO on 2026-08-29, as were
both replacements and a sampled MF counterpart (GO:0004813, GO:0050567).

Annotation counts below are from the QuickGO annotation API on 2026-08-29,
exact-term. exp counts ECO:0000269 and descendants; all is all evidence.

Block 1 → GO:0006418 tRNA aminoacylation for protein translation

Obsoleted term ID 1:1 MF counterpart exp all
alanyl-tRNA aminoacylation GO:0006419 GO:0004813 alanine-tRNA ligase activity 10 35,014
arginyl-tRNA aminoacylation GO:0006420 GO:0004814 arginine-tRNA ligase activity 5 40,567
asparaginyl-tRNA aminoacylation GO:0006421 GO:0004816 asparagine-tRNA ligase activity 9 18,878
aspartyl-tRNA aminoacylation GO:0006422 GO:0004815 aspartate-tRNA ligase activity 4 31,298
cysteinyl-tRNA aminoacylation GO:0006423 GO:0004817 cysteine-tRNA ligase activity 5 31,142
glutamyl-tRNA aminoacylation GO:0006424 GO:0004818 glutamate-tRNA ligase activity 7 46,903
glutaminyl-tRNA aminoacylation GO:0006425 GO:0004819 glutamine-tRNA ligase activity 6 13,762
glycyl-tRNA aminoacylation GO:0006426 GO:0004820 glycine-tRNA ligase activity 9 35,598
histidyl-tRNA aminoacylation GO:0006427 GO:0004821 histidine-tRNA ligase activity 5 34,734
isoleucyl-tRNA aminoacylation GO:0006428 GO:0004822 isoleucine-tRNA ligase activity 6 29,914
leucyl-tRNA aminoacylation GO:0006429 GO:0004823 leucine-tRNA ligase activity 8 29,621
lysyl-tRNA aminoacylation GO:0006430 GO:0004824 lysine-tRNA ligase activity 13 32,872
methionyl-tRNA aminoacylation GO:0006431 GO:0004825 methionine-tRNA ligase activity 9 30,699
phenylalanyl-tRNA aminoacylation GO:0006432 GO:0004826 phenylalanine-tRNA ligase activity 26 52,601
prolyl-tRNA aminoacylation GO:0006433 GO:0004827 proline-tRNA ligase activity 5 30,613
seryl-tRNA aminoacylation GO:0006434 GO:0004828 serine-tRNA ligase activity 11 28,880
threonyl-tRNA aminoacylation GO:0006435 GO:0004829 threonine-tRNA ligase activity 11 31,390
tryptophanyl-tRNA aminoacylation GO:0006436 GO:0004830 tryptophan-tRNA ligase activity 5 30,758
tyrosyl-tRNA aminoacylation GO:0006437 GO:0004831 tyrosine-tRNA ligase activity 8 28,429
valyl-tRNA aminoacylation GO:0006438 GO:0004832 valine-tRNA ligase activity 2 28,254
cytosolic valyl-tRNA aminoacylation GO:0061475 GO:0004832 valine-tRNA ligase activity 1 1
cytoplasmic alanyl-tRNA aminoacylation GO:1990762 GO:0004813 alanine-tRNA ligase activity 0 0
glutaminyl-tRNAGln biosynthesis via transamidation GO:0070681 GO:0050567 glutaminyl-tRNA synthase (glutamine-hydrolyzing) activity 14 56,741

Block 2 → GO:0070127 tRNA aminoacylation for mitochondrial protein translation

All 20 mitochondrial terms map to the same MF as their cytosolic twin.

Obsoleted term ID exp all
mitochondrial alanyl-tRNA aminoacylation GO:0070143 2 2,876
mitochondrial arginyl-tRNA aminoacylation GO:0070144 0 1
mitochondrial asparaginyl-tRNA aminoacylation GO:0070145 2 231
mitochondrial aspartyl-tRNA aminoacylation GO:0070146 2 160
mitochondrial cysteinyl-tRNA aminoacylation GO:0070147 0 0
mitochondrial glutaminyl-tRNA aminoacylation GO:0070148 0 0
mitochondrial glutamyl-tRNA aminoacylation GO:0070149 0 1
mitochondrial glycyl-tRNA aminoacylation GO:0070150 2 5,030
mitochondrial histidyl-tRNA aminoacylation GO:0070151 0 0
mitochondrial isoleucyl-tRNA aminoacylation GO:0070152 0 0
mitochondrial leucyl-tRNA aminoacylation GO:0070153 1 1
mitochondrial lysyl-tRNA aminoacylation GO:0070154 2 1,054
mitochondrial methionyl-tRNA aminoacylation GO:0070155 1 1
mitochondrial phenylalanyl-tRNA aminoacylation GO:0070156 1 17
mitochondrial prolyl-tRNA aminoacylation GO:0070157 0 0
mitochondrial seryl-tRNA aminoacylation GO:0070158 4 310
mitochondrial threonyl-tRNA aminoacylation GO:0070159 1 273
mitochondrial tryptophanyl-tRNA aminoacylation GO:0070183 2 3,113
mitochondrial tyrosyl-tRNA aminoacylation GO:0070184 4 296
mitochondrial valyl-tRNA aminoacylation GO:0070185 0 0

Six of the twenty mitochondrial terms have zero annotations of any kind
(GO:0070147, GO:0070148, GO:0070151, GO:0070152, GO:0070157,
GO:0070185), as does GO:1990762. The mitochondrial block carries 24 experimental annotations
across all 20 terms — fewer than GO:0006432 alone. This is strong independent
support for the upstream position: the compartment-qualified terms were created
to mirror the cytosolic set, not because curators needed them.

On the count gap

The QuickGO experimental totals come to 203 (179 in block 1, 24 in block 2)
against the upstream group tally of 251. The gap is almost certainly a filter
difference rather than a real disagreement — the upstream spreadsheet's
ComplexPortal (28), CAFA (7) and PINC (5) contributions largely do not carry
ECO:0000269-descendant evidence codes. Reconcile against the spreadsheet before
quoting either number as authoritative.

The four inverted judgments

This is the most consequential finding for this repo, and it is not a mechanical
id swap.

Four reviews here carry an explicit review action on GO:0006418 — the term
everything is being merged into — arguing that it is too general and is
superseded by the amino-acid-specific child:

Review Action on GO:0006418 Recorded reason
PSEPK/glnS MARK_AS_OVER_ANNOTATED "GO:0006425 already captures the direct substrate-specific process."
PSEPK/gltX MARK_AS_OVER_ANNOTATED "GO:0006424 already captures the defined glutamate and tRNA(Glu) substrates."
PSEPK/serS MODIFY → GO:0006434 "SerS has defined serine and tRNA substrates. GO:0006434 preserves its direct role in translational tRNA charging while recording the known amino-acid specificity."
human/AARS1 MODIFY → GO:0006419 "Use the alanine-specific aminoacylation process."

The obsoletion asserts the opposite: the substrate-specific child is the wrong
place to record specificity, and GO:0006418 is the correct BP. After the merge,
each of these four says "replace the surviving term with an obsolete one", and
two of them mark the surviving term as an over-annotation.

These reviews are not wrong about the biology — SerS really does charge serine —
they applied a general "prefer the most specific term" heuristic to an axis
(substrate identity) where GO has now decided specificity belongs on the MF. The
fix is to move the specificity claim to the MF slot (GO:0004828 serine-tRNA ligase activity etc., which these reviews already carry) and let the BP sit at
GO:0006418. This is a reusable lesson beyond tRNA charging: "more specific
is better" is not unconditional, and cross-aspect redundancy is the signal that a
BP is encoding something that belongs in MF. It belongs in
OVER_ANNOTATION_PATTERNS.

A fifth, smaller inversion: human/AARS2 has a MODIFY on GO:0006419 whose
proposed_replacement_terms is GO:0070143 — a term that is itself in the
obsoletion batch. That replacement target must be re-pointed to GO:0070127
regardless of how the rest of the review is handled.

Impact on this repo

Fifteen reviews touch the obsoleted terms or their replacements. Per CLAUDE.md,
existing_annotations[].term.id is GOA-sourced and not hard-validated, so
those will not break validation — but core_functions ids are strictly
validated, and eleven reviews use an obsoleted term inside
core_functions.directly_involved_in.

Reviews needing a core_functions change (11)

Review Accession core_functions term existing_annotations on obsoleted terms
human/AARS1 P49588 GO:0006419 ×6 — IBA, IEA, IMP (PMID:33909043), IDA (PMID:28493438), IDA (PMID:27622773), TAS (PMID:7761427); all ACCEPT
human/AARS2 Q5JTZ9 GO:0070143 GO:0070143 IBA + IMP (PMID:21549344) ACCEPT; GO:0006419 IEA MODIFY→GO:0070143
POPTR/ALARS B9HQZ6 GO:0006419 IBA + IEA, both ACCEPT
POPTR/GATC B9INH0 GO:0070681 IBA + IEA, both ACCEPT
PSEPK/gatA Q88PB9 GO:0070681 IEA ACCEPT
PSEPK/gatB Q88PC0 GO:0070681 IEA ACCEPT
PSEPK/gatC Q88PB8 GO:0070681 IEA ACCEPT
METTP/gatC A0B5K3 GO:0070681 IEA ACCEPT
PSEPK/glnS Q88IU5 GO:0006425 GO:0006425 IEA ACCEPT; GO:0006424 IEA REMOVE
PSEPK/gltX Q88LF6 GO:0006424 IEA ACCEPT
PSEPK/serS Q88FT2 GO:0006434 IEA ACCEPT

Reviews affected only in existing_annotations (2)

Reviews carrying only the replacement terms (2, unaffected)

human/AIMP1 (Q12904) and human/AIMP2 (Q13155) each carry GO:0006418
(NAS, ACCEPT) and no obsoleted term. They gain company rather than losing scope.
Worth a re-check pass only.

The gtpbp3 case is worth flagging upstream

GO:0070153 mitochondrial leucyl-tRNA aminoacylation and
GO:0070155 mitochondrial methionyl-tRNA aminoacylation have exactly one
annotation each in all of GOA, and in both cases it is zebrafish gtpbp3
(UniProtKB:Q501Z5, IMP, PMID:30916346, ZFIN) — verified via the QuickGO
annotation API on 2026-08-29. The entire existence of those two terms in the
annotation corpus rests on a single paper about a protein that is not a
synthetase: GTPBP3 is a tRNA-modifying GTPase that installs τm⁵U at the wobble
position.

This repo's existing gtpbp3 review already reached that conclusion independently,
noting that "gtpbp3KO zebrafish showed increased efficiencies of tRNA
aminoacylation", which is inconsistent with gtpbp3 acting as a direct ligase and
instead reflects an indirect consequence of altered tRNA modification.

The merge would silently roll all six of these into one GO:0070127, converting
a visible six-fold over-annotation into a single plausible-looking one. These
six annotations should be withdrawn rather than migrated, and that is worth
saying on go-annotation#6525 while the batch is still being assembled — it is
exactly the kind of case a bulk term-replacement will otherwise launder.

Module impact

modules/bacterial_aminoacyl_trna_charging.yaml is affected twice:

See also the
PSEPK ppu00970 aminoacyl-tRNA biosynthesis batch,
which curated eight of the PSEPK reviews listed above.

Mappings flagged for redirection

Upstream lists a large InterPro2GO / UniRule / HAMAP2GO / MetaCyc2GO block —
roughly 100 mappings across the 43 terms. These are the source of the ~30–50k
electronic annotations per term. Not independently verified here; see
go-annotation#6525 for the full list. Two observations:

Scope

Candidate genes for initial review

Priority order.

  1. PSEPK/glnS, PSEPK/gltX, PSEPK/serS, human/AARS1 — highest priority. These
    four carry the inverted GO:0006418 judgments. They need a re-argued review,
    not an id swap, and they are the ones that will look actively wrong once the
    merge lands.
  2. human/AARS2 (Q5JTZ9) — the proposed_replacement_terms: GO:0070143 is
    already a dangling target. Also carries an IMP on PMID:21549344 that is a
    genuine upstream experimental annotation.
  3. DANRE/gtpbp3 (Q501Z5) — the six-annotation over-annotation cluster; the
    sole basis for two of the obsoleted terms. Should drive an upstream comment
    before the batch is finalised.
  4. The five GatCAB reviews (POPTR/GATC, PSEPK/gatA, PSEPK/gatB, PSEPK/gatC,
    METTP/gatC) — all hinge on GO:0070681, whose inclusion in the
    batch is the open question above. Hold until that is settled.
  5. POPTR/ALARS (B9HQZ6) — clean mechanical case; IBA + IEA, both ACCEPT, one
    core_functions entry.
  6. human/AARSD1 (Q9BTE6) — already REMOVE; confirm the removal survives the
    merge rather than being migrated to GO:0006418.
  7. Not yet in repo, worth adding: E. coli valS (P07118) is the sole
    annotation to GO:0061475, and yeast MSR1 (P38714) / MSE1 (P48525) are
    the sole annotations to GO:0070144 / GO:0070149 (all verified via QuickGO,
    2026-08-29). Three single-annotation terms, three reviewable genes — a cheap
    way to cover the long tail of this batch.

Proposed approach

  1. Do not edit gene reviews yet. go-ontology#15375 is still open and the
    43-vs-42 question is unresolved. Editing core_functions now would desynchronise
    eleven reviews from GOA for no gain.
  2. Comment on go-annotation#6525 with the two findings this repo can
    contribute that are not in the upstream thread: (a) the gtpbp3 cluster should
    be withdrawn rather than migrated, with the single-annotation evidence above;
    (b) confirm whether GO:0070681 is in the batch, since its MF counterpart and
    MetaCyc mapping make it structurally unlike the other 42.
  3. When the obsoletion lands: re-point the eleven core_functions entries to
    GO:0006418 / GO:0070127, re-run just validate per gene, and re-fetch GOA
    so existing_annotations pick up the replacements.
  4. Re-argue, don't re-point, the four inverted reviews. Each needs its
    specificity claim relocated to the MF slot and its reason text rewritten.
    Record the general pattern in OVER_ANNOTATION_PATTERNS.
  5. Refresh the module: update modules/bacterial_aminoacyl_trna_charging.yaml
    — the GO:0043039 evidence title, and the GO:0070681 concepts term if it is
    confirmed in the batch.

Priority

High — the highest of the obsoletion projects currently in this repo. Fifteen
reviews are affected, eleven contain strictly-validated core_functions ids that
must change, and four contain reasoning the obsoletion directly contradicts. The
upstream ontology work is actively moving (two PRs merged in the last week), so
the window for contributing the gtpbp3 finding is now.

Status

Slides