ACTG2: a conventional actin with an entirely inferred GO record

Generated by uv run python analyze_actg2.py. Every number below is computed at run time
from the UniProt REST API, RCSB coordinate files, QuickGO, and the repository's cached
PANTHER PTHR11937 PAINT table. Nothing is hard-coded from a previous run.

1. Reference-sequence length audit (run before any conservation number)

The ACTL10 review showed that a Swiss-Prot entry starting mid-fold manufactures fake
substitutions out of residues the sequence never reaches. So the panel is length-audited
first, and the audit is required to flag the known-short entry before it may clear ACTG2.

Conventional-actin modal length: 377 aa (P60709 375, P63261 375, P68133 377, P68032 377, P62736 377, P63267 376).

Entries shorter than 85% of that: {'Q5JWF8': 245} (ACTL10 is the entry the ACTL10 review traced to a lost initiator; it is flagged here by
the same criterion, which is what licenses trusting the rest of the panel).

ACTG2 is 376 aa and matches every reviewed one-to-one orthologue: Actg2 Rattus norvegicus 376 aa, Actg2 Mus musculus 376 aa, ACTG2 Gallus gallus 376 aa.
So ACTG2's own reference sequence is full length and no part of this analysis is a
truncation artefact.

2. Filament protomer interface, with the sibling panel reproduced first

PDB 6DJO chain C was used; interface residues are every residue
within 4.0 A of any other protein chain: 38 residues.

Reproduction check against the merged ACTL8 review's committed panel (genes/human/ACTL8/ACTL8-bioinformatics/RESULTS.md, section 2): all 14 published columns reproduced exactly. Only
because of that is the new column trustworthy.

ACTG2 scores 38/38 chemically compatible (38 identical, 0 conservative, 0 non-conservative, 0 gaps) and ACTA2 38/38. Both sit with the conventional actins
at the top of the panel, not in the divergent band where the actin-like proteins reviewed
earlier in this campaign fall. The protomer interface that builds a two-stranded filament is
therefore intact in both smooth muscle actins, which is the positive structural counterpart
to the annotation gap reported in section 5.

3. ACTG2's own filament structure and its bound nucleotide

PDB 8V2O (wild-type ACTG2 filament) chain C gives 67 inter-protomer interface residues under the same 4.0 A criterion, i.e. ACTG2 is observed as a multi-protomer filament in
its own right rather than inferred to be one. That count is deliberately not compared with
the 38 of section 2: 6DJO is a four-protomer reconstruction and 8V2O a five-protomer one, so the chosen chain has a different number of neighbours.
The comparable statement is the one in section 2, where the same 38 positions are scored
across every sequence.

Bound heteroatoms per ACTG2 structure, read from the coordinates:

HIC is 4-methyl-histidine: the recombinant human-cell-expressed ACTG2 used for these
structures carries the SETD3 methyl mark at His74 in every protomer, which independently
corroborates UniProt's experimentally-evidenced MOD_RES 74. DTH/EEP/HYP in 8V30 are
phalloidin components, matching that structure's need for phalloidin to polymerise R40C.

Every ACTG2 filament structure carries ADP, not ATP, in every protomer. Following the
principle used by the merged ACTR1A/ACTR1B/ACTR5 reviews (annotate the ligand actually
observed), the directly-supported nucleotide term for ACTG2 is ADP binding; ATP binding is
supported instead by UniProt's own catalytic-activity reaction and by the source paper's
statement that hydrolysis happened during sample preparation.

4. WITH/FROM audit (source lists built from the GOA field, never by hand)

16 of ACTG2's GOA rows carry a WITH/FROM field, with 54 tokens in total. The counts below are asserted equal to the GOA field
inside the script, so a hand-edited source list cannot drift from GOA.

Distinct source proteins across all rows: 31; of these, 31 carry their own experimental-code annotation for the term they donated, and 28 are reviewed (Swiss-Prot) entries. Those are two different numbers and are
reported separately, because an unreviewed entry's protein NAME is an automatic label even
when its GO annotations are genuine experimental ones.

Tokens that no query resolved: none. An unresolved WITH/FROM token
can be deferred but not dismissed, so any that remain are named here rather than dropped.

Per-donor detail, including Swiss-Prot/TrEMBL status and which lookup answered:

GOA line 1: GO:0005576 extracellular region (IBA, GO_REF:0000033)

GOA line 2: GO:0015629 actin cytoskeleton (IBA, GO_REF:0000033)

GOA line 3: GO:0005200 structural constituent of cytoskeleton (IBA, GO_REF:0000033)

GOA line 4: GO:0005856 cytoskeleton (IEA, GO_REF:0000044)

GOA line 5: GO:0007010 cytoskeleton organization (IEA, GO_REF:0000108)

GOA line 6: GO:0045471 response to ethanol (IEA, GO_REF:0000107)

GOA line 7: GO:0071354 cellular response to interleukin-6 (IEA, GO_REF:0000107)

GOA line 8: GO:1905641 cellular response to acetaldehyde (IEA, GO_REF:0000107)

GOA line 9: GO:0016887 ATP hydrolysis activity (ISS, GO_REF:0000024)

GOA line 10: GO:0032982 myosin filament (ISS, GO_REF:0000024)

GOA line 11: GO:0005737 cytoplasm (ISS, GO_REF:0000024)

GOA line 12: GO:0010628 positive regulation of gene expression (ISS, GO_REF:0000024)

GOA line 13: GO:0030027 lamellipodium (ISS, GO_REF:0000024)

GOA line 14: GO:0030175 filopodium (ISS, GO_REF:0000024)

GOA line 15: GO:0044297 cell body (ISS, GO_REF:0000024)

GOA line 16: GO:0090131 mesenchyme migration (ISS, GO_REF:0000024)

5. PAINT node audit inside PTHR11937: which node carries which term

For each term, the cached PAINT table gives the node-level assertions and negations, and
QuickGO gives the human genes each asserting node actually reaches.

GO:0005884 actin filament

• asserted at PTN000233075 (IBD, taxon taxon:32523, 20190301), seeds: UniProtKB:P68133
• reaches 1 human gene(s): ACTA1 (enumeration complete: True, 54/54 annotations scanned)
• asserted at PTN000748220 (IBD, taxon taxon:32523, 20190301), seeds: UniProtKB:P68032
• reaches 1 human gene(s): ACTC1 (enumeration complete: True, 54/54 annotations scanned)
• asserted at PTN002631586 (IBD, taxon taxon:117571, 20190301), seeds: MGI:MGI:87906, UniProtKB:P63261
• reaches 9 human gene(s): ACTB, ACTBL2, ACTG1, ACTL8, POTEE, POTEF, POTEI, POTEJ, POTEKP (enumeration complete: True, 54/54 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 0 none

GO:0005200 structural constituent of cytoskeleton

• asserted at PTN000940351 (IBD, taxon taxon:, 20250805), seeds: SGD:S000001855, UniProtKB:P61160, MGI:MGI:87906, dictyBase:DDB_G0269234, SGD:S000002513, UniProtKB:P61158, dictyBase:DDB_G0289811, RGD:1304556, UniProtKB:P60709, SGD:S000001171
• reaches 10 human gene(s): ACTA1, ACTA2, ACTC1, ACTG2, ACTL10, ACTL9, ACTR10, ACTRT1, ACTRT2, ACTRT3 (enumeration complete: True, 55/55 annotations scanned)
• negated (IRD) at PTN000233596 (20260416), against PANTHER:PTN000940351
• negated (IRD) at PTN000233752 (20250805), against PANTHER:PTN000940351
• negated (IRD) at PTN000233796 (20260416), against PANTHER:PTN000940351
• negated (IRD) at PTN000233887 (20250805), against PANTHER:PTN000940351
• negated (IRD) at PTN000234048 (20250805), against PANTHER:PTN000940351
• negated (IRD) at PTN001732543 (20250805), against PANTHER:PTN000940351
• negated (IRD) at PTN007551901 (20260416), against PANTHER:PTN000940351
• negated (IRD) at PTN008986528 (20250805), against PANTHER:PTN000940351
• ACTG2's own annotations to this term or a descendant: 1 GO:0005200 IBA GO_REF:0000033

GO:0015629 actin cytoskeleton

• asserted at PTN002631484 (IBD, taxon taxon:2759, 20260528), seeds: RGD:1304556, WB:WBGene00000064, UniProtKB:P60709, UniProtKB:Q8I4X0, MGI:MGI:87906, PomBase:SPBC32H8.12c, WB:WBGene00000066, SGD:S000001855, dictyBase:DDB_G0269234, RGD:628837, UniProtKB:Q6QAQ1, dictyBase:DDB_G0289811, WB:WBGene00000067, MGI:MGI:87909, UniProtKB:P68032, UniProtKB:P08023, dictyBase:DDB_G0289487, dictyBase:DDB_G0275023, WB:WBGene00000065, RGD:621676, UniProtKB:P68133, CGD:CAL0000191211, FB:FBgn0011743, UniProtKB:P63261
• reaches 18 human gene(s): ACTA1, ACTA2, ACTB, ACTBL2, ACTC1, ACTG1, ACTG2, ACTL10, ACTL8, ACTL9, ACTRT1, ACTRT2, ACTRT3, POTEE, POTEF, POTEI, POTEJ, POTEKP (enumeration complete: True, 349/349 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 2 GO:0015629 IBA GO_REF:0000033, GO:0030175 ISS GO_REF:0000024

GO:0005576 extracellular region

• asserted at PTN004322804 (IBD, taxon taxon:, 20260528), seeds: RGD:621676
• reaches 2 human gene(s): ACTA2, ACTG2 (enumeration complete: True, 1420/1420 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 4 GO:0005576 IBA GO_REF:0000033, GO:0005576 HDA PMID:23580065, GO:0070062 HDA PMID:23533145, GO:0072562 HDA PMID:22516433

GO:0017022 myosin binding

• asserted at PTN000748220 (IBD, taxon taxon:32523, 20190301), seeds: UniProtKB:P68032
• reaches 1 human gene(s): ACTC1 (enumeration complete: True, 23/23 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 0 none

GO:0033275 actin-myosin filament sliding

• asserted at PTN000748220 (IBD, taxon taxon:32523, 20190301), seeds: UniProtKB:P68032
• reaches 1 human gene(s): ACTC1 (enumeration complete: True, 2/2 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 0 none

GO:0007015 actin filament organization

• asserted at PTN000748220 (IBD, taxon taxon:32523, 20190301), seeds: MGI:MGI:87905
• reaches 1 human gene(s): ACTC1 (enumeration complete: True, 275/275 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 0 none

GO:0005198 structural molecule activity

• asserted at PTN008986528 (IBA, taxon taxon:, 20250805), seeds: PANTHER:PTN000940351
• reaches 2 human gene(s): ACTL7A, ACTL7B (enumeration complete: True, 484/484 annotations scanned)
• ACTG2's own annotations to this term or a descendant: 1 GO:0005200 IBA GO_REF:0000033

6. Reference-projection check

For each literature reference behind an ACTG2 row (directly, or through a donor), how many
distinct gene products does it annotate, and does the functional term spread across the
set or stay on the perturbed gene? An annotation count is not an entity count, and a page
total is not a whole result, so incomplete enumerations are reported as unavailable.

6a. Is the chicken-ACTA2 ISS block also on human ACTA2?

The donor of ACTG2's ISS block is P08023 = ACTA_CHICK, gene ACTA2, Gallus gallus, 377 aa (Swiss-Prot) - "Actin, aortic smooth muscle".
The chicken orthologue that was available and not used is P63270 = ACTH_CHICK, gene ACTG2, 376 aa (Swiss-Prot).

• ACTG2 ISS rows from that donor (6): GO:0005737, GO:0010628, GO:0030027, GO:0030175, GO:0044297, GO:0090131
• human ACTA2 ISS rows from the same donor (5): GO:0010628, GO:0030027, GO:0030175, GO:0044297, GO:0090131
• shared (5): GO:0010628, GO:0030027, GO:0030175, GO:0044297, GO:0090131
• on ACTG2 only (1): GO:0005737

So the transfers are not unique to ACTG2: the donor's true human orthologue already
carries them, and the ACTG2 copies are the same inference aimed at the other smooth
muscle actin.

7. Canonical source labels for the review YAML

check_source_labels.py asserts the review's source_label strings against these, so a
retyped residue count cannot drift (six of them did on the first round of this review).