Exact Q8IRM9 is the 230-residue CG32706-PA protein, with an ABT1/Esf2-specific InterPro classification and RRM at residues 62–140. PMID:15964808(https://pubmed.ncbi.nlm.nih.gov/15964808/), DOI https://doi.org/10.1128/MCB.25.13.5523-5534.2005, is cached with full text and experimentally establishes the yeast homolog’s early rRNA-processing and processome-assembly function. The GOA IBA sources include yeast Esf2; this is a conserved family transfer with a characterized mechanism, not a donor-count argument. The provided ProtNLM paragraph is already reflected in those existing annotations.
The original phmmer donor O74362 is fission-yeast Esf2, whose identical UniProt functional paragraph is itself by-similarity. The direct experimental anchor is the budding-yeast paper, not repetition of the donor paragraph. PMID:18245331(https://pubmed.ncbi.nlm.nih.gov/18245331/) remains abstract-only: both PMC and EuropePMC full-text attempts failed. The specific defense-versus-endurance mutant classification cannot be checked, so its experimental GO annotation remains UNDECIDED.
The selected frozen UniProt record is retained in CG32706-uniprot-source.json, with the complete original prediction metadata in CG32706-predictions-source.json. It maps the exact accession to this FlyBase gene; current sequence identity does not establish the historical predictor input. Gene-level experiments support conserved functions unless a relevant isoform difference is established. Falcon research was requested with perplexity-lite fallback alongside publication caching. Primary publications and sequence observations, rather than AI summaries or ARBA assertions, support the curated conclusions.
The completed Falcon report was inspected. It supports the Esf2 family-based ribosome-biogenesis inference and identifies additional yeast Dbp8 cofactor experiments, but no direct fly reaction. The broad SSU-processome/18S narrative claims are CNN; the precise requirement at yeast-named A0/A1/A2 sites is separately UNC because target site correspondence and requirement are not established. Broad conservation does not validate exact site names.
The full Falcon output was inspected. It supports the conserved Esf2 mechanism while identifying uncertainty about exact fly RNA contacts; no yeast experiment is labeled as direct fly evidence.
Identity. Q8IRM9 (230 aa, TrEMBL, CG32706-PA) is the only UniProt entry for FlyBase
FBgn0052706, so the benchmark accession is the canonical product, not a fragment or
alternative model. It starts with Met, carries the full RRM (62-140, SMART) and the
ABT1/Esf2 post-RRM core. It is shorter than yeast Esf2 (316 aa) and human ABT1 (272 aa)
because it lacks the yeast acidic N-terminal extension and the yeast C-terminal tail, ending
instead in a Lys/Ala-rich repeat (CG32706-bioinformatics/RESULTS-rereview.md, new script
compare_sequences_rereview.py; identity 33.6% to P53743, 29.4% to Q9ULW3 over ~215 columns).
Evolutionary grounding. PANTHER PTHR12311 has a single subfamily, SF7, holding fungal
Esf2, Dictyostelium esf2, vertebrate ABT1 and CG32706 (interpro/panther/PTHR12311/PTHR12311-entries.csv).
The family is one-to-one in fly, so yeast Esf2 (P53743) is the ortholog. All IBA rows come from
PAINT node PTN000267771. PAINT propagated RNA binding, nucleolus, SSU-processome assembly and
the three early cleavages, but not Esf2's Dbp8 ATPase-stimulating activity
PMID:16772403;
I treat that omission as deliberate, and the fly protein has only part of the yeast C-terminal
Dbp8-binding region, so no ATPase-activator term is proposed.
Changes vs previous review.
- The three endonucleolytic-cleavage IBA rows went from ACCEPT to KEEP_AS_NON_CORE. Esf2 is a
processome component but not the nuclease, and the yeast paper attributes the cleavage block
to failed assembly PMID:15964808. Core = GO:0034462 + RNA binding + nucleolus.
- PMID:16772403 added as direct support for RNA binding
PMID:16772403.
- GO:0042742 defense response to bacterium (IMP, PMID:18245331) went from UNDECIDED to
KEEP_AS_NON_CORE. The full text was read online at https://pmc.ncbi.nlm.nih.gov/articles/PMC2278058/
(not cacheable locally: Europe PMC and BioC return nothing). Table 3 lists PBac line 10881
"Within 100 bp of 3' end of CG32706", and Table 4 scores it + S. aureus, 0 S. typhimurium,
+ bacterial load, i.e. the defense class, which matches the curator. It is a single insertion
allele with no reported revertants, and could be an indirect ribosome-biogenesis effect, so non-core.
- The description was rewritten as a project-independent summary. Status set to COMPLETE.
Relative for next review. Yeast ESF2 (P53743, SGD:S000005337) is the experimentally
characterised one-to-one ortholog in the same PANTHER subfamily and the PAINT donor. No review
exists in the repo (no genes/*/ESF2, nor P53743, Q9ULW3, O74362 or Q9QYL7 ids).
Family review / modules. interpro/panther/PTHR12311/PTHR12311-review.yaml agrees with the SF7
placement and lists CG32706 as SF7's representative; there is no subfamily disagreement. It leaves
GO:0034462 scope UNRESOLVED family-wide, but this review accepts it for CG32706 via the PAINT node.
No module under modules/ models the SSU processome. The nearest is modules/dmel_small_nucleolar_rnps.yaml
(the U3 box C/D snoRNP, whose partner Esf2 is in yeast); CG32706 is not a participant there and should
not be, since it is not a snoRNP subunit. CG32706 is not in gocams/index.tsv.