CG4793 is a noncatalytic serine protease homolog with experimentally supported roles in cellular immunity. ProtNLM's serine-type endopeptidase activity prediction is contradicted by the absence of the catalytic serine in the selected sequence. This is a pseudoenzyme error, not merely disagreement with a preferred gene name or a generic caution in a database record.
The reproducible sequence analysis maps the catalytic serine of two active trypsin controls to Gly290 in Q8IP30. The full 910-residue sequence is identical to RefSeq NP_723941.2; UniProt maps it to FlyBase FBgn0028514 and CG4793. The N-terminal S1 domain covers residues98–340, followed by a long repetitive tail. Its length and architecture differ from small single-domain SPH proteins. Alignment covers about97% of each mature control, and the separate active-control run recovers an intact triad. The more divergent histidine loop makes the precise Glu145 site mapping less secure than Gly290. Neither historical prediction-time input nor isoform-specific experimental action is established by present-day accession identity.
Kr et al. 2021, PMID:33644706, DOI 10.17912/micropub.biology.000370, provides full-text gene-specific RNAi evidence. Figure1 and its legend state: “Conversely, knockdown of CG4793 in lamellocytes resulted in a significant increase in the encapsulation rate.” Two independent constructs show this effect; plasmatocyte knockdown does not. This supports NEW GO:0140540, negative regulation melanotic encapsulation of foreign target, in the tested Leptopilina clavipes infection model. The paper's mechanistic interpretation remains explicitly provisional: “However, a detailed mechanistic understanding of how these S1A family genes regulate cellular immunity and how their activity may be linked to JAK-STAT pathway signaling remain to be established.” The experiments establish regulation, not direct cleavage or a particular binding partner. The published reagent table appears internally inconsistent in its TRiP genotype/stock labeling; the distinct NIG4793R RNAi line and the reported convergent gene-specific results provide a second experimental handle, but construct identity would merit checking in follow-up replication.
Bina et al. 2010, PMID:20168330, DOI 10.1038/embor.2010.1, is abstract-only in the standard cache. The full article was accessed and read separately: PDF, unmodified extracted text, and download provenance. Figure3F and the page206 results place CG4793 among the genes whose knockdown suppresses melanotic tumour formation in the constitutively active hopTumL JAK model. The text states: “CG4793, CG1572 and CG13559—were identified whose knockdown significantly reduces the development of tumours in the adult fly”. Direct circulating-haemocyte counts in Figure3G/H concern baz, not CG4793. Retain FlyBase's positive-regulation-of-haemocyte-proliferation IGI annotation as KEEP_AS_NON_CORE, interpreted in this activated-JAK tumour context. This is distinct from the physiological encapsulation response and does not establish a universal positive regulator of cell division.
The four GOA NOT activity annotations (three serine-type endopeptidase, one peptidase) are accepted. They agree with target-specific catalytic-site loss and the published SPH framework. Cao and Jiang2018, PMID:30367934 explicitly classify SPs versus SPHs by presence/absence of the His–Asp–Ser triad. Its family-level discussion is consistent with the local analysis; it is not a substitute for checking Q8IP30. Ross et al.2003, PMID:12568721 is available as an abstract and records the original genome-wide framework; its detailed target alignment was not accessible here.
The extracellular-region IBA is accepted as an inference consistent with the exact sequence's predicted signal peptide, residues1–18. ARBA family text and keywords are not counted as independent support. The PTHR24256 PAINT export places both the extracellular-region and proteolysis IBDs at PTN001207268 and, after refetching with --extra-uniprot Q8IP30, shows the nested PTN000668109 IKR loss call that prunes the non-negated serine-type endopeptidase IBD from CG4793's branch. Both positive proteolysis annotations remain UNDECIDED. An inactive domain cannot catalyze peptide-bond hydrolysis, and merely regulating an active protease would be regulation of proteolysis rather than proteolysis itself; the open possibility, grounded in Cao and Jiang 2018's SPH/proPO review, is that CG4793 acts as a noncatalytic SPH cofactor in a protease activation step. Neither gene-specific study establishes or excludes that proteolytic participation.
The frozen source JSON is Q8IP30-predictions.json, copied exactly in content from the project's 2026-09-08 release snapshot. It records ProtNLM2, model score0.99, phmmer accessionQ7Z410, score226.0 for GO:0004252. These values describe prediction provenance. They cannot compensate for loss of the catalytic nucleophile. A separate secreted-location prediction is present in raw metadata but is outside this functional GO claim review.
Falcon deep research returned successfully through the repository workflow, concurrently with publication caching. The report correctly separates family annotation from intrinsic catalytic activity, but did not retrieve the gene-specific primary experiments in enough detail: its statement that the accessible evidence does not identify whether CG4793 controls encapsulation is resolved by the full 2021 article and Figure1. Its catalytic-status uncertainty is resolved at the sequence-inference level by the independently reproduced replacement of the catalytic serine by Gly290; no purified-protein assay is claimed. The report also repeats a generic CLIP architecture from supplied family metadata, whereas the target has a long repetitive C-terminal extension and an N-terminal protease-like domain. The primary articles, exact sequence, and local computations support the specific conclusions above. The current review's unresolved molecular activity and process assignments are intentional; a specific peptidase-regulator molecular function would require a demonstrated target interaction.
Scope: every existing annotation, the description and core functions were re-examined against the cached literature, the exact-sequence analysis in CG4793-bioinformatics/RESULTS.md, and PANTHER/PAINT placement. ProtNLM2 output was not consulted.
Identity. Q8IP30 (910 aa, unreviewed) is the only UniProtKB entry cross-referenced to FlyBase FBgn0028514 (UniProt search xref:flybase-FBgn0028514, 2026-10-20) and matches RefSeq NP_723941.2. It is the canonical full-length product of the gene, not a fragment, isoform or alternative gene model. The S1 domain spans residues 98–340; the C-terminal ~570 residues are a repetitive extension with no second protease domain.
Phylogenetic placement. PANTHER 19 geneinfo places Q8IP30 in PTHR24256:SF462 (official name "FI06461P-RELATED"; leaf node PTN000668118). The InterPro PANTHER match (residues 1–341) carries no subfamily. PANTHER matchortho lists only "O" (non-LDO) orthologs. In Anopheles gambiae these are the AGAP011781–011790 cluster (CLIPA5, CLIPA6, CLIPA12, CLIPA13, CLIPE6, A0A453YY93, A0A1S4H7K7, A0A1S4H8N3), and all checked members are in SF462. Vertebrate PRSS21/42/43/44/48/54 appear only because of a deep, unresolved Eukaryota-to-Eumetazoa duplication series. The nearest Drosophila paralogs come from Drosophila-specific duplications (SPH93, CG18563, SPH97, c-SPH125, CG31827, LD13269, c-SPH94, c-SPH35) and a Diptera duplication (CG8738, CG8586). None of these has experimental GO annotation: QuickGO, 2026-10-20, shows only IBA/IEA/IKR rows for CG8738, CG8586 and the four CLIPA/CLIPE SF462 members. The well-characterised fly proteases in the family, such as easter (ea) and Sp7/MP1, separate from CG4793 at the deep Eumetazoa-level duplication. No subfamily-specific function can therefore be transferred to CG4793 from a characterised relative. Its functional statements rest on its own experiments plus its own catalytic-site loss.
PAINT. interpro/panther/PTHR24256/PTHR24256-paint.tsv puts the extracellular-region, proteolysis and positive GO:0004252 IBDs at PTN001207268. A nested IKR at PTN000668109 (one of eight sibling IKR loss nodes) removes GO:0004252 from CG4793's branch. The curators negated the MF and did not negate proteolysis. That is the usual PAINT practice of pruning MF only, and it does not settle the question for the BP.
Family review. interpro/panther/PTHR24256/PTHR24256-review.yaml lists CG4793 under "Benchmark members without resolved subfamily placement" and assigns no subfamily id. That is not a disagreement, because the family review asserts nothing. It could now record SF462 (PANTHER 19) for Q8IP30, but that would be a family-review edit outside this gene folder.
Literature. A Europe PMC search for "CG4793" (2026-10-20) found no functional study beyond Bina et al. 2010 (PMID:20168330) and the 2021 microPublication (PMID:33644706). The other hits are expression or screen papers that only mention the gene.
Annotation decisions (re-checked; unchanged).
- NOT serine-type endopeptidase (IBA, 2×IKR) and NOT peptidase activity (IKR): ACCEPT. Gly290 sits at the catalytic Ser position.
- extracellular region (IBA): ACCEPT. Consistent with the SignalP 1–18 prediction.
- proteolysis (IBA, IEA): UNDECIDED. The protein cannot catalyse cleavage. Whether it acts as a cofactor in a cleavage step (the "cofactor" case in CLAUDE.md, as for PPO-activating SPHs) has been neither tested nor excluded, so it is not called over-annotated.
- positive regulation of hemocyte proliferation (IGI, PMID:20168330): KEEP_AS_NON_CORE. The full text supports a hopTumL tumour-suppression interaction; the effect is context-dependent.
- NEW GO:0140540 negative regulation melanotic encapsulation of foreign target (IMP, PMID:33644706): retained. Comparator check: FlyBase already uses this term for IMP/RNAi regulators (Spn27A PMID:15749104; lwr and Uba2 PMID:21203476). QuickGO has no GO annotation from PMID:33644706 at all, so its absence reflects a paper that has not been curated, not a deliberate decline. The term is a regulation term, not a claim that CG4793 participates in encapsulation.
Edits made. The description was rewritten to state the SPH status, the SF462 placement and the Kc167 context of the JAK/STAT target finding. Status was set to COMPLETE. Two suggested experiments were added (partner identification; G290S catalytic-restoration rescue). No annotation action, reference or bioinformatics file was changed or removed.
Modules. No module under modules/ contains CG4793. modules/dmel_toll_signaling.yaml models a related family proteolytic cascade (Toll pathway), but CG4793 has no reported role in it.