CG18547 (Q9VGF3) bioinformatics: PANTHER tree placement and AKR tetrad

Scripts (run with uv run <script>): panther_tree_placement.py (output
panther_tree_placement.out) and tetrad_check.py (output tetrad_check.out).
Both fetch live data (PANTHER treeinfo API, UniProt REST); nothing is hard-coded
except the reference positions in O81884, which come from its UniProt record.

Tree placement (PANTHER 19, PTHR42686)

PAINT context (QuickGO, checked 2026-10-21): GO:0010349 L-galactose dehydrogenase activity
is propagated to O81884 from PTN000920391 (plant node), and ARA2's D-arabinose
dehydrogenase / dehydro-D-arabinono-1,4-lactone biosynthesis IBAs come from PTN000920371
(fungal node). CG18547 receives only GO:0016491 oxidoreductase activity from the SF1 root
PTN002448388. Neither specific function was placed at a node ancestral to CG18547.

AKR catalytic tetrad (global pairwise alignment, BLOSUM62)

O81884 (LGALDH) CG18547 (Q9VGF3) ARA2 (Q04212)
Asp54 Asp71 Asp53
Tyr59 (proton donor, UniProt ACT_SITE) Tyr76 Tyr58
Lys87 Lys102 Lys86
His124 (substrate binding, UniProt BINDING) His143 His124

Identity over aligned positions: CG18547 vs LGALDH 42.6% (133/312);
ARA2 vs LGALDH 34.3% (104/303). CG18547 retains the complete Asp-Tyr-Lys-His tetrad,
so it is expected to be a catalytically competent AKR. Retention of the tetrad says
nothing about substrate: L-galactose, D-arabinose, D-threo-aldose and pyridoxal
dehydrogenases in the same subfamily all carry it.