FANCG (FANCG) — AIGR vs Affinage

Affinage record: run 2026-06-09 · 30 discoveries · self-eval pairwise "win" (faith 100%) · gates passed.

Agreement (brief)

Both sources converge on the central biology: FANCG (= XRCC9) is a TPR-repeat
scaffold subunit of the nuclear Fanconi anemia (FA) core complex, with no intrinsic
catalytic or DNA-binding activity. It binds FANCA directly, is required for complex
assembly/stability/nuclear accumulation, and the assembled core complex drives
damage-induced FANCD2(-FANCI) monoubiquitination and replication-coupled interstrand
crosslink (ICL) repair. Affinage's narrative (FANCA-FANCG mutual stabilization and
nuclear import, cryo-EM contacts, TPR-motif interaction scaffold, requirement for
FANCD2 monoubiquitination, mitochondrial/PRDX3 oxidative-stress pool) is fully
consistent with the AIGR review's core_functions (adaptor activity + ICL repair +
DNA damage response, nucleoplasm/chromatin, FA nuclear complex) and its non-core
classification of the mitochondrial role.

Disagreements

Topic	Affinage says	AIGR review says	Verdict (who is right + why)
MF grounding	mechanism_profile: GO:0060090 molecular adaptor activity and GO:0005198 structural molecule activity	Adaptor activity grounded specifically as GO:0030674 protein-macromolecule adaptor activity (NEW); no structural-molecule term	AIGR. GO:0030674 is the precise adaptor term for a protein-protein bridging scaffold; GO:0005198 (structural constituent, e.g. cytoskeleton/ribosome) is the wrong sense for a signaling/assembly adaptor. Affinage's layer is coarse and lists both parents.
HR-mediated DSB repair	Foregrounds a Ser7-dependent D1-D2-G-X3 (BRCA2/FANCD2/XRCC3) complex and "FANCG required for efficient HR repair of DSBs" (PMID:12861027, PMID:18212739, PMID:16621732)	Originally folded HR only into the core_functions prose; no discrete HR annotation	Affinage flags a real gap. AIGR now adds GO:0000724 (double-strand break repair via HR), NEW/non-core, citing PMID:12861027 (DT40 knockout, ~9-fold HR drop) + PMID:18212739. Direct experimental, correctly branched.
DNA-binding MF	Narrative does not claim FANCG binds DNA; grounding lists only adaptor/structural MF	GOA legacy TAS "damaged DNA binding" (GO:0003684) flagged MARK_AS_OVER_ANNOTATED	Agreement in substance. Affinage's narrative implicitly agrees FANCG has no DNA-binding activity; AIGR makes the over-annotation call explicit.
Localization	mechanism_profile: nucleus, cytosol, mitochondrion (flat list, no core/non-core weighting)	Nucleus/nucleoplasm/chromatin = core; cytosol/cytoplasm/mitochondrion/nuclear-speck = non-core	AIGR. Same locations, but AIGR correctly weights the nuclear/chromatin pool as functional and demotes the minor cytoplasmic/mito pools.
Reactome pathway	Collapses to R-HSA-73894 "DNA Repair" and R-HSA-1643685 "Disease" (very general parents)	Uses specific Reactome ICL-repair reactions for nucleoplasm localization	AIGR. Affinage's Reactome grounding is at uninformative top-level nodes.

Papers incorporated into the review

PMID	Supports	How used
PMID:12861027	FANCG required for efficient HR repair of I-SceI-induced DSBs (DT40 knockout, ~9-fold)	New reference (HIGH/VERIFIED) + supported_by for the new GO:0000724 annotation
PMID:18212739	Ser7-phospho FANCG promotes D1-D2-G-X3 (BRCA2/FANCD2/XRCC3); FANCG–XRCC3 epistatic for crosslink sensitivity	New reference (HIGH/VERIFIED) + supported_by for the new GO:0000724 annotation

Net assessment

AIGR was already the stronger, better-grounded record: correct precise MF term
(GO:0030674 vs Affinage's coarse GO:0060090/GO:0005198 pair), correct core/non-core
weighting of localizations, and an explicit over-annotation call on legacy "damaged
DNA binding" that Affinage's coarse GO layer would not surface. Affinage's value here
was the dense, correctly-cited narrative, which surfaced one genuine gap: FANCG's
direct, experimentally demonstrated requirement for HR-mediated DSB repair (via the
Ser7-dependent D1-D2-G-X3 complex). That function is now added as a single NEW,
non-core GO:0000724 annotation backed by two verified primary papers. No Affinage
mechanism_profile GO ids were imported, and no existing AIGR decision was weakened.