Target: horse A0A9L0RGD6. Source: frozen ProtNLM2 API snapshot2026-09-08, exact output in projects/PROTNLM_EVALUATION/mammal-benchmark/horse40-predictions.csv. This is an external prediction, separate from GOA.
Component of the shieldin complex, which plays an important role in repair of DNA double-stranded breaks (DSBs). During G1 and S phase of the cell cycle, the complex functions downstream of TP53BP1 to promote non-homologous end joining (NHEJ) and suppress DNA end resection. Mediates various NHEJ-dependent processes including immunoglobulin class-switch recombination, and fusion of unprotected telomeres
| Claim | Assessment | Evidence and limitation |
|---|---|---|
| Shieldin-complex component | COR | The sequence is a close SHLD2-family match, including the long-form insertion; this supports family/complex identity. Complex recruitment alone does not demonstrate an intact DNA-binding module. |
| Downstream of TP53BP1 in DNA-break repair | UNC for the exact horse model | Human primary studies place shieldin downstream of 53BP1. The selected horse C-terminal OB region is divergent/gapped; biological family membership and functional repair are distinct claims. |
| Promotes NHEJ and suppresses end resection | UNC for the exact horse model | The human short splice product can assemble/localize yet fail to suppress homologous recombination. The horse model cannot be certified from recruitment alone. |
| G1/S-phase activity | UNC | The human pathway provides a cell-cycle context, but the specified timing and repair competence have not been resolved for this horse model. |
| Immunoglobulin class switching | UNC | A demonstrated mammalian shieldin-dependent NHEJ outcome requires a functioning end-protection module; that transfer remains uncertain. |
| Fusion of unprotected telomeres | UNC | The mammalian pathway result is plausible at gene-family level but remains conditional on the exact horse protein's repair function. |
The principal sources are Noordermeer et al., PMID:30022168, DOI 10.1038/s41586-018-0340-7, and Dev et al., PMID:29789392. See sequence and isoform analysis and review notes. Neither the human AI review text nor ARBA agreement supplies independent validation.
COR here denotes a biologically supported claim absent from the frozen target GOA, not proof of absence from training data. The paragraph is retained separately because the PredictionReview schema currently represents GO/EC terms rather than narrative claims.