Version history

{
  "disease": "aging, cellular senescence",
  "summary": "FOXO4 sequesters p53 in nuclear bodies in senescent cells, protecting them from apoptosis. The FOXO4-DRI all-D-amino-acid retro-inverso peptide (Baar et al. 2017, Cell) proved the concept: disrupting the FOXO4–p53 interaction triggers senescent-cell-selective apoptosis and rejuvenates aged mice. This plan designs improved disruptors using a structure-grounded approach: (1) identify the FOXO4 FHD–p53 interaction interface from available co-crystal or modeled structures (PDB 4IFR / AF2-predicted complex); (2) compute hotspot residues on FOXO4 that are both buried at the interface and contribute high ddG; (3) run BindCraft targeting those hotspots to generate de novo binder candidates; (4) score candidates by AF2-Multimer pTM/ipTM and ESM-2 log-likelihood; (5) validate top hits by independent AF3 forward-fold. Success criteria: at least one candidate with AF2-Multimer ipTM > 0.6, predicted interface area > 600 Ų, ESM-2 pseudo-perplexity < 8, and Foldseek TM-score < 0.5 against PDB (novel fold). Reference benchmark: FOXO4-DRI has estimated IC50 ~1–5 µM in NMR competition assays — new designs must show tighter predicted binding.",
  "assay_spec": "Literature-anchored hotspot mapping of FOXO4 FHD–p53 DBD interface from published NMR/crystal contacts, followed by retro-inverso peptide variant enumeration and ESM-2 pseudo-log-likelihood scoring of 8–12 candidate sequences (60–80 aa window around FOXO4 residues 28–73), capped as a methods-design plan until PDB entry for FOXO4–p53 complex is confirmed; FOXO4-DRI next-gate gate-check notebook: confirmed PDB accession for FOXO4 FHD, interface residue list (hotspots), FOXO4-DRI reference sequence, and 8–12 variant sequences with ESM-2 pseudo-log-likelihood scores — planned; {'kind': 'claim', 'text_template': 'ESM-2 pseudo-log-likelihood of variant [SEQ_ID] exceeds FOXO4-DRI reference score by ≥ 0.5 nats/residue, suggesting improved sequence fitness at the p53-binding interface — planned, pending ESM-2 run'}; Updated FOXO4-DRI binder design research plan: confirmed target PDB, hotspot table, variant shortlist (≤5 sequences), and explicit next-gate — RFdiffusion partial-diffusion on confirmed PDB when budget resumes — planned",
  "hypothesis": "A de novo protein binder targeting the FOXO4 forkhead domain surface that contacts p53 will disrupt FOXO4–p53 nuclear-body sequestration in senescent cells, triggering selective apoptosis with higher potency and lower off-target risk than the original FOXO4-DRI retro-inverso peptide.",
  "target_ref": "UniProt:P98177",
  "kill_criteria": "Abort or revise if: No co-crystal structure of FOXO4–p53 may exist in PDB; fallback is AlphaFold P98177 model + literature-mapped interface, which reduces structural confidence; FOXO4-DRI is an all-D-amino-acid retro-inverso peptide — ESM-2 is trained on L-amino acid sequences; perplexity scores on DRI variants are a proxy only and must be flagged as such in the artifact; research_plan:5e4168c3 context ref is unread this tick; must check before creating duplicate FOXO4 research_plan",
  "primary_endpoint": "Correct FOXO4 FHD PDB accession confirmed (resolution ≤ 3.0 Å or high-quality NMR ensemble; not 53BP1 or unrelated Tudor domain)",
  "identification_strategy": "in_silico_KO"
}