Version history

{
  "disease": "aging",
  "summary": "CD38 (UniProt P28907, PDB 2I65/4CMH/5F1K) catalytic-cleft binder design. Ticks 1-55: hotspot geometry confirmed — 8 consensus hotspot residues (S126, D155, D156, L157, D179, K190, Q226, G245); all 8 present in 2I65 chain A with full neighbor profiles. PDB 5F1K confirmed at 2.3 Å (nanobody MU1053 complex). Benchmark anchors: daratumumab KD ~1-5 nM (PMID 21248249), TNB-738 IC50 ~0.1 nM (PMID 35867844). BindingDB Kd/IC50/Ki queries all returned not_found=true; SKEMPI API returned 422 (uniprot_id not a valid param); EuropePMC sweeps (binder/peptide/structure-based and nanobody/engineering queries) returned no on-point de novo protein binder papers against CD38 catalytic cleft in 2023-2025. Novelty hurdle declared clear. Tick 56: BindCraft invoked — 50 designs, PDB 2I65 chain A, hotspots S126/D155/D156/L157/D179/K190/Q226/G245, binder length 50-80 aa.",
  "assay_spec": "Tick 56 — BindCraft batch dispatch. Planned post-BindCraft pipeline (pending tool return): ESM-2 pseudo-perplexity ≤8.0 → AF2-Multimer ipTM ≥0.6 → Foldseek novelty screen (TM-score <0.5 vs PDB) → Rosetta ddG ≤-10 REU → protein_design artifact creation for passing candidates. Success criterion: ≥1 of 50 designs passes ESM-2 pseudo-perplexity ≤8.0 filter.",
  "hypothesis": "A de novo protein binder engaging the CD38 catalytic cleft at hotspot residues S126, D155, D156, L157, D179, K190 (2I65 chain A) will competitively inhibit CD38 NADase activity, restore NAD+ levels in aged tissues, and outperform small-molecule inhibitors (e.g., 78c, apigenin) on selectivity against BST-1 (CD157) based on in-silico binding metrics (ipTM > 0.7, AF2 pLDDT > 80, Rosetta ddG < -10 REU).",
  "target_ref": "UniProt:P28907 / PDB:2I65",
  "kill_criteria": "Abort or revise if: (1) pdb_hotspot_profile returns no neighbors within 4.5 Å for ≥3 of 7 hotspot residues — requires PDB re-inspection; (2) AF2-Multimer ipTM < 0.5 on all 50 BindCraft designs — redesign with adjusted hotspot weighting; (3) ESM-2 pseudo-perplexity > 12.0 on all 50 designs — indicates sequence space collapse, widen binder length range or loosen hotspot constraints; (4) Foldseek TM-score ≥ 0.7 for majority of passing designs vs PDB — novelty insufficient, re-run with diversified scaffold seeds.",
  "primary_endpoint": "At least 1 of 50 BindCraft designs passes ESM-2 pseudo-perplexity ≤ 8.0",
  "identification_strategy": "in_silico_KO"
}