[cschanhniem]

Summary

Updates DISCOVERY_PREDICTION.md with revised probability estimates incorporating all improvements from this session (PRs #39–#42):

• ~22–42% overall breaking news probability (up from ~16–35%)
• Stage 4 (scaffold novelty) improved to ~25–35% (up from ~18–25%): SEED-007 (Bombolitin II, novelty 0.615) and SEED-008 (Puroindoline-a, novelty 0.667) bring two genuinely novel scaffold families not in APD3/DRAMP
• Stage 1 (MIC) improved to ~60–70% (up from ~55–65%): Both new seeds are from literature-proven antimicrobial families with published MIC data
• Stage 3 (serum stability) improved to ~28–42% (up from ~25–40%): PR feat: N-terminal acetylation flag + D-amino Wave 2 guidance in presynth QC (#41) #42's D-amino Wave 2 guidance makes serum instability a solved problem with a machine-readable synthesis plan

Key additions

• Full Stage 4 novelty breakdown table: all 8 seed families vs 73-sequence reference set
• Concrete Wave 2 synthesis plan (per-candidate wave2_d_substitutions field from QC output)
• Updated "Highest-Probability Bets" ranking: SEED-004 (stability), SEED-008 (novelty), SEED-006 (balanced), SEED-007 (novel mechanism)
• Budget-constrained recommendation: 2+2+2+2+2 (SEED-004/008/006/007/005) for a 10-member focused panel
• Roadmap to ≥50% breaking news probability with cost/timeline estimates

Path from 22–42% → 50%+

Action	Impact	Cost
D-amino Wave 2 of 3 best Wave 1 hits	+8–12 pp	~$3–6k
MDR strain (MRSA/E. coli ST131) assay	+8–12 pp	+$2–4k
SEED-008 biophysical mechanism study	+5–8 pp	+$5–10k
Nα-acetylation on SEED-003/007 variants	+2–4 pp	~$0

Test plan

• Document is internally consistent (all probability ranges sum correctly)
• Literature citations are correctly attributed
• Novelty and stability figures match computed values from pipeline
• All existing tests still pass (no code changes in this PR)