docs: track PR #110 — expanded benchmark + DKP flag (SEED-008 synthesis note)

docs/DISCOVERY_PREDICTION.md

@@ -854,3 +854,14 @@ proline-rich, RPMI-1640 requirement restated, Pro-bond serine-protease resistanc
 Serum stability section header updated to "All 7 Seeds" with SEED-006 and SEED-007 rows
 added to the per-seed table. All 7 families now have equal documentation depth in the
 wet-lab handoff package.
+
+**Tracking note (post-PR #110):** Benchmark expanded from 43+44 to 95 AMPs + 96 decoys
+(n=191). AUROC updates: pipeline.yaml 0.8420 → 0.7832 (CI₉₅: 0.72–0.84); phase3.yaml
+0.8266 → 0.7448. The lower point estimate on the larger, class-diverse benchmark is
+expected — the helix-centric scorer was not calibrated for defensins, proline-rich peptides,
+or lantibiotics. The synthesis gate (AUROC > 0.70) still holds with comfortable margin.
+DKP_RISK QC flag added: all 4 SEED-008 pilots (F-Pro N-terminus) require Nα-acetylation
+in the synthesis order to prevent cyclo(F-Pro) (MW≈244 Da) truncation. ASSAY_PREREGISTRATION
+and WET_LAB_HANDOFF SEED-008 sections updated with REQUIRED Nα-Ac guidance and MS receipt
+rejection criterion (satellite > 5% → reject batch). EXPERT_REVIEW_PACK and
+DISCOVERY_PREDICTION doc corruption artifacts corrected. 1337 tests.

docs/ROADMAP.md

@@ -94,6 +94,7 @@ Implemented during the pre-wet-lab improvement loop (PRs #31–#54):
 - test(qc): SEED-009_VAR_033 regression guard — asserts both MET×1 (Nle) and PROLINE_RICH_INTRACELLULAR fire simultaneously for RRLPRPGYMPRP; 1312 tests total (PR #107)
 - EXPERT_REVIEW_PACK.md: Pipeline Limitations table updated — PYROGLUTAMATE_RISK row added (N-terminal Q, physiological pH, none of 20 pilot candidates affected); SEED-009_VAR_033 Met row added (PR #108)
 - test: close remaining 1% coverage gap — 6 modules to 100% branch coverage; 1321 tests total; only 6 structural CLI guard lines remain uncovered; all source modules at 100% branch coverage (PR #109)
+- feat(bench+qc): expanded benchmark (95 AMPs + 96 decoys, n=191, AUROC=0.7832, CI₉₅: 0.72–0.84); 52 new public-domain AMPs across 12 classes (defensins, proline-rich, lantibiotics); DKP_RISK flag for N-terminal X-Pro diketopiperazine cyclization — all 4 SEED-008 pilots (F-Pro) flagged with dynamic mass (244 Da), MS receipt check, Nα-Ac REQUIRED; WET_LAB_HANDOFF.md SEED-008 synthesis guidance updated; ASSAY_PREREGISTRATION.md SEED-008 DKP note added; 1337 tests (PR #110)
 
 ## v1.0 — Validated dry-lab-to-wet-lab loop
 

docs/WET_LAB_HANDOFF.md

@@ -369,7 +369,16 @@ correct** for this non-helical AMP class. See the Helix-Wheel Amphipathic Face A
 sites are cut 3–8× slower (Wu & Ding 2016, *J Pept Sci*). Assay before committing to D-Trp Wave 2.
 
 **Synthesis guidance:**
-- No Met, no Cys → standard Fmoc SPPS; no special modifications.
+- **⚠️ ALL 4 SEED-008 pilots start with F-Pro: Nα-acetylation (Ac-) REQUIRED** on the synthesis
+  order. The N-terminal Phe-Pro dipeptide spontaneously cyclizes to a diketopiperazine (DKP,
+  cyclo(F-Pro), MW≈244 Da) at physiological pH (t½ hours–days at 37°C). DKP formation truncates
+  the peptide at position 3 (losing the key Phe residue), likely abolishing activity.
+  Nα-acetylation (zero extra cost — specify "Nα-acetyl / Ac-" in the order form) blocks the
+  free amine completely.
+- **MS identity check mandatory at receipt:** confirm full-length peptide by ESI-MS or MALDI;
+  a peak at MW−244 Da indicates DKP contamination — reject the batch if satellite exceeds 5%
+  relative peak area. (Literature: Smyth *et al.* 2000 *J Peptide Res* 55:105.)
+- No Met, no Cys in VAR_009/VAR_019/VAR_044 (3/4 pilots) → standard Fmoc SPPS otherwise.
 - High Trp content (3–4 Trp/13 AA): dissolve in DMSO (≤ 10% final), then dilute to assay
   concentration. Use 0.01% BSA to prevent Trp-driven tube adhesion.