Repository navigation
fix(docs+qc): correct SEED-006 Met synthesis error and close expert-pack gaps - #104
Merged
Merged
Conversation
…ack gaps
ALL 4 SEED-006 pilot variants inherit Met at position 9 from the
mastoparan-X template (INWKGIAAMAKKLL). The prior synthesis guidance
("No Cys, no Met — standard Fmoc SPPS") was incorrect and would have
caused the lab team to skip Nle substitution, receive oxidised batches,
and miss activity.
Changes:
- docs/WET_LAB_HANDOFF.md: SEED-006 synthesis section rewritten —
"No Met" (WRONG) replaced with Nle/M9Nle REQUIRED guidance,
inert-atmosphere storage, HPLC purity at receipt
- docs/ASSAY_PREREGISTRATION.md: SEED-006 Met-at-position-9 entry
added to the pre-registered synthesis-risk block (parallel to the
existing SEED-007 Met entry)
- docs/EXPERT_REVIEW_PACK.md: Pipeline Limitations table updated —
three new rows added: SEED-006 Met oxidation (all 4 variants, M9Nle
REQUIRED), N-terminal X-Pro DKP_RISK (SEED-008, 244 Da, Nα-Ac
REQUIRED, MS receipt check), short-peptide serum stability (6 pilots
< 14 AA, model calibrated for 18–30 AA); stale AUROC values (0.8420)
corrected to 0.7832 (expanded benchmark, CI₉₅: 0.72–0.84, n=191)
- tests/test_presynth_qc.py: regression guard —
test_seed006_all_pilots_have_met_at_position9 asserts MET×1 flag and
Nle recommendation fire for all 4 SEED-006 pilot variants; prevents
future edits from silently losing the Met detection on these sequences
Tests: 1338 passed (unchanged count — all existing tests green)
…#111 Fixes found during code review of fix/seed006-met-and-expert-pack-gaps: HIGH — arithmetic error: SHORT_PEPTIDE count "6 candidates" was wrong; parenthetical listed SEED-007 ×4 + SEED-003 ×2 + SEED-009 ×4 = 10, not 6. Corrected to "10 pilot candidates" in EXPERT_REVIEW_PACK.md. MEDIUM — ASSAY_PREREGISTRATION.md SEED-007 used "preferred" while the new SEED-006 entry used "REQUIRED" for the same Nle substitution risk. Aligned SEED-007 to "REQUIRED" for synthesis-protocol consistency. MEDIUM — test class mislocation: test_seed006_all_pilots_have_met_at_position9 was appended to TestProlineRichMediaFlag; moved to TestOxidationRisk alongside the other Met/Cys oxidation guards (test_met_triggers_oxidation_risk, etc.). MEDIUM — test name promised position-9 specificity but no assertion checked it. Added assert seq[8] == "M" per-sequence to guard against Met position drift. LOW — EXPERT_REVIEW_PACK.md SEED-006 limitation row said "store at −80°C"; corrected to "Store lyophilized at −80°C" to match WET_LAB_HANDOFF.md and ASSAY_PREREGISTRATION.md. Tests: 1338 passed (net count unchanged — 1 test moved, not added/removed).
5 tasks
cschanhniem
added a commit
that referenced
this pull request
Jun 29, 2026
…tor confirmation (#109) * feat(panel): integrate wave 0.5b top picks — SEED-020 external predictor confirmation External predictor screening (CAMPR4, AMPScanner v2, Macrel, AMPActiPred, HemoFinder, AntiCP2) of 23 wave 0.5b candidates confirmed SEED-020_VAR_004 (RLRIRVLKRLLK) and SEED-020_VAR_002 (KVRIRVLKRLLK) as the best safety-activity leads in the batch. Broad novelty check confirms both are NOVEL (< 50% similarity to the 72-AMP curated reference database; max similarity 41.7% vs REF-RLK-001 and 37.5% vs REF-LL37-003). Key external predictor metrics: - SEED-020_VAR_004: AMPScanner 0.9928, HemoFinder LOW, Non-AntiCP 0.29 (cleanest in batch) - SEED-020_VAR_002: AMPScanner 0.9960 (highest in batch), HemoFinder LOW, Non-AntiCP 0.32 Both outperform SEED-019_VAR_004 on the AntiCP off-target safety criterion while maintaining equivalent activity. Both are NOVEL — confirmed activity is publishable. Changes: - docs/WET_LAB_HANDOFF.md: new SEED-019/020 section — scaffold description, synthesis guidance (Nα-Ac REQUIRED for 5 trypsin sites, C-amide recommended, SHORT_PEPTIDE serum note, Macrel hemolysis false-positive warning), assay priority note (SAR triad comparison with SEED-019_VAR_004 within same assay plate) - docs/EXPERT_REVIEW_PACK.md: §5 family table extended with SEED-019/020 row (P(≥1 active) ~78–87%, best ext-predictor confirmed); wave 0.5b addendum block added below table summarising the confirmation rationale - docs/ROADMAP.md: PR tracking entries for PR #104 (SEED-006 Met fix) and this PR wave1_final_panel.csv updated locally (outputs/ is gitignored): SEED-020_VAR_004 and SEED-020_VAR_002 added as BALANCED_LEAD (WAVE0_5B). Tests: 1372 passed. * fix(review): address all HIGH/MEDIUM issues from code review of wave05b PR HIGH — SEED-019_VAR_004 classified NOVEL in new docs but is CLOSE_RELATIVE with POSSIBLE_PATENT_RISK in official novelty audit v2 (66.7% to DRAMP05502). Added explicit IP NOTE block in WET_LAB_HANDOFF.md; added patent risk flag to EXPERT_REVIEW_PACK.md §5 row and wave 0.5b addendum; publishability claim scoped to SEED-020 pair only. HIGH — Arg count/charge density errors: "≥4 Arg" doesn't apply to KVRIRVLKRLLK (3 Arg + 3 Lys = 6 cationic); net charge "+4 to +5" incorrect — should be "+5 to +6". Fixed to "3–4 Arg plus 2–3 Lys, 6 total cationic, net charge ~+5 to +6." HIGH — Nα-Ac rationale conflated exopeptidase + endopeptidase protection. Added explicit clarification: Nα-Ac blocks aminopeptidase (exopeptidase) from the free α-amine; trypsin endoproteolysis at 5 K/R sites requires D-amino Wave 2. MEDIUM — Motif descriptor removed (had positional errors: K at pos1 for VAR_002, V/L order mismatch for SEED-019); replaced with prose description of the pattern. MEDIUM — Macrel bias characterization changed from "all cationic helices" to "non-discriminating in this dataset — 60/60 candidates flagged regardless of class" (matches documented evidence in METRICS_CURRENT.md). MEDIUM — P(≥1 active) range adjusted from 78–87% to 75–85% in EXPERT_REVIEW_PACK §5; "best ext. predictor confirmed leads" clarification added to reviewer focus. LOW — Melittin cross-reference removed; now cites docs/METRICS_CURRENT.md. LOW — SEED-022 documented in addendum (excluded at pre-screen, no viable candidates after safety scoring). Tests: 1372 passed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
WET_LAB_HANDOFF.mdsaid "No Cys, no Met" — would have caused the lab team to skip Nle substitution, receive oxidised batches, and miss activity.test_seed006_all_pilots_have_met_at_position9assertsMET×1flag and Nle recommendation fire for all 4 SEED-006 pilot variants.Test plan
make test→ 1338 tests pass (verified locally)docs/WET_LAB_HANDOFF.mdSEED-006 section explicitly says M9Nle REQUIRED, −80 °C storage, HPLC puritydocs/ASSAY_PREREGISTRATION.md§5 bullet for SEED-006 present alongside SEED-007docs/EXPERT_REVIEW_PACK.mdPipeline Limitations table has SEED-006 Met row with correct AUROC (0.7832)🤖 Generated with Claude Code