Skip to content

feat: add SEED-007 (Bombolitin II) + SEED-008 (Puroindoline-a) for novel scaffold diversity (#42) - #41

Merged
cschanhniem merged 1 commit into
mainfrom
feat/seed-007-008-novel-scaffolds
Jun 28, 2026
Merged

cschanhniem merged 1 commit into
mainfrom
feat/seed-007-008-novel-scaffolds

Conversation

@cschanhniem

Copy link
Copy Markdown
Collaborator

Summary

  • SEED-007: Bombolitin II (IKITTMLKKLG) from Bombus terrestris bumblebee venom

    • Novelty: 0.615 vs 73-sequence reference set (higher than SEED-001–005, comparable to SEED-006)
    • Serum stability score: 0.636 (moderate-good, 2 interior trypsin sites)
    • Mechanism: membrane disruption via amphipathic helix (distinct from mastoparan SEED-006)
    • Literature: Argiolas & Pisano (1985) J Biol Chem
  • SEED-008: Puroindoline-a Trp domain (FPVTWRWWKWWKG) from Triticum aestivum wheat endosperm

    • Novelty: 0.667 vs reference set — HIGHEST of all 8 seeds
    • Serum stability: 0.385 (moderate risk from Trp-rich chymotrypsin sites; offset by high novelty)
    • Mechanism: Trp-mediated lipid bilayer disruption — genuinely different from helical AMPs
    • Literature: Dubreil et al. (1997) Plant J; Blochet et al. (1993) FEBS Lett
  • Pool regenerated: ~461 → 583 entries (8 seeds × ~73 variants each)

Impact on Discovery Probability

  • SEED-007 variants occupy 65 new pool positions with novelty ~0.58–0.65 (fills gap between SEED-005 and SEED-006)
  • SEED-008 variants occupy 57 new pool positions with novelty ~0.60–0.72 (highest-novelty cohort)
  • Pilot panel will select 4 candidates from each new seed family (max_per_seed=4)
  • Expected scaffold novelty gate improvement: ~18–25% → ~22–30%

Test plan

  • Pool regenerates without error (583 entries, 8 seeds)
  • SEED-007 variants present in pool (65 entries confirmed)
  • SEED-008 variants present in pool (57 entries confirmed)
  • All 944 tests pass

…vel scaffold diversity

- SEED-007: IKITTMLKKLG from Bombus terrestris bumblebee venom (Argiolas & Pisano 1985);
  novelty=0.615, serum_stability=0.636; amphipathic helix mechanism distinct from SEED-006
- SEED-008: FPVTWRWWKWWKG puroindoline-a Trp domain from Triticum aestivum (Dubreil et al. 1997);
  novelty=0.667 (highest of all 8 seeds); Trp-mediated lipid bilayer disruption
- Pool regenerated: 461 → 583 entries (8 seeds × ~73 variants each)
@cschanhniem
cschanhniem merged commit ef4e493 into main Jun 28, 2026
3 checks passed
@cschanhniem
cschanhniem deleted the feat/seed-007-008-novel-scaffolds branch June 28, 2026 05:37
cschanhniem added a commit that referenced this pull request Jun 28, 2026
…th QC (#41) (#42)

* feat: add N-terminal acetylation flag and D-amino Wave 2 guidance to presynth QC

- SynthQC gains n_acetylation_recommended/n_acetylation_reason fields:
  when interior trypsin sites exist, flag recommends Nα-Ac at zero
  synthesis cost to block aminopeptidase entry (Creighton 1993;
  Dhankhar et al. 2023)
- SynthQC gains wave2_d_substitutions list: up to 3 most-N-terminal
  interior K/R sites yield D-Lys/D-Arg substitution guidance for Wave 2
  synthesis (expected t½ gain 3–10×; Wade et al. 1990 PNAS)
- Both guidance flags appended AFTER synthesis_difficulty is computed
  (zero synthesis cost — must not inflate difficulty rating)
- to_dict() exposes all three new fields
- WET_LAB_HANDOFF.md: two new rows in the QC flags table
- 10 new tests in TestWave2Guidance; full 954-test suite green

* test: add missing edge-case coverage for Wave 2 guidance (C-terminal K/R, 1-indexed positions, empty seq)

- HIGH: lock in that a C-terminal-only K/R does NOT trigger N-acetylation flags (regression guard)
- MEDIUM: assert 1-indexed position strings in wave2_d_substitutions for RRWQWRMKKLG
- MEDIUM: empty-sequence smoke test (parallel to TestCTerminalAmidation)
- LOW: fix misleading comment in interior-trypsin-sites test
- LOW: assert n_acetylation_reason is non-empty and contains "acetylat"
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant