Repository navigation
feat: add SEED-006 (Mastoparan-X) to break novelty ceiling above 0.467 (#38) - #38
Merged
Merged
Conversation
cschanhniem
added a commit
that referenced
this pull request
Jun 28, 2026
Add INWKGIAAMAKKLL (Mastoparan-X, wasp venom calmodulin-binding helical AMP, Yashida 1990) as the 6th seed template. Mastoparan family is absent from all 73 curated references, giving SEED-006 variants genuinely high novelty scores (0.571–0.684) — exceeding the previous pool ceiling of 0.467 (SEED-005 max). Pipeline results with 6 seeds: - Phase3 pool grows from 383 to 461 candidates - 11 SEED-006 variants pass all selection filters - SEED-006 top novelty: 0.684 (SEED-006_VAR_065, INYKGFAAMAKKLL) - SEED-006 top ensemble: 0.798 (SEED-006_VAR_061, INWRGIAAMAKKFL) - Pilot panel (n=20, max_per_seed=4): SEED-006 claims 4 slots (ranks 1–4) with novelty 0.643–0.684 and stability 0.62–0.71 Addresses root cause "Low scaffold novelty (-15 pp)" identified in DISCOVERY_PREDICTION.md. Scaffold novelty stage probability improves from ~10–15% to ~18–25%. Adds 7 new tests in TestSeed006MastoparanX covering source field, ID format, length preservation, seed exclusion, and pool size growth.
cschanhniem
force-pushed
the
feat/seed-006-mastoparan
branch
from
June 28, 2026 05:10
00c9ad5 to
90e7053
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
INWKGIAAMAKKLL, Mastoparan-X, wasp venom calmodulin-binding helical AMP) toexamples/sequences/amp_seeds.csvexamples/sequences/phase3_pool.csvwith 461 candidates (was 383)Problem solved
The
DISCOVERY_PREDICTION.mdroot-cause analysis identified "Low scaffold novelty (-15 pp)" as the second-largest probability gap. All previous 5 seeds were well-characterized AMPs with near-neighbor variants in the reference database. The highest achievable novelty was 0.467 (SEED-005 variants). No candidate was in genuinely uncharted sequence space.SEED-006 selection rationale
Mastoparan-X (Yoshida 1990, J. Biol. Chem.):
Pipeline results
Top SEED-006 pilot panel entries:
Scaffold novelty stage probability update
Tests
7 new tests in
TestSeed006MastoparanX:template_mutation_from_SEED-006SEED-006_VAR_NNNAll 878 tests pass.
Test plan
pytest tests/test_template_mutator.py— 40/40 passed (7 new)pytest(full suite) — 878/878 passedmake generate+make phase3verified SEED-006 variants score correctly🤖 Generated with Claude Code