Repository navigation
feat: proline synthesis penalty + helix bonus weight (PR #51) - #51
Merged
Merged
Conversation
- synthesis_feasibility_score: penalise pro_fraction > 0.15 by -0.10 (DKP formation risk at XP junctions during Fmoc SPPS; Barlos 1989, Quibell 1994, Fischer 2003) - activity_likeness_score: helix_bonus weight 0.01 → 0.03; score ceiling rises from 0.95 to 0.97 (helical AMPs represent ~70% of membrane-active families; Huang 2000, Tossi 2000) - AUROC improved 0.8138 → 0.8164 on retrospective benchmark - Add 6 proline-penalty tests to test_scoring_synthesis.py - Update test_scoring_activity.py docstring to reflect new helix weight - Makefile: bump test-suite count to 1164
- Add TestHelixBonusWeight with 3 numeric tests that pin the 0.03 helix bonus weight: exact delta at helix_pa=1.20, half-value at pa=1.10, and score ceiling assertion at 0.97 - Add test_proline_stacks_with_aggregation_propensity to cover combined agg+pro penalty interaction - Fix synthesis.py docstring: DKP formation is N-terminal dipeptide risk during piperidine deprotection, not exclusive to XP junction coupling - Makefile: bump test count to 1168
4 tasks done
cschanhniem
added a commit
that referenced
this pull request
Jun 28, 2026
- Add PR #51 to completed improvements header - Add detailed PR #51 section: proline synthesis penalty (DKP risk) and helix bonus weight 0.01→0.03 with literature citations - Update AUROC from 0.811/0.8138 → 0.8164 throughout - MIC gate updated ~60-70% → ~61-71% to reflect AUROC improvement - Breaking news probability updated ~27-47% → ~28-48% - Add item 15 to "What the Pipeline Got Right" - Update summary table, confidence calibration, roadmap gap references - Scoring model limitation: ~-5 pp → ~-4 pp (AUROC improved)
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
1 participant
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
synthesis_feasibility_scorenow deducts −0.10 whenproline_fraction > 0.15. XP junctions are prone to diketopiperazine formation during Fmoc SPPS; high-proline peptides require extended activation and show reduced coupling efficiency (Barlos 1989, Quibell 1994, Fischer 2003).activity_likeness_scorehelix_bonus weight raised from 0.01 → 0.03. Helical AMPs (LL-37, magainin, mastoparan) represent ~70% of membrane-active AMP families; this bonus now better distinguishes them from structurally disordered or beta-sheet peptides at otherwise equal composition scores (Huang 2000 Biochim Biophys Acta; Tossi 2000 Biopolymers).Changes
scoring/synthesis.pyscoring/activity.pytests/test_scoring_synthesis.pytests/test_scoring_activity.pyMakefileTest plan
>)proline_fractionkey defaults to 0.0 → backward compatible🤖 Generated with Claude Code