comp-012: DAF/CD55 SCR1-4 Truncated Construct — Shio-Koji Protease Stability¶
Question: Does the DAF/CD55 SCR1-4-only construct (aa 35–285, stalk removed) survive the shio-koji protease environment if expressed in A. oryzae?
Verdict: LOW
| Scope | Verdict | Max risk score | Worst protease |
|---|---|---|---|
| SCR1-4 only (aa 35–285; stalk aa 286–353 removed) | LOW | 0.039 | NPr |
The key finding: Truncating the Ser/Thr-rich stalk (aa 286–353) from the DAF/CD55 ectodomain drops the verdict from HIGH (comp-006, full ectodomain, max score 0.388) to LOW (comp-012, SCR1-4 only, max score 0.039). The stalk was the sole driver of the HIGH verdict — every exposed site across all three proteases (9 NPr sites, 48 ALP sites, 1 acid_protease site) was in the stalk. After truncation, zero exposed sites remain. All 242 recognition sites in the SCR1-4 construct (157 ALP + 60 NPr + 25 acid_protease) are buried (pLDDT ≥ 80, risk weight 0.1).
The SCR1-4 construct (aa 35–285) is structurally comparable to uricase (comp-001): mean pLDDT 96.7 vs. 97.1, min pLDDT 85.6 vs. 80.5, 100% residues pLDDT > 80 in both constructs. This is the benchmark for protease stability in the shio-koji platform.
Platform implication: A stalk-truncated DAF/CD55 SCR1-4 construct is the first computationally validated complement regulator candidate compatible with shio-koji fermentation. This closes the computational feasibility gate for the CP0 platform gap (complement priming via C5a). The next question — whether the expressed fragment retains complement-regulatory activity (C3b deposition inhibition, C3 convertase decay acceleration, C5a reduction in a gut-lumen assay) — is a wet-lab question.
Predecessor: comp-006 (full ectodomain, HIGH) — the HIGH verdict that motivated this follow-on
Informs: wiki/modality-chokepoint-matrix.md — "Engineered soluble complement regulators" row (CP0 platform gap)
Interpretive wiki page: wiki/daf-cd55-scr14-truncated-computational.md
How to reproduce¶
No external packages required (stdlib only: json, pathlib). Outputs land in outputs/.
The core algorithm lives in experiments/lib/protease_stability.py — shared with comp-001, comp-005, and comp-006. This script is the orchestrator; the library exports functions only.
Verify: Construct length analyzed should be 251 aa (aa 35–285 inclusive). Confirmed in terminal output on completion.
File index¶
comp-012-daf-cd55-scr14-truncated/
analyze.py ← analysis script (run this)
inputs/
P08174.fasta ← human DAF/CD55 sequence (copied from comp-006)
alphafold_P08174_plddt.json ← per-residue pLDDT scores (copied from comp-006)
protease_specificities.json ← koji protease rules + shio-koji conditions (copied from comp-006)
provenance.md ← sources + relationship to comp-006
outputs/ ← generated by analyze.py; committed as artifacts
cleavage_sites.json ← machine-readable full results
summary.md ← human-readable; cited in the wiki
README.md ← this file
Key results¶
| Protease | Recognition sites (SCR1-4) | Buried | Partially exposed | Exposed | Max risk score |
|---|---|---|---|---|---|
| ALP (alkaline subtilisin) | 157 | 157 | 0 | 0 | 0.019 |
| NPr (neutral metalloprotease) | 60 | 60 | 0 | 0 | 0.039 |
| acid_protease (aspergillopepsin) | 25 | 25 | 0 | 0 | 0.020 |
Zero exposed sites in the SCR1-4 construct across all three proteases. The verdict floor for buried-only sites is risk score = accessibility_weight × effective_activity = 0.1 × 0.388 (NPr) = 0.039 — which is the exact max score observed.
Compare with comp-006 (full ectodomain, NPr): 9 exposed stalk sites driving max risk score 0.388.
Disagreement protocol¶
If you reproduce the outputs and disagree with the methods or numbers, file a GitHub issue referencing this folder (comp-012-daf-cd55-scr14-truncated). Primary candidates for revision: truncation boundary (aa 285 per UniProt P08174 SV=4 SCR4 C-terminus; verify against current UniProt entry), disulfide bond correction factor, or protease pH factors.