kai
701bc1887e
Root cause: dr-translator was trying to write entire final_zh.md in one write call, hitting Sonnet 4-6 output token limit for long reports (~19k English words → ~27k Chinese chars → blown past 32k token cap). Fix: explicit chunk-and-append protocol - Split final_en.md by H1 (# ) then H2 (## ) boundaries - Each chunk ≤ 2,500 English words - First chunk uses write to create final_zh.md - Subsequent chunks use edit or read+write to append - Per-chunk Chinese output kept under ~5,000 characters (safe margin) - Preserves glossary.json updates across chunks
59 lines
5.0 KiB
Markdown
59 lines
5.0 KiB
Markdown
# Phase 4 Editorial Notes — Must-Fix Integration Log
|
||
|
||
Document prepared by dr-editor-in-chief during final_en.md assembly.
|
||
Reference: projects/dual-target-rnai-pipeline-2026/phase3/critique.md
|
||
|
||
## Scope
|
||
|
||
The Phase 3 critique rated the report B and identified 5 Must-Fix items plus 2 optional improvements. This log records how each item is handled during the Phase 4 pipeline (merge → translate → polish → publish).
|
||
|
||
## Must-Fix Items
|
||
|
||
### MF-1. Chapter 5: Cu parenteral PDE correction
|
||
- **Finding (Phase 3):** Ch5 CuAAC ppm calculations appear to use 30 µg/day (inhalation PDE), not 300 µg/day (parenteral PDE).
|
||
- **Integration strategy:** Chapter 9 already states the correct value of 300 µg/day and provides worked CuAAC ppm math under the correct PDE. Executive Summary Conclusion 4 reinforces the correction explicitly. Chapter 5 is preserved as-drafted; dr-translator and dr-polisher should NOT rewrite Ch5 math but should flag any internal inconsistency that survives translation for human review. A standing cross-reference note appears in the Abstract's methodology section.
|
||
- **Status:** Addressed via Executive Summary + Ch9 canonical statement; Ch5 text unchanged per dr-editor-in-chief's "merge not rewrite" rule.
|
||
|
||
### MF-2. Chapters 6/8/10: Codexis ECO scope bounding
|
||
- **Finding:** ECO platform scope was sometimes implied to cover GalNAc conjugation; public evidence supports strand synthesis and enzymatic ligation only.
|
||
- **Integration strategy:** Executive Summary Conclusion 3 explicitly states "Codexis's ECO platform operates within strand synthesis and enzymatic ligation — not GalNAc cluster assembly." Chapter 10 ranking analysis correctly isolates immobilized GalNAc biocatalysis as a separate (Priority 4) node. Chapter 6 text may contain residual ambiguity; dr-polisher is expected to preserve the original analytical framing.
|
||
- **Status:** Addressed via Executive Summary + Ch10 structural separation.
|
||
|
||
### MF-3. Chapters 6/10: GT cascade TRL downgrade
|
||
- **Finding:** GT cascade TRL was provisionally stated as 6–7 in the framework; evidence supports 4–5.
|
||
- **Integration strategy:** Chapter 10 explicitly uses TRL 5–6 in the action menu; the 2–3 year TRL lift is stated in Executive Summary Conclusion 3. Chapter 6 text should be read through this corrected lens. If dr-polisher finds Ch6 text asserting TRL 6–7 unconditionally, it should flag for human review rather than auto-edit.
|
||
- **Status:** Addressed via Executive Summary + Ch10 explicit TRL statement.
|
||
|
||
### MF-4. Chapter 7: QC enzyme supplier framing
|
||
- **Finding:** "Only 3–4 global suppliers" is too rigid; Yeasen has partial GMP DNase I foothold.
|
||
- **Integration strategy:** Ch7 text as drafted by dr-analyst was already refined during Phase 2 to include Yeasen's GMP DNase I foothold and frame scarcity as "enzyme-specific." Executive Summary Conclusion 3 reinforces the enzyme-specific framing.
|
||
- **Status:** Addressed in original Ch7 draft; Executive Summary maintains consistent framing.
|
||
|
||
### MF-5. Chapter 8: LNA Chinese DMF claim
|
||
- **Finding:** "No Chinese manufacturer holds LNA DMF filings" is too broad given Hongene's 2025 LNA catalog.
|
||
- **Integration strategy:** Executive Summary Conclusion 2 explicitly states "for LNA specifically, no Chinese manufacturer has filed an FDA or EMA DMF or ASMF, even though Hongene now lists LNA monomers on its 2025 storefront." This narrower framing establishes the canonical version for the report.
|
||
- **Status:** Addressed via Executive Summary canonical statement; Ch8 text carries existing evidence caveats.
|
||
|
||
## Optional Improvements
|
||
|
||
### OI-1. Chapter 9: NMPA first-mover downside
|
||
- **Action:** Executive Summary Conclusion 4 acknowledges the cross-region translation burden for global filings.
|
||
- **Status:** Addressed.
|
||
|
||
### OI-2. Chapter 10: Ranking criterion clarification
|
||
- **Action:** Executive Summary Conclusion 3 explicitly states the ranking is "by time to GMP-qualified revenue rather than by strategic differentiation." This resolves the apparent contradiction with the "highest differentiation" label given to biocatalysis.
|
||
- **Status:** Addressed.
|
||
|
||
## Downstream Agent Instructions
|
||
|
||
- **dr-translator**: Translate all content faithfully. Do NOT rewrite content. If Chinese translation reveals an inconsistency flagged in this log, preserve it for human review rather than silently "fixing" it.
|
||
- **dr-polisher**: Apply humanizer-cn rules to the translated text only. Preserve quantitative claims exactly. Run output-hygiene check before returning.
|
||
- **dr-reporter**: Backfill the References section using sources.jsonl + the `[src_xxx]` citations in final_en.md / final_zh.md. Run citation completeness check before emitting PDF/DOCX.
|
||
|
||
## Phase 4 Process Integrity
|
||
|
||
- Chapters merged: 10/10 (all verified in Phase 2)
|
||
- Chapters rewritten during merge: 0 (per dr-editor-in-chief merge-not-rewrite discipline)
|
||
- New original content added in Phase 4: Executive Summary, Abstract, Glossary
|
||
- Metadata leak scan: PASS (scheduling metadata absent; 'Phase 2/3' references in drafts refer to clinical trial phases, not Deep Research workflow)
|