Skip to content

Wiki Rules

Purpose

These rules adapt the research-wiki-maintainer principle to the ILC_in_lung wiki.

Layer Rules

RAW/

  • RAW/ is the incoming source layer.
  • Successfully ingested PDFs are moved to RAW/processed/ because this was explicitly requested for this project.
  • Do not overwrite, rename, or alter processed PDFs.

wiki/

  • wiki/ contains source-aware markdown pages.
  • Every durable biological claim should be traceable to one or more source pages.
  • Bulk-generated source pages are provisional knowledge-oriented notes until manually reviewed. They may record title, abstract, early extracted-text, and tag-derived clues, but they should not be treated as fully verified biological summaries.

Ingest Modes

The ILC_in_lung wiki formally distinguishes two ingest modes. Every source page should state its current mode.

Provisional Bulk-Ingest Mode

Use this mode for batch intake of new PDFs.

Allowed outputs:

  • source registration in wiki/sources/
  • full-text extraction into private extracted text artifacts omitted from public export
  • conservative evidence tags
  • provisional source notes from title, abstract, early extracted text, and automated signals
  • manifest, index, and log updates
  • movement of successfully processed PDFs to RAW/processed/

Forbidden outputs:

  • durable biological synthesis in topic, entity, project, or digest pages
  • high-confidence mechanistic claims based only on automated extraction
  • claim-level confidence statements about file presence, text extraction, source-page creation, or tag usefulness

Focused Manual Crystallization Mode

Use this mode for priority papers that should change the reusable knowledge layer.

Required work:

  • read the source page and extracted text
  • check figures, methods, model system, species, tissue, time point, perturbation, and assay context when available
  • compare against related topic, entity, project, and digest pages
  • state observations, interpretations, caveats, contradictions, and confidence levels

Allowed outputs:

  • upgraded source pages with source-supported findings
  • updates to topic/entity/project/digest pages
  • high-confidence claim sinking into entity hubs when justified
  • contradiction and supersession tracking
  • audit pages for broad reinterpretation or maintenance-rule changes

Required Scientific Discipline

  • Do not invent biology, metadata, methods, or mechanistic claims.
  • Label assumptions explicitly.
  • Distinguish observation, interpretation, and hypothesis.
  • Distinguish mouse vs human when relevant.
  • Distinguish lung, gut, skin, nervous-system, and systemic contexts.
  • Distinguish transcript, protein, cytokine secretion, flow phenotype, and functional evidence.
  • Distinguish in vivo, ex vivo, and in vitro evidence.
  • Record contradictions rather than silently harmonizing them.
  • Use Claim-Level Confidence only for source-supported biological or knowledge claims, never for PDF presence, text extraction, tag detection, source-page creation, or ingest completeness.
  • If no reusable biological claim has been reviewed yet, write Not assigned rather than assigning low confidence because a page was generated automatically.

Preferred Page Families

  • sources/: one page per ingested PDF
  • topics/: cross-source synthesis pages
  • entities/: reusable cell, cytokine, receptor, pathway, or method pages
  • projects/: project-level question and working model pages
  • digests/: crystallized cross-source working models
  • audit/: schema changes, broad re-reviews, or reviewer-triggered corrections

Knowledge-Page Naming Style Guide

This naming guide applies to reader-facing knowledge pages in:

  • entities/
  • topics/
  • digests/

It does not apply to audit/, log.md, or most sources/ pages, where workflow history and maintenance language are expected.

Core naming principles

  • Use biology-first, reader-facing headings.
  • Use section names that describe what the reader will learn, not what the curator did.
  • Keep heading vocabulary stable across pages so the wiki feels navigable rather than improvised.
  • Use Title Case for ## section headings.
  • Put workflow provenance, batch history, migration notes, and maintenance details in audit/ or log.md, not in the main section structure of knowledge pages.

Preferred shared heading vocabulary

Use these names whenever they fit the page:

  • Scope
  • Evidence Tags
  • Confidence Snapshot
  • Established Observations
  • Interpretation
  • Open Questions
  • Related Pages
  • Future Expansion Directions

Use these when a page needs stronger synthesis or navigation structure:

  • At A Glance
  • How To Use This Page
  • Integrated Working Model
  • Review Map
  • Major Biological Branches
  • Regulatory Architecture
  • Claim-Level Confidence Boundaries
  • Interpretation Guardrails
  • Reading Routes
  • Core Claims
  • Evidence Layers
  • Interpretation Boundaries
  • How This Companion Fits The Wiki
  • How To Use This Digest
  • When To Revisit This Page
  • Representative Source Spine
  • Disease-Oriented Reading Guide
  • Conceptual Timeline
  • Knowledge Evolution Flowchart

Heading names to avoid in knowledge pages

Avoid these as section headings in entities/, topics/, and digests/ unless the page itself is explicitly about workflow or schema:

  • Batch
  • Update
  • Update Triggers
  • Next Ingest Targets
  • Crystallization
  • Focused Manual Crystallization
  • Working-Model Source Spine
  • Migration Notes
  • Rename Map
  • TODO
  • Task List

These terms may still appear in:

  • audit/
  • log.md
  • source-page ingest metadata
  • guide pages that explicitly explain ingest modes or curation workflow

entities/

Recommended order:

  1. Scope
  2. Evidence Tags
  3. At A Glance
  4. How To Use This Page
  5. Confidence Snapshot
  6. Integrated Working Model
  7. Review Map
  8. Major Biological Branches
  9. Regulatory Architecture
  10. Claim-Level Confidence Boundaries
  11. Interpretation Guardrails
  12. Contradiction And Supersession
  13. Open Questions
  14. Reading Routes
  15. Related Pages
  16. Future Expansion Directions

topics/

Recommended order:

  1. Scope
  2. Evidence Tags
  3. Confidence Snapshot
  4. Established Observations
  5. Interpretation
  6. Contradiction And Supersession
  7. Open Questions
  8. Related Pages
  9. Future Expansion Directions

Allow narrower subheadings under Established Observations when they reflect real biological branches, for example Metabolic Regulation or Glucocorticoid Resistance And Inflammatory Signaling.

digests/

Digest pages can vary more, but they should still use reader-facing section names. A typical digest should include most of the following:

  1. Scope
  2. Evidence Tags
  3. one orienting synthesis section such as Working Model, Disease-First Map, or Conceptual Timeline
  4. one evidence-organization section such as Core Claims, Disease-Oriented Reading Guide, or How Understanding Changed Over Time
  5. one caution or evidence-structure section such as Evidence Layers, Interpretation Boundaries, or Claim-Level Confidence Boundaries
  6. one navigation section such as How To Use This Digest or How This Companion Fits The Wiki
  7. When To Revisit This Page
  8. Representative Source Spine

Provenance placement rule

When a reader-facing knowledge page needs provenance:

  • keep provenance at the bottom of the page
  • use Representative Source Spine rather than workflow-heavy wording
  • do not place provenance sections before the main biological interpretation

Maintenance-language rule

If a section mainly tells the reader what evidence would strengthen the page later, use Future Expansion Directions.

If a section mainly tells the reader when the synthesis should be reconsidered, use When To Revisit This Page.

If the content is really curator workflow rather than reader guidance, move it to audit/ or log.md instead of keeping it in a knowledge page.

Evidence Tags

Preferred tag families include:

  • source type: #source/literature_pdf #source/primary #source/review
  • species: #species/mouse #species/human #species/mixed
  • tissue/context: #tissue/lung #tissue/gut #tissue/skin #tissue/nasal_polyp #tissue/systemic
  • cell focus: #cell/ILC1 #cell/ILC2 #cell/ILC3 #cell/NK #cell/macrophage #cell/monocyte #cell/T_cell #cell/B_cell
  • assay: #assay/flow #assay/scRNAseq #assay/RNAseq #assay/in_vivo #assay/in_vitro #assay/KO
  • outcome: #outcome/inflammation #outcome/infection #outcome/repair #outcome/homeostasis #outcome/airway_hyperresponsiveness
  • project axis: #axis/ILC_lung_homeostasis #axis/ILC_lung_infection #axis/ILC_airway_inflammation #axis/ILC_plasticity

Topic, entity, and digest pages should store tags in two compatible layers:

  • YAML frontmatter tags: without the leading #, so Obsidian can index them as page tags.
  • A visible ## Evidence tags section using inline-code tags, for example `#cell/ILC2` `#tissue/lung`.

Do not place raw #tag text at the start of a line in page body text. MkDocs/Python-Markdown interprets a line-start #tag as a heading, causing oversized browser rendering and incorrect document structure.

Ingest Rules

  1. Confirm the source exists in RAW/.
  2. Choose and record the ingest mode.
  3. Extract full text into private extracted text artifacts omitted from public export.
  4. Create one source page in wiki/sources/.
  5. Use conservative tags for automated pages, but keep Claim-Level Confidence claim-centered. Do not assign low confidence merely because the page came from automated extraction.
  6. Move the processed PDF to RAW/processed/ only after source-page and text extraction succeed.
  7. Update index.md, log.md, and the manifest.
  8. Do not update durable synthesis pages from provisional bulk ingest alone.

Manual Review Rules

Before promoting claims into topic, entity, digest, or project pages:

  1. Read the source page and extracted text.
  2. Verify species, tissue, model, intervention, and assay.
  3. Add source-supported findings with confidence levels.
  4. Note contradictions or context boundaries.
  5. Update related topic/entity/project pages only when justified.