Update evidence IDs in Fanconi Anemia

[caufieldjh]

The Fanconi Anemia entry contains evidence statements linked to specific files, with file: locations.
Example:

These don't resolve to anything and should be replaced with resolvable IDs.

[github-actions]

Issue Summary

The Fanconi Anemia entry (kb/disorders/Fanconi_Anemia.yaml) contains 69 evidence items with non-resolvable file: references pointing to:

file:research/Fanconi_Anemia_Clinical_Care_Guidelines_5thEdition_web.extracted.md

These references need to be replaced with resolvable identifiers that comply with the dismech reference validation system.

Background

Source Document

The referenced file exists in the repository at /research/Fanconi_Anemia_Clinical_Care_Guidelines_5thEdition_web.extracted.md and contains:

• Title: Fanconi Anemia Clinical Care Guidelines, 5th Edition (2020)
• Publisher: Fanconi Anemia Research Fund (FARF)
• Source URL: https://fanconi.org/wp-content/uploads/2024/01/Fanconi_Anemia_Clinical_Care_Guidelines_5thEdition_web.pdf
• Format: PDF (276 pages) extracted to markdown
• Editors: Isis Sroka, Lynn Frohnmayer, Sherri Van Ravenhorst, Leanne Wirkkula

This is a clinical care guideline document published by a patient advocacy organization, not a peer-reviewed journal article. It does not have a PMID or DOI.

Current Reference Schema

According to src/dismech/schema/dismech.yaml:797-802, the reference field expects:

• Type: PMID
• Example: PMID:35533128
• Range: PMID (base: str, uri: xsd:string)

However, other prefixes are supported throughout the schema including:

• clinicaltrials: (for clinical trial references)
• geo:, arrayexpress:, dbgap:, etc. (for datasets)
• phenopacket-store:, synapse:, etc.

Affected Evidence Items

The 69 file references are distributed across phenotype descriptors for:

• Hematologic manifestations (thrombocytopenia, neutropenia, anemia)
• Skeletal anomalies (radial ray defects, thumb abnormalities, vertebral anomalies)
• Facial features (micrognathia, epicanthus, abnormal ears)
• Cardiac defects (VSD, ASD, PDA, coarctation)
• Genitourinary abnormalities
• CNS malformations
• Skin pigmentation changes
• And many other congenital anomalies

All of these evidence items have evidence_source: HUMAN_CLINICAL and supports: SUPPORT, but lack the required snippet and explanation fields that would normally accompany PMID references.

Recommended Solutions

Option 1: Replace with Primary Literature (Preferred)

Each file reference should be replaced with PMID(s) from peer-reviewed publications that support the specific phenotype association. For example:

• GeneReviews FA entry: PMID:20301575 (comprehensive review, regularly updated)
• FA cohort studies with specific phenotype frequencies
• Individual case series for rare manifestations

This approach ensures:

• ✅ References are externally resolvable
• ✅ Evidence can be validated via just validate-references
• ✅ Snippets can be extracted from PubMed abstracts
• ✅ Complies with dismech SOP for evidence (see CLAUDE.md)

Option 2: Create a Custom Prefix for Guidelines

Add a new prefix to the schema for clinical guidelines:

prefixes:
  fanconi-guidelines: https://fanconi.org/wp-content/uploads/2024/01/

Then use references like:

reference: fanconi-guidelines:Fanconi_Anemia_Clinical_Care_Guidelines_5thEdition_web.pdf

However, this approach:

• ❌ Creates a precedent for citing non-peer-reviewed materials
• ❌ Still requires manual snippet validation (no automated checking)
• ❌ May not align with dismech's evidence standards

Option 3: Move to Notes Field

For phenotypes that are well-established but lack strong individual publication support, the information could be moved from evidence blocks to the notes field, which doesn't require references:

phenotypes:
- preferred_term: Thrombocytopenia
  term:
    id: HP:0001873
    label: Thrombocytopenia
  notes: Listed in FA Clinical Care Guidelines (5th Ed, 2020) as a common manifestation of bone marrow failure. Present in approximately X% of FA patients.

Next Steps

Recommendation: Use Option 1 - systematically replace file references with primary literature PMIDs.

This would involve:

1. Reviewing each of the 69 affected phenotypes
2. Searching PubMed for supporting evidence (starting with PMID:20301575 - GeneReviews)
3. Extracting appropriate snippets from abstracts
4. Replacing the file references with proper PMID references
5. Running just validate-references kb/disorders/Fanconi_Anemia.yaml to verify

This is a significant curation task that could benefit from semi-automated tooling or delegation to @dragon-ai-agent with appropriate supervision.

---

Related Files:

• /kb/disorders/Fanconi_Anemia.yaml (lines 215, 912, 928, 1188, 1199, 1215, 1231, 1247, 1263, 1279, 1295, 1427, 1443, 1459, 1475, 1491, 1507, 1523, 1539, 1555, 1571, 1587, 1603, 1625, 1654, 1670, 1686, 1702, 1718, 1734, 1750, 1766, 1782, 1798, 1814, 1830, 1846, 1862, 1913, 1929, 1945, 1961, 1992, 2021, 2037, 2053, 2069, 2085, 2101, 2130, 2146, 2162, 2178, 2194, 2426, 2442, 2458, 2474, 2490, 2506, 2522, 2538, 2554, 2570, 2586, 2616, 2746, 2762, 2797)
• /research/Fanconi_Anemia_Clinical_Care_Guidelines_5thEdition_web.extracted.md