Fanconi Anemia Phenotype Gap Analysis

Overview

Connelly E, Laraway B, Mullen KR, Mungall CJ, Haendel MA, Hurwitz EG. A custom phenotypic profile for Fanconi anemia: Addressing gaps in existing disease annotations. medRxiv, 2026. PMID:41728284, PMCID:PMC12919158, DOI:10.64898/2026.02.10.26346018.

The paper publishes a 264-term custom HPO profile for Fanconi anemia, curated from the Fanconi Cancer Foundation Clinical Care Guidelines (5th Edition, 2020) via OntoGPT extraction plus manual review, and shows existing OMIM/Orphanet annotations under-represent musculoskeletal, genitourinary, and limb phenotypes. This project diffs that profile against the dismech Fanconi_Anemia entry and tracks closing the gap with snippet-verified evidence.

Scope and Method

• Profile data (CC-BY-4.0): https://github.com/ehurwitz/FA-custom-profile (FA-custom-profile.hpoa, HPO release 2026-02-16) — mirrored as custom_profile.hpoa below.
• Extracted the 264 custom-profile HP IDs and the 446-term OMIM/Orphanet/custom comparison table (FA-custom-profile.xlsx) with anatomical-system labels.
• Diffed against the HP IDs already present in kb/disorders/Fanconi_Anemia.yaml.
• Evidence policy: the .hpoa annotations cite the FCF guideline URL (evidence code TAS), which does not satisfy the dismech PMID/structured snippet policy. Every added term is independently sourced and snippet-verified against ORPHA:84 (Orphanet structured record) or a fetched PubMed abstract.

Result (at project start)

Many of the 203 are finer-grained children of terms Fanconi_Anemia already carries (e.g. dismech has generic "Radial Ray Defects"; the profile enumerates absent radius/scaphoid/trapezium), so the gap is partly granularity, not pure absence.

Catalog Artifacts

Batch 1 — added (raises Fanconi_Anemia from 98 → 115 HP terms)

Prioritised the paper's emphasis systems. Sources are deterministic and snippet-verified (see batch1_added.tsv).

• ORPHA:84 (exact phenotype-table row substrings): Triphalangeal thumb (Triphalangeal thumb), Spina bifida (HP:0002414), Tetralogy of Fallot (HP:0001636), High palate (HP:0000218), Frontal bossing (HP:0002007), Short palpebral fissure (HP:0012745), Azoospermia (Azoospermia), Hydroureter (HP:0000072), Hyperreflexia (HP:0001347).
• PMID:19622403 (Auerbach AD, Fanconi anemia and its diagnosis, Mutat Res 2009 — IFAR series; exact-substring quotes): Microtia (Microtia), Low-set ears (HP:0000369), Posteriorly rotated ears (HP:0000358), Fusion of middle ear ossicles (HP:0005473), Aplastic clavicle (HP:0006660), Oligozoospermia (HP:0000798), Obesity (HP:0001513), Abnormal circulating lipid concentration (HP:0003119).

Batch 2 — added (raises Fanconi_Anemia from 115 → 147 HP terms)

Curating phenotypes as recorded in the authoritative Orphanet record (ORPHA:84): every uncovered phenotype-table row was added, skipping only top-of-hierarchy "Abnormality of the [system]" umbrellas, exact concept duplicates of existing entries, and one clinically implausible Orphanet annotation (pyridoxine-responsive sideroblastic anemia — not characteristic of FA, likely a mapping artifact). All 32 carry an exact ORPHA:84 row snippet (see batch2_added.tsv). Spans renal/GU (recurrent UTI, renal insufficiency/hypoplasia, abnormal kidney position, decreased male fertility, preputial anomaly), craniofacial (dolichocephaly, facial asymmetry, sloping forehead, choanal atresia, uvular hypoplasia), ocular (astigmatism, visual impairment, proptosis, upslanted fissures, nystagmus, iris hypoplasia), cardiac/ vascular (HCM, carotid anomaly, AVM), skeletal/limb (hip dislocation, finger syndactyly, finger aplasia/hypoplasia, pes planus, toe clubbing), GI (umbilical hernia, Meckel diverticulum, Hirschsprung, duodenal stenosis), plus cranial nerve paralysis, weight loss, and oligohydramnios.

Cross-source finding: none of these 32 Orphanet-recorded phenotypes appear in the Connelly 264-term custom profile — a direct illustration of the paper's thesis that no single source is complete. The custom profile and the Orphanet record are complementary, not nested.

Remaining work — ~186 unmapped profile terms

Counts per anatomical system (from missing_terms_by_system.tsv, status TODO):

Next batches (proposed)

• [x] Batch 2 — Orphanet-record sweep (incl. renal/genitourinary) — done: 32 ORPHA:84-recorded phenotypes added (recurrent UTI, renal insufficiency, renal hypoplasia/aplasia, abnormal kidney position, decreased male fertility, preputial anomaly, plus craniofacial/ocular/cardiac/skeletal/GI rows).
• [ ] Full-text-only structural specifics (horseshoe/ectopic kidney, hydronephrosis, chordee, phimosis): documented only in full-text results, not snippet-validatable against abstract-only cache — blocked by #4884.
• [ ] Batch 3 — radial-ray granularity (absent radius/scaphoid/trapezium, thumb-hypoplasia grading, carpal hypoplasia, ulnar bowing): map to the existing generic radial-ray nodes.
• [ ] Batch 4 — oral-mucosal / head-and-neck (oral ulcer, gingivitis, periodontitis, xerostomia, leukoplakia-adjacent): relevant to the FA head-and-neck SCC surveillance context.
• [ ] Batch 5 — GI-functional and metabolic (GERD, dysphagia, constipation, diabetes mellitus, insulin resistance).
• [ ] Re-run the diff after each batch and update missing_terms_by_system.tsv status flags.

Notes

• Every future addition must carry a verified ORPHA:/PMID: snippet; do not import the OntoGPT TAS annotations as evidence.
• Watch the Fanconi_Anemia vs Fanconi renotubular syndrome named-entity confusion when sourcing literature (the renotubular disorder shares "Fanconi" but has an unrelated phenotype set).