A quantitative disorder-phenotype atlas of inherited metabolic disorders reveals shared symptom spaces and...

Inherited metabolic disorders (IMDs) are classified by pathway and gene, but their multisystemic presentation complicates phenotype-first diagnosis.

Using a frozen IEMbase export, we analyzed expert-curated disorder-level phenotype annotations for 1737 IMDs across 21 phenotype domains, 24 disorder groups, and 122 subgroups. We quantified annotation frequency, multi-domain burden, and group-domain enrichment using binomial z-scores, with clustering, principal component analysis, and bipartite networks for higher-order structure. Primary analyses were age-independent; age coverage was assessed exploratorily.

Disorders involved a mean of 4.01 ± 2.54 phenotypic categories. Neurologic involvement was annotated in 70.5%, followed by Eye (38.3%) and Other (34.7%). Enrichment identified hotspots, including endocrine metabolic disorders → Endocrine (z ≈ 12.6) and congenital disorders of glycosylation → Dysmorphic/Skeletal (z ≈ 10.3/8.0), with glycosylation and mitochondrial phenotype clouds. Age-specific information was available for 98.7% of disorders in at least one interval, but only 28.9% spanned all five.

This first-generation knowledgebase-level atlas quantifies reported disorder-level phenotype architecture rather than patient-level prevalence or natural history. It provides a scalable framework for phenotype standardization, diagnostic navigation, targeted curation, and future integration of age, variants, treatment, and real-world evidence.