IEMbase

IEMbase provides an expert-curated knowledge base and a computational diagnostic support system to match biochemical and clinical phenotypic profiles to inborn errors of metabolism.

Key Features:

• Expert-Curated Knowledge Base: Contains 530 well-defined profiles of biochemical markers and clinical symptoms associated with specific inborn errors of metabolism.
• Diagnostic Support System: Implements a mini-expert system using cosine similarity and semantic similarity to match user-provided phenotypic profiles to candidate diagnoses or genes, with reported performance of 62% exact diagnosis match and 86% within the top five candidates.
• Biochemical Annotation: Incorporates biochemical features in annotations, yielding a 41% higher rate of exact phenotype matches compared with clinical features alone.
• Phenomics Integration: Integrates biochemical phenotypes into phenomics resources including the Human Phenotype Ontology to support improved gene–disease relationship predictions.
• Community-Expandable Descriptions: Supports expansion of computationally accessible descriptions of biochemical phenotypes through contributions from the clinical and research community.

Scientific Applications:

• Clinical Diagnostic Interpretation: Matching biochemical and clinical profiles to prioritize candidate IEM diagnoses for genetic diagnostic centers and clinical teams.
• Gene–Disease Prediction: Using integrated biochemical phenotypes and HPO mappings to inform gene–disease relationship predictions in research studies.
• Phenomics Resource Development: Enabling expansion of computable biochemical phenotype descriptions for phenomics projects and computational phenotype analysis.

Methodology:

IEMbase combines user-inputted biochemical and clinical symptoms with its curated disease profiles and computes cosine similarity and semantic similarity between input and stored profiles to generate a ranked list of candidate IEMs, providing explanations for results and suggesting additional tests.