HGDiscovery

HGDiscovery predicts functional and phenotypic impacts of HGD missense variants to support interpretation of mutations associated with alkaptonuria.

Key Features:

• Training dataset: Contains a curated set of 111 pathogenic and 190 non-pathogenic HGD missense mutations.
• Structural and sequence analyses: Integrates protein structural information with sequence-based analyses to generate variant scores.
• Graph-based signatures (mCSM): Applies graph-based signature methods including mCSM to assess effects on protein stability and protein–protein interactions.
• Evolutionary conservation: Evaluates evolutionary conservation to inform functional impact assessments.
• Machine learning classifier: Trains a supervised machine learning classifier on structure- and sequence-derived scores to predict phenotypic outcomes of novel HGD missense mutations.
• Performance: The classifier achieved 84% accuracy on the provided dataset.

Scientific Applications:

• Variant interpretation: Predicts functional and phenotypic consequences of novel HGD missense variants for research and clinical variant interpretation.
• Disease mechanism investigation: Supports investigation of alkaptonuria mechanisms by indicating impacts on HGD stability, interactions, and conservation.
• Genetic diagnosis support: Provides evidence to assist genetic diagnosis of alkaptonuria-associated HGD variants.
• Therapeutic research: Informs therapeutic research by identifying mutations that alter HGD function or molecular interactions.

Methodology:

Integrates protein structural information with graph-based signature methods (including mCSM) and sequence-based analyses using scores derived from 111 pathogenic and 190 non-pathogenic HGD missense mutations, and trains a supervised machine learning classifier to predict phenotypic outcomes (84% accuracy reported).