Klarigi

Klarigi provides multivariable semantic explanations for groups of biomedical entities annotated with ontology classes, combining compositional and discriminatory scoring to characterize and distinguish entity groups.

Key Features:

• Multivariable Semantic Explanations: Identifies sets of ontology classes that jointly characterize groups of entities rather than relying on univariate associations.
• Scoring Heuristics: Applies multiple scoring heuristics to assess both compositional (characteristic) and discriminatory importance of ontology classes within groups.
• Ontology Reasoner Integration: Performs live semantic inference using an ontology reasoner to derive explanations grounded in formal semantics.
• Discriminatory Power Identification: Implements a classification-based method to evaluate and identify candidate sets of ontology classes that distinguish between groups.
• Novel Algorithm for Multivariable Explanations: Uses a bespoke algorithm tailored to derive multivariable semantic explanations from annotated entity groups.

Scientific Applications:

• Gene expression analysis: Characterizes and distinguishes groups of gene expression profiles using ontology-based annotations.
• Phenotype analysis: Provides characteristic and discriminatory explanations for patient phenotypes annotated with ontology classes.
• Interpretation of complex biomedical datasets: Produces semantic explanations that complement traditional enrichment analyses to aid interpretation of annotated biomedical entity groups.

Methodology:

Implements a novel algorithm for multivariable semantic explanations; employs multiple scoring heuristics to assess compositional and discriminatory aspects of ontology classes; integrates an ontology reasoner for live semantic inference; and was evaluated using two test cases.