PhenoMiner

PhenoMiner extracts and analyzes phenotypes from scientific literature to identify and harmonize phenotype descriptions and establish phenotype–disease associations for Mendelian disorders.

Key Features:

• Full parsing and conceptual analysis: Uses full parsing and conceptual analysis techniques to systematically identify and harmonize phenotype descriptions across heterogeneous text.
• Apriori association mining: Applies Apriori association mining to derive relationships and hypotheses linking phenotypes to human diseases.
• Semantic Distribution Analysis: Evaluates the semantic distribution of extracted terms against linked ontologies to characterize term usage.
• Comparison with Human Phenotype Ontology (HP): Assesses term overlap and consistency with the Human Phenotype Ontology (HP).
• Support assessment in OMIM and literature: Quantifies support for phenotype–disorder pairs within OMIM and the broader literature, reporting moderate agreement where observed.
• Association with known disease–gene pairs via PhenoDigm: Links phenotype–disorder pairs to known disease–gene pairs using the PhenoDigm framework and reports strong associations for many pairs.
• Large-scale extraction results: Identified 13,636 candidate phenotypes and produced 28,155 phenotype–disorder hypotheses encompassing 4,898 distinct phenotypes and 1,659 Mendelian disorders.
• Output generation: Produces comprehensive outputs including lists of identified phenotypes, phenotype–disorder associations, and association-filtered linked data.

Scientific Applications:

• Phenotype discovery at scale: Enables large-scale extraction of candidate phenotypes from literature collections such as the BMC open access collection.
• Ontology mapping and harmonization: Supports mapping of text-mined phenotype mentions to ontologies like HP for ontology-driven curation.
• Hypothesis generation for Mendelian disorders: Generates phenotype–disorder hypotheses to support discovery of novel phenotype–disease relationships in Mendelian disorders.
• Cross-referencing phenotype–gene relationships: Facilitates cross-referencing of phenotype–disorder associations with known disease–gene pairs via PhenoDigm for genetic research.

Methodology:

Applied full parsing and conceptual analysis to the BMC open access collection; performed Apriori association mining to derive phenotype–disorder hypotheses; conducted semantic distribution analysis and term-overlap assessment with the Human Phenotype Ontology (HP); and linked phenotype–disorder pairs to known disease–gene pairs using the PhenoDigm framework.