Linking human diseases to animal models using ontology-based phenotype annotation.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Traditional phenotype descriptions use variable natural language, hindering gene-disease relationship discovery.
• A computationally tractable method is needed to mine mutant phenotypes from data resources.

Purpose of the Study:

• To test if ontological annotation of disease phenotypes can facilitate discovering new genotype-phenotype relationships within and across species.
• To establish a standardized, computable method for phenotype description and comparison.

Main Methods:

• Developed an Entity-Quality (EQ) methodology using ontologies to describe phenotypes.
• Annotated 11 human diseases from Online Mendelian Inheritance in Man (OMIM) using the EQ method.
• Integrated human disease annotations with model organism mutant phenotypes in an Ontology-Based Database (OBD).
• Employed four similarity metrics and a cross-species anatomical ontology for phenotype comparison.

Main Results:

• EQ-annotated phenotypes can be computationally compared based on ontological hierarchies and annotation frequency.
• Identified biologically meaningful similarities between genes by comparing phenotypes alone.
• Successfully identified other alleles of the same gene, pathway members, and orthologous genes across species.

Conclusions:

• EQ-based phenotype annotation, cross-species ontologies, and similarity metrics effectively identify gene similarities.
• This novel method efficiently identifies gene candidates and animal models for human diseases.
• The approach can shorten the path to understanding the genetic basis of human diseases.