Ontological Descriptions of Phenotypes, with a Case Study in Maize

The phenotypes linked to classic gene models traditionally have been described quite casually, using free text. These descriptions may be informative to human readers but are inefficient for comparing across many genotypes and phenotypes, and they may be completely unusable for computerized comparisons. Across species, traditional phenotype descriptions are even less comparable, due to different naming traditions. For example, the phenotypes “adherent leaf” in Zea mays and “fiddlehead” in Arabidopsis thaliana are very similar, but this would be impossible to discern based on their names. Consistent, semantically meaningful descriptions of phenotypes are essential to the effective use of high-throughput phenotyping data and bio-informatic approaches to genotype-phenotype analysis. Ontologies provide the structured, semantically enriched vocabulary needed to describe phenotypes for contemporary biology.

This presentation will briefly introduce ontologies and cover how ontologies can be used to describe phenotypes, including a discussion of the differences and similarities (from an ontological perspective) among phenotypes, traits, characters, and character states. It will include a brief overview of how different groups are using ontologies to describe phenotypes (for both plants and animals), with a more detailed description of MaizeGDB’s efforts to create and store ontology-based phenotype descriptions.  The presentation will conclude with a discussion of how phenotypes that have been scored using ontologies can be compared across model species to search for similar genotype to phenotypes mappings, and how such comparisons can be used to identify candidate genes associated with known phenotypes in non-model species.