The description of phenotype

A great strength of modern genetic analysis has been the ability of researchers to understand both the functions of individual genes and the roles of complex biological processes by the study of the consequences of mutation. Mutations are seen as a consequence of their effects on the phenotype of organisms. For just about 100 years these phenotypes have been described in free text. A mutant fly might be described as having 'white eyes', a mutant mouse as having a 'kinked tail' or a mutant plant as 'having delayed flowering'. Expressive as these descriptions are they suffer from being opaque to computational reasoning, and they hinder comparisons of phenotypes both within and between organisms. This is becoming increasingly important, since much of the justification for mutant screens is that the analysis of phenotypes will inform human biology or the biology of major crop plants. The solution is reasonably clear, it is to annotate phenotypes using a controlled vocabulary, an ontology. Our proposal is to develop an ontology of phenotypic qualities that then be used to annotate, within the context of open databases, the entities affected by mutation. For example, we would then express a fly with white eyes as, 'Entity = [fly] eye with Quality = white'. An entity need not be an anatomical part, it could equally be a biological process, such as 'flowering' of a higher plant, thus a plant with delayed flowering could be described as 'Entity = flowering; Quality = delayed'. In practice these so-called EQ statements are rather more formal and complex than these examples. One principle is that the entities are always taken from another available ontology, for example that of Drosophila anatomy or that of Biological Processes. Phenotypes described in this way are, however, computationally tractable, and a computer could reason that a fly with 'Entity = [fly] eye with Quality = white' is an example of all flies with an abnormal eye colour. Our proposal is to build and maintain this ontology of qualities and to promote its use by the community.

We propose to develop an ontology of qualities to be used for the annotation of phenotypes. This will enable mutant phenotypes of model organisms (including Drosophila, zebrafish, mouse and Arabidopsis) to be described using an Entity:Quality (EQ) model, where the entity may be an anatomical part (e.g. a mouse tail), drawn from the appropriate model organism anatomical ontology, or a biological process (e.g. flowering) drawn from the Gene Ontology. Associated with this ontology, which we call the Phenotype and Trait Ontology (PATO) will be a Units Ontology, so that quantitative measurements can be formally expressed in the syntax. Working with colleagues in the National Center for Biomedical Ontology (NCBO) we have developed an annotation tool called Phenote which has been field tested by several communities (including FlyBase, ZFIN) for the use of multiple ontologies for annotation of phenotypes. While no annotation per se is proposed here we will work very closely with three communities in particular to encourage the use of PATO and the communal warehousing of PATO-annotations in a common database, the Open Biomedical Database of the NCBO. These will be the Model Organism Database community, the international communities now conducting high-throughput mutant screens for mice, zebrafish and other organisms, and the evolutionary biology community, who wish to use PATO for the annotation of organismal phenotypes within an evolutionary context.