Ontologies

The term ontology is in its current usage substantially different from its original philosophical meaning, which was the "study of the nature of being." However, as most modern linguists will acknowledge, correct usage in the end is determined by widespread current practice and currently, the term ontology encompasses the study of formalized descriptions of all entities within the area of interest or research and all the relationships between those entities. In the discipline of functional genomics, bio-ontologies refer to organized and formalized systems of concepts of biological relevance in which the relationship between these concepts can be qualified. Before we proceed to some of the technical aspects of ontologies and examples of the more successful bio-ontologies, we list some notable efforts at creating usable bio-ontologies. First, we list here some additional motivations for the development of bio-ontologies.

• The need for a communal memory in the context of exponential growth in the amount of genomic information. That is, at the very least, we want to be able to share the biological knowledge that we are able to derive from our experiments, both to avoid duplication and to build on the experience of others.

• If we are to automate the process of sharing knowledge and use the prior inferences of functional dependencies, then the ontology must have sufficient formalization to enable a computational process that is both sound and efficient. As will be discussed briefly below (section 5.2), there is a natural tension between developing ontologies that are sufficiently expressive and those that are sound and efficiently computable.

• In order for a bio-ontology to be truly useful, it needs to be able to communicate the definition of terms in a reader- and context-independent way. This is particularly important in the domain of functional genomics because of the multiplicity of function of genes. For example, a gene's products will often have diverse functions within the same cell or even within the same subcellular compartment, and most often in different tissue types and under divergent environmental conditions. Stated in the converse, a gene's function will be highly determined by the subcellular context of its protein and the other coexpressed genes. For example, medium-chain dehydrogenase serves as a metabolic catalyst in one context, and in another it serves as a component of the structure of the eye lens (one of the crystallins) in some animals.

• An ontology should only represent the terms or relationships that are absolutely necessary (minimal ontological commitment). The lesser the ontological commitment, the greater the number of purposes for which a particular ontology can be used. For instance, consider a bio-ontology that is specifically designed to capture the structural protein motifs associated with a gene and to capture the potential sites of interaction between proteins. This ontology may be extremely useful in the screening of genes that might be coding for proteins that are discovered to physically interact, but it may be less useful in determining whether or not a particular protein is a known transcriptional factor for a particular gene. This latter information might best be represented in a different formalism than the protein structure representation.

• There is a significant need for organism-specific ontologies, as genes do not always have the same function across different species. A gene in the human genome may have a single homolog in yeast, but the way in which it is regulated, the biological systems in which it participates, and the nature of the additional structural complexity of its protein in the human may be obscure without a species-specific bio-ontology.

• There also needs to be cross-organism ontologies because many genes have identical or analogous function across organisms. If a finding is made in a yeast experiment, e.g., then the ontology should inform the functional genomicist of the potential for finding a human gene with the same function, if one is found with similar sequence homology. Conversely, a gene in one species may have similar function to that of another nonhomologous gene in another species. Understanding the parallels in the regulation and function of different genes across species may provide important biological insights.

Was this article helpful?

0 0

Post a comment