Ontologies for Biomedicine – How to make and use them

From NCBO Wiki
Revision as of 20:21, 15 December 2011 by JoanneLuciano (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
 When: July 21, 2007
 Where: ISMB 2007, Vienna, Austria
 What: Tutorial on Ontologies for Biomedicine – How to make and use them
 Faculty: Nigam Shah (Stanford), Barry Smith (Buffalo)

Brief: The tutorial provides an overview of current uses of ontologies in bioinformatics, instruction on ontology development using OWL, and an outline of principles of good ontology design. It comprises an interactive session devoted to developing an ontology, and a session illustrating how good design practices effect the usefulness of ontologies in molecular biology.

Tutorial level: Introductory/Intermediate

Prior knowledge required: Participants should be at least aware of ontologies such as the Gene Ontology and BioPAX. Familiarity with basic biology terms such as genes, proteins, promoter, intron and exon is required.

Abstract: Ontologies are becoming essential with the rapid increase in the quantities and types of data we need to handle in the molecular biology domain. Simultaneously, the need to organise, co-ordinate and disseminate ontologies as well as coherent ontology development methods is now accepted and is evidenced by the funding of the National Center for Biomedical Ontology (NCBO). Though the need to use ontologies is widely appreciated, the right manner in which they should be developed and applied is not well understood. Researchers still resort to ad hoc methods in developing and using ontologies, with the result that many opportunities for integration and cross-disciplinary communication are lost, and many obstacles to cross-domain reasoning are created. This tutorial will provide an overview of the various ways in which ontologies are used in bioinformatics and biomedicine, along with pointers to innovative applications of ontologies such as to validate pathway data and enable cross-database reasoning. It will educate the participants on what ontologies are and in how they are currently used in molecular biology. It will provide guidance on how to create ontologies using OWL and on best practices for ontology development and use. This tutorial will be aimed at advanced graduate students and active researchers who need to understand the basic principles underlying ontology development in order to use ontologies more effectively for interpreting their own data (for example interpreting microarray data using the Gene Ontology or pathway information from Reactome). It will be of value, too, to researchers who need to participate in ontology development efforts in their scientific respective communities.

 Download the handout
 See the slides

Overview of current applications of ontologies in bioinformatics

  • As a controlled vocabulary to annotate genes and gene products (e.g The Gene Ontology)
  • As a data exchange format and for data integration (e.g. MGED, SBML and BioPAX)
  • To define a knowledgebase schema (e.g. BioCyc and Reactome)
  • For driving natural language processing (e.g. Textpresso and Geneways)
  • For semantically rich querying of federated databases (e.g. TAMBIS)
  • Creating formal representations of biological processes for hypothesis evaluation (e.g. HyBrow)

Ontologies – What they are and what they are not

  • We will discuss “ontology” as understood in philosophy, computer science and information science to explain how the computer/information science meanings are different - but related to - the philosophical meaning of ontology. In practice, ontologies provide standardized labels which are used to annotate different experimental data. We will discuss the implications of this annotation-based view of ontology to clarify the difference between terminologies, taxonomies, application ontologies and reference ontologies.

Building ontologies using OWL

  • We will provide an overview of the various constructs available in OWL and of how they can be used to represent biomedical knowledge in an ontology
  • We will demonstrate an example of how DL-reasoners can be used to verify ontologies created in OWL and how such reasoning will help in reducing errors in biomedical ontologies.
  • We will provide indications of some of the problems facing ontology developers using OWL in the bioinformatics domain, and offer solutions to these problems.

Interactive session: developing a small ontology

This session will consist of an audience-driven discussion with the goal of developing an ontology to represent DNA sequences. We will focus on the task of representing relationships between introns, exons, start sites, promoters, genes and chromosomes. (We have used this topic before and have found that it is accessible to participants with a wide background and has enough complexity that a large fraction of the common mistakes do happen)

The do’s and don’ts of ontology creation (highlight common mistakes and recommended design strategies)

  • We will describe the key principles of ontology design that address the most common sources of mistakes in biomedical ontologies.
  • We will present rules of thumb for good ontology design, along with examples of how they will enhance the quality of the resulting ontologies. We will also outline the guiding principles of the OBO Foundry, and describe how they can enable cross-ontology reasoning.

Wrap up question and answers with discussion

Total time = 3.5 hours [2 sessions of 105 minutes each] + 30 min break