Experimental Factor Ontology

EFO logo The Experimental Factor Ontology (EFO) is an application focused ontology modelling the experimental factors in ArrayExpress. The ontology has been developed to increase the richness of the annotations that are currently made in the ArrayExpress repository, to promote consistent annotation, to facilitate automatic annotation and to integrate external data. The methodology employed in the development of EFO involves construction of mappings to multiple existing domain specific ontologies, such as the Disease Ontology and Cell Type Ontology. This is achieved using a combination of automated and manual curation steps and the use of a phonetic matching algorithm. The ontology is evaluated with use cases from the ArrayExpress repository and ArrayExpress Atlas.

Publications on EFO

Malone J, Holloway E, Adamusiak T, Kapushesky M, Zheng J, Kolesnikov N, Zhukova A, Brazma A, Parkinson H: Modeling Sample Variables with an Experimental Factor Ontology. Bioinformatics 2010, 26(8):1112-1118 PDF

Malone J, Rayner TF, Bradley XZ and Parkinson H (2008) Developing an application focused experimental factor ontology: embracing the OBO Community. In Proc. of ISMB 2008 SIG meeting on Bio-ontologies. PDF

Explore EFO:

see the complete overview , or browse EFO in the NCBO BioPortal

Download EFO

Latest public EFO release is available in OWL format at:
http://www.ebi.ac.uk/efo/efo.owl

in OBO format (lossy conversion from OWL using OWL::Simple::OBOWriter) at:
http://www.ebi.ac.uk/efo/efo.obo

inferred view (inferred relations explicitly asserted) in OWL format at:
http://www.ebi.ac.uk/efo/efo_inferred.owl

View the Release Notes for the ontology at the following link: EFO Release Notes .

You can also learn more details about EFO by following this link on meta data, data model and process documentation.

About EFO - Modelling Experimental Variables

The driving use case in developing the EFO is based on the need for annotating experimental data in ArrayExpress. Our methodology involves transforming terms in an experimental variable vocabulary into an ontological representation, forming classes and relationships between those classes. This includes relations such as subclass which link general classes to more specific classes subtypes (such as cancer to subclass leukemia), part of relation which links classes to those which they are part of (such as cochlea is part of inner ear) and derives from which relates a class to something it was originally derived from (such as Hela cell is derived from cervix).

EFO is built using a combination of techniques; semi-automated text mining is used to map class labels from EFO to external resources, such as NCI Thesarus, Diease Ontology and ChEBI, and the mapping is maintained as an ID from this external resource into a definition_citation annotation property. In this way, we are able to map equivalent classes from EFO into multiple other ontologies, for example neoplasia in EFO maps to neoplasia in NCI Thesaurus and this is seen in the definition_citation property which has the value NCI thesaurus:C3262 to encapsulate this. This means it is easy to map data annotated with EFO to data annotated with the many other bioontology efforts that exist. EFO also involves a large degree of manual curation, checking each class, adding axioms and definitions and has also prototyped the use of agent technology to automate some aspects of ontology validation. A recent presentation at SWAS 2009: PPT - 2002KB

EFO is now used in the ArrayExpress Atlas to assist with complex querying and can be seen in use here http://www.ebi.ac.uk/microarray-as/atlas/

Acknowledgements

We would like to thank the NCBO BioPortal and Annotare developers for their assistance.

Contact Us

You can reach us at the EFO mailing list with any questions or general comments or contact us (Tomasz Adamusiak and James Malone) directly.

General ArrayExpress queries to: arrayexpress@ebi.ac.uk