The Ontology Lookup Service: bigger and better

INTRODUCTION

Ontologies and controlled vocabularies (CVs) have more than demonstrated their essential function when dealing with large volumes of complex data currently being generated by high-throughput multi-domain analysis techniques (1). They provide a framework around which large data sets can be systematically annotated and queried. For this framework to function efficiently, however, the ontologies and CVs must be made available to the user community.

The Ontology Lookup Service (OLS) has been in production since mid-2005 and has quickly become one of the most accessed services in the Proteomics Services team at the EBI, with monthly usage figures in the millions of hits. This includes both the programmatic as well as the interactive interfaces that the service offers. The OLS has been previously described and readers are invited to refer to the original publication for in-depth information on the technical architecture and data models (2,3).

The core functionality of the OLS has remained largely unchanged since its inception, allowing users to query ontologies and CVs by name or identifier as well as obtaining metadata, such as synonyms, definitions, cross references and other annotations, for a given term. Users can also traverse the relationships between terms. The usability and volume of data captured, however, has been enhanced and this will be expanded below.

The OLS has always been designed to be used in other projects as a means to integrate ontology and CV annotation and query functionality. A SOAP web service has been available since the OLS went into production. A full description of the web service has already been published (2,3) and users who wish to make use of it are encouraged to go to the OLS web service developer section for the most up-to-date documentation and code samples (http://www.ebi.ac.uk/ontology-lookup/WSDLDocumentation.do).

AVAILABLE DATA

The first OLS publication described it as containing 42 ontologies, accounting for roughly 135 000 terms. Over a 4-year period, the data loaded into the OLS has been expanded to 79 ontologies, representing over 971 000 unique terms (Figure 1). These cover far-ranging topics such as model organism anatomy and development, physiology and disease, instrumentation and methods and many others. In the 2 years since the OLS was previously published in NAR, 25 new ontologies have been added (Table 1). Users are encouraged to go online at http://www.ebi.ac.uk/ontology-lookup/ontologyList.do to access a full listing of currently available ontologies and CVs.

Figure 1.

Growth chart of the OLS data content. The amount of data loaded into the OLS, based on unique terms, has shown a 7-fold increase since the service went online. The first large increase is due to the incorporation of the NEWT taxonomy and the second large increase is due to the addition of a large number of ontologies at once.

The ontologies and CVs loaded in the OLS are maintained by various external groups that are domain experts in their fields. To maintain the OLS as up-to-date as possible with the current state of knowledge, the ontology providers are polled on a daily basis and updated files are downloaded and parsed to update the core OLS database. Currently, the OLS loaders poll six different Concurrent Versioning System (CVS) repositories, complemented with three Subversion (SVN) Version Control repositories thanks to the recently added SVN support. A mechanism to download individual files available by HTTP or FTP has also been implemented, which allows the loaders to track changes in files that are not in CVS or SVN

The OLS codebase is made available under the permissive Apache 2.0 Open Source License and is freely available from the Google Code project repository (http://code.google.com/p/ols-ebi/). A weekly updated MySQL database dump is also made available from the EBI FTP server (ftp://ftp.ebi.ac.uk/pub/databases/ols).

DATA MODEL IMPROVEMENTS

The OLS data loaders have been upgraded to be able to parse ontologies produced according to the Open Biomedical Ontology (OBO) 1.2 specification (http://www.geneontology.org/GO.format.obo-1_2.shtml) and can now capture previously unavailable information, such as custom name–value pairs and new synonym types (Figure 2). Another important feature of the OBO 1.2 specification is the ability to ‘import’ other ontologies and create relationships between local and imported terms.

Figure 2.

An example of custom synonyms and annotations. The synonyms in the blue box are uniquely defined in the PSI-MOD ontology. The annotations in the red box are examples of how the OLS can capture user-defined name–value attribute pairs.

In order to avoid loading multiple copies of imported ontologies, the loaders and database back end has been refactored such that each ontology is only loaded once. The OLS loaders are configured so that ontologies and CVs define one or more term prefixes that are local to itself (e.g. GO for the Gene Ontology). If the loaders encounter term identifiers that begin with a non-local prefix, they will query the OLS database and retrieve the latest version of the term in question and then proceed as normal. In this way, relationships across linked ontologies always refer to the most up-to-date data. These cross-ontology links can now also be queried and browsed, as shown in Figure 3.

Figure 3.

An example term hierarchy graph from the Ontology browser of the OLS. When using the ontology browser, selecting a term will provide a graphical display of all paths from that term to the ontology root term(s). Users can click on the terms to zoom the ontology browser to a particular term. Note the cross-ontology links in this example. The term scan start time from the MS (mass spectrometry) ontology has a relation to the second and minute terms of the unit ontology (UO), which in turn has relations to the PATO (phenotypic quality ontology).

INTERACTIVE USER INTERFACE IMPROVEMENTS

Users of the OLS website typically do one of two things: query the database using the auto-suggestion search box or browse an ontology (or a subset thereof). Once a term has been highlighted, either from the search suggestions or from the ontology browser, the user will be shown a table containing all the metadata associated with this term (synonyms, definitions, comments, cross-references and any other annotations that were captured during the loading process). When using the ontology browser, a graph showing either all the possible paths from the selected term to the root term(s) of the ontology and the relationships between all involved terms (Figure 3) or a local relationship graph with only the direct parent terms and children terms will also be shown. The type of graph to be displayed is configurable from the ontology browser interface. These graphs are clickable image maps that will zoom and re-root the ontology browser to the selected term.

REUSABLE CODE COMPONENTS

As mentioned previously, the OLS has always provided a SOAP web service. This service has been used by several large projects, such as PRIDE (4), IntAct (5), CheBI (6) and the Proteomics Standards Initiative (PSI) (7). However, the main drawback to its wider acceptance and uptake has been the lack of a simple GUI component that could easily be plugged in to existing code projects.

This has now been solved through the release of the open source OLS Dialog GUI component (8) (Figure 4). The OLS Dialog can easily be integrated into existing Java applications and gives access to the full range of query types supported by the OLS. Users can search for terms by name or by identifier, as well as use a graphical ontology browser to navigate an ontology and select a term. It is also possible to query terms from the PSI protein modification (PSI-MOD) (9) based on captured annotations from the source ontology. Users can select the type of annotation to query and enter a mass in Daltons and a desired precision and obtain all of the PSI-MOD entries that fit those parameters (e.g. find all PSI-MOD entries whose annotated monoisotopic mass is 120 D ± 1 D)

Figure 4.

Two screenshots of the OLS Dialog GUI component. The OLS Dialog allows Java application developers to seamlessly integrate OLS functionality in existing tools. Users can query the OLS by term name or ID. They can also locate terms by browsing an ontology and search the PSI-MOD ontology entries by term annotations specific to the ontology. In the left panel, a search on term names will also include partial matches and synonyms. In both cases, when a term is selected, the relevant associated metadata will be displayed and a graph similar to Figure 3 can be shown (not shown in these examples).

The OLS Dialog has been developed as part of the PRIDE Converter toolkit (10), which allows users to convert multiple mass spectroscopy file formats into PRIDE XML in preparation for submission to the PRIDE database and requires users to annotate their submission files with terms from specific ontologies. User feedback has indicated that the PRIDE Converter and OLS Dialog have made submissions to PRIDE much easier and this has been made apparent in the submission figures to PRIDE (11).

DISCUSSION

The OLS has matured into a stable system and has proven to be popular beyond our initial expectations. Besides being used as a stand-alone system, its functionality has been incorporated into several independent tools and large-scale projects and is also being used by several ontology developers as the primary ontology browser (12, 13 as examples).

When it went into production in mid-2005, the OLS was without peer. While it was true that each major ontology provider (GO, TAIR, FlyBase, Wormbase, etc.) generally provided its own website to browse their individual ontology, there was no unified resource to interactively and programmatically query multiple ontologies using a single, constant interface. Other services quickly followed suit and current systems that perform a similar function now include the National Center for Biomedical Ontology (NCBO) BioPortal (14) and the National Cancer Institute BioPortal (http://bioportal.nci.nih.gov/ncbo/faces/index.xhtml), which uses a scaled-down version of the NCBO BioPortal codebase.

A continuous increase in the number and scope of ontologies and CVs made available, coupled with an enhanced data model and better cross-ontology support will ensure that the OLS keeps its place as a valuable tool for a broad segment of the scientific community. The development team and its collaborators are always trying to make it easier to integrate OLS functionality into other projects, and the release of the OLS Dialog will go a long way towards achieving this goal. Ontology developers who wish to make their ontology available to the OLS can do so easily and through a variety of means, thanks to a versatile and automated loading process.

The OLS team is always looking for feedback to improve the project. Users are encouraged to contact pride-support@ebi.ac.uk for comments, problems and suggestions for new functionality.

FUNDING

The OLS is funded by the European Commission ‘Serving Life-science Information for the Next Generation’ (SLING), grant agreement number 226073 (Integrating Activity) within Research Infrastructures of the FP7. Formerly OLS was funded by the Biotechnology and Biological Sciences Research Council (BBSRC) ISPIDER grant and European Union (EU) FP6 ‘Free European Life-science Information and Computational Services’ (FELICS) [contract number 021902 (RII3)] grants. Funding for open access charge: EU FP6 ‘Felics’ [contract number 021902 (RII3)].

Conflict of interest statement. None declared.