Webulous and the Webulous Google Add-On - a web service and application for ontology building from templates
© Jupp et al. 2016
Received: 20 October 2015
Accepted: 11 March 2016
Published: 1 April 2016
Authoring bio-ontologies is a task that has traditionally been undertaken by skilled experts trained in understanding complex languages such as the Web Ontology Language (OWL), in tools designed for such experts. As requests for new terms are made, the need for expert ontologists represents a bottleneck in the development process. Furthermore, the ability to rigorously enforce ontology design patterns in large, collaboratively developed ontologies is difficult with existing ontology authoring software.
We present Webulous, an application suite for supporting ontology creation by design patterns. Webulous provides infrastructure to specify templates for populating ontology design patterns that get transformed into OWL assertions in a target ontology. Webulous provides programmatic access to the template server and a client application has been developed for Google Sheets that allows templates to be loaded, populated and resubmitted to the Webulous server for processing.
The development and delivery of ontologies to the community requires software support that goes beyond the ontology editor. Building ontologies by design patterns and providing simple mechanisms for the addition of new content helps reduce the overall cost and effort required to develop an ontology. The Webulous system provides support for this process and is used as part of the development of several ontologies at the European Bioinformatics Institute.
KeywordsOWL Ontology Spreadsheet Webulous Google App
Like most data resources, ontologies are rarely complete, and healthy ontologies are continually growing and improving, as the state of knowledge progresses [1, 2]. Typically, authoring ontologies is a task performed by trained experts, familiar with ontology development practices and the complexities of languages such as the Web Ontology Language (OWL). This presents a major bottleneck to the ontology development process as the time and availability of trained experts is limited and ontology development is hard to fund . Tools are now being developed to simplify the addition of content to ontologies that are based on populating ontology design patterns via data entry templates.
Ontology design patterns (ODPs) are commonly used in ontology development in guiding the ontology developer in the modeling of knowledge [4, 5]. They also help in enforcing consistency and best practice in ontology design whilst reducing arbitrary class descriptions within an ontology that can lead to both errors and ontologies that are difficult to maintain. Whilst ODPs can provide a sound methodological framework, ontology expertise is still required to establish and apply modeling patterns for real-world entities from a particular domain of interest .
Several tools have been previously developed to support building OWL ontologies from design pattern templates [1, 7, 8]. The main aim of these tools is to provide a simple interface for populating a design pattern that shields the users from the underlying OWL vocabulary. These systems help to enforce rigour and adherence to a design pattern and allow new content to be added in bulk in a reproducible manner. Although these tools help in enforcing consistency of ontology development, in order to truly mediate content contributions from non-ontology experts, tools that use a familiar paradigm to domain experts are required. Such tools should enable non-ontologists to contribute whilst also tackling the issues of translating input into OWL ontologies.
In this paper we describe the Webulous framework that provides software for the management of ontology design patterns and ontology building templates. Webulous is built around a client/server architecture, where the server hosts a number of ontology building templates that can be served to any number of client applications. Data submitted to the server is translated into OWL assertions according to design patterns expressed in the Ontology Pre-Processing Language (OPPL) . We have developed a client application for Webulous using the Google Sheets Add-On framework that allows design pattern templates to be loaded into Google Sheets and submitted back to a Webulous server for processing. The Webulous client is aimed at domain experts adding new content to ontologies and is demonstrated as a term submission tool for the Experimental Factor Ontology .
Webulous provides a public service for the creation and management of ontology design templates. A Webulous server can host a number of ontology building templates that use OPPL statements to translate input data into OWL axioms. A Webulous template specifies a series of fields for the input data, and fields can can be restricted to only allow values from a list of ontology terms. The Webulous API can be used by client side applications to automatically build the user interface for a given template. Once a user populates a template with data this is submitted back to a Webulous server where the patterns are instantiated to create new OWL statements ready for import into the target ontology.
Google Sheets Add-on
Providing Webulous as a service means that a range of client-side applications can be developed for populating a template. We built a Google Sheets Add-On that supports loading Webulous templates from a server and submitting populated templates back to the server for processing. We chose Google Sheets for their convenient document management and sharing functionality and for the familiarity of the spreadsheet format for users.
When a Webulous template is loaded via the Google Add-On, each template input field represents a column in the sheet. Columns can be restricted to a set of allowed ontology terms by using term labels to create data validation. This data validation provides the user with convenient term autocomplete when entering data into a cell and will alert the user when an invalid term has been entered. Data submitted from Google Sheets is associated with the user’s Google account so the server can notify both the user and template admin via e-mail once the template has been processed.
Application of Webulous
The Experimental Factor Ontology contains descriptions of experimental variables ranging from diseases, cell types, cell lines, anatomy, assays, chemical compounds and phenotypes. It is developed as an application ontology that integrates and bridges several external reference ontologies (such as ChEBI and the Gene Ontology). EFO enriches these existing ontologies by including additional axioms that connect terms like diseases to tissues and anatomical systems; cell lines to cell types, diseases and tissue; and link common and rare diseases through associated anatomical parts and phenotypes. EFO is used to annotate resources spanning multiple omics including; transcriptomics data in ArrayExpress  and Expression Atlas , genomics data in the NHGRI-EBI GWAS Catalog , proteomics data in PRIDE  and cell line data in Encode . EFO is also used by the Centre for Therapeutic Target Validation (CTTV)1 as their core data annotation resource.
One of the appealing features of EFO is that many of the design patterns are well established and applied consistently across large portions of the ontology. This use of design patterns makes EFO nicely amenable to the generation of new content using templates. Prior to the work presented here most cell lines have been added to EFO for the ENCODE project using Excel-based spreadsheets that were processed with Populous . As more resources adopt EFO there is an increasing pressure on the editors to add new content, much of which remains in a spreadsheet-based format on submission.
Summary of EBI Webulous templates
Number of fields
Number of design patterns
Add EFO cell Line
Add EFO synonyms
Add EFO assay
Add EFO disease
Edit EFO annotation
Add EFO measurement
Users load these templates directly from the Google Sheets Webulous Add-On by simply connecting to the Webulous server running at the EBI. Once a pattern is selected a spreadsheet-based template will be created in the Google Sheet. Figure 1 shows the Cell line pattern loaded into a Google Sheet. It also shows how the data restrictions on fields have been used to create data validations on some of the columns to assist the user in data entry.
Summary of EFO content generated via Webulous data submissions
ATLAS - cell lines
ATLAS - diseases
COSMIC - cancers
UniProt - OMIM xRef
Total (as of EFO 2.65)
A public Webulous server is currently being hosted by EMBL-EBI at http://www.ebi.ac.uk/spot/webulous, where users can create their own custom templates and access them from Google Sheets. Data submitted to the EBI server is processed on the EBI Load Sharing Facility (LSF) computing cluster to provide highly scalable infrastructure for executing OPPL patterns over large ontologies. The Webulous Service for submitting EFO terms is available at http://www.ebi.ac.uk/efo/webulous. The Webulous Google Sheets Add-on is available to install form the Google Chrome store at https://goo.gl/KoHA8k. Webulous is open source and the code is kindly hosted by GitHub at https://github.com/EBISPOT/webulous.
Webulous provides a client-server architecture for the both the management of design patterns and the transformation of data to OWL axioms according to a set of applied patterns. Patterns are expressed in the OPPL language and the Java OPPL API is used to process data into OWL. For example, in OPPL we can define a simple design pattern for modeling cell nucleation as follows:
Example 1: OPPL pattern for cells and nucleation ?cell:CLASS, ?nucleation:CLASS BEGIN ADD ?cell SubClassOf hasNucleation some ?nucleation END;
This pattern defines two variables, ?cellType and ?nucleation, that are typed as OWL classes, and an OWL subclass axiom that represent the cell nucleation design pattern. By assigning concrete classes for cell type and nucleation, such as blood cell and anucleate, the OPPL API could be used to generate new OWL axioms.
Webulous works by linking fields in the template to variables in the OPPL patterns. Consider the following template that could be used to add new terms to an existing ontology. For this template we want to define a field for the new term, a field for the parent class and a field for a term definition.
Create a new template called “Add terms”.
Add the source ontology as an imported ontology.
- 3.Create three data restrictions, one for each input field
The first field is where we want users to input the new term name. We call this field “New term” and assign it to a variable called ?newTerm
The second input field will be the parent class of the new term. We call this field ?parent and we want to restrict the valid entries to any term in the source ontology. We do this by specifying the DL query as owl:Thing and selecting the descendants option. We assign the field to the ?parent variable.
Finally we want a field where the user can enter a textual definition. We call this field “definition” and leave it as an unrestricted field with the variable name ?definition.
- 4.We can use two OPPL patterns to transform any input data to OWL
Pattern 1 is used to create a subclass relation between the new newly created class in field 1 and the named class in field 2. The OPPL pattern uses the variable names to refer to the input fields.
ADD ?newTerm subClassOf ?parent
Pattern 2 is used to create an annotation assertion between the newly created class in field 1 and the textual definition supplied in field 3.
ADD ?newTerm.IRI definition ?definition
On saving this template Webulous will load the source ontology in order to prepopulate the list of allowed values in field 2 so that it is ready to serve the template to a client application.
This template is then ready for loading into a client application for user input. Once the user has populated the template with data, this can be submitted back to the Webulous server for processing and conversion to OWL. The Webulous server processes a submission by taking each row in the input data and applying the OPPL patterns associated with the template. Once all the data has been processed by OPPL, the axioms are collected together and written to a single file and made available on the Webulous FTP server4.
Webulous templates can be further configured to specify if fields are mandatory or optional. Users can enter data using the primary label rather than id for fields that are restricted to existing ontology terms. As Webulous is primarily aimed at generating new ontology content, any value that is not recognised by label in the source ontology will be created as a new term in the ontology. If the label already exists, Webulous will use the URI for that term, making it possible to refer to existing terms in an ontology. By default a random URI will be generated for the term with the user entered value set as the label. Webulous can be configured to create new URIs according to an incremental id strategy or can connect to a URIGen server for minting new term URIs.
The Webulous server is built in Java and includes an embedded Apache Tomcat5 server so it can be run directly or deployed in any other Java servlet container. The primary database for storing templates is MongoDB6 and the Spring Data and MVC7 frameworks are used to provide the REST API. Webulous includes a series of scripts for updating templates and processing data submissions that can be run as part of scheduled job. Using the scripts means that CPU and memory-intensive tasks such as executing DL queries and OPPL scripts are run outside the web server so as to avoid memory bottleneck issues with the web service. The OPPL patterns are processed with the OPPL 2 Java API8 and further processing done using the Java OWL API. OPPL2 uses the HermiT  reasoner for querying the target ontology.
RightField  and OntoMaton  are examples of ontology-aware data input tools. They are aimed at creating spreadsheet-based templates where regions of the spreadsheet are restricted to values from a list of ontology terms. Spreadsheets are a popular data entry tool and have the benefit of being both familiar to user and support the input of data in bulk. OntoMaton is built as an Add-On to Google Sheets, so it has the added benefits of the collaborative support offered by Google Documents9. However, neither tool provides support for transforming the input data into OWL axioms.
Whilst data can be readily transformed to OWL using APIs such as the OWL API , there is an increasing demand for domain specific languages (DSLs) for working with OWL that are decoupled from any particular programming language and provide a more abstract representation of the design pattern. The Manchester OWL syntax  was designed as a more user-friendly syntax for expressing OWL constructs and as such provides a good basis for a DSL. Implementations of DSLs based on the Manchester OWL syntax that can be used for ontology design patterns include the Ontology Pre Processing Language (OPPL)  and M2 . Both OPPL and M2 are designed to support a form of Manchester OWL syntax that uses variables that can be assigned to values to form new OWL axioms.
Dedicated applications like MappingMaster  and OntoRat  were developed to support the conversion of data from spreadsheets into OWL, but they don’t provide support for the management and creation of the data entry templates. Populous  was developed as an extension to RightField to provide support for both the template creation and the data transformation in a single application. Populous uses the RightField component to create Microsoft Excel templates and extends this with support for transforming populated templates into OWL using OPPL. The Populous application demonstrated how Excel spreadsheets provided a familiar user-interface for users that could be used to populate ontology templates en masse. Populous has been used in the development of several ontologies , including the Experimental Factor Ontology (EFO).
The OPPL language is extremely powerful but the lack of support and documentation for OPPL makes writing new design pattern difficult. Another limitation is that OPPL currently requires the HermiT OWL reasoner. HermiT is a DL reasoner and it cannot classify many of the ontologies available in the life science in a reasonable time, even when run on large computing clusters with lots of allocated memory and CPU. We are currently investigating the use of different reasoners with OPPL, in particular highly-scalable EL reasoners, like ELK10. The Webulous system has been designed to support different types of DSLs, other than OPPL, so could be readily extended in the future to support others OWL pattern languages such as Tawny-OWL or DOS-DP11.
We have presented the Webulous Service along with a client application that runs as a Google Sheets Add-On. Webulous has been developed to provide domain neutral support for building ontology by design patterns. Webulous is being used in the development of ontologies at EMBL-EBI and is proving to be a successful service for the bulk submission of term requests by our users. Webulous is now the new primary submission route for a range of terms in EFO, including submission of new cell lines for databases such as BioSamples and projects such as Encode. New disease-to-phenotype bridging axioms are being generated using Webulous as part of the Centre for Therapeutic Target Validation (CTTV) knowledge base. Webulous is also being used for the development for the Cellular Microscopy Phenotype Ontology (CMPO)12, an ontology being developed to annotate several cellular imaging databases including the BBSRC Image Data Repository (IDR).
Webulous is designed to complement existing development strategies and free up the time expert ontologists spend manually inserting new terms and to allow expert users of ontologies to perform knowledge representation directly on spreadsheets rather than using tools such as Protégé. The move to building the ontology by design patterns means that we can apply more rigour to ontology development. This kind of automation is especially important when an ontology grows to a size where human curators can no longer evaluate the content of the ontology as a whole before each release.
A range of tools are required to support large-scale ontology development projects, and Webulous is designed to support a specific scenario where domain experts wish to submit new ontology terms following well-established design patterns. Applications like TermGenie have shown that this approach can be successful for individual term requests, and Webulous extends this to provide an alternative interface that supports batch submissions and is configurable for any ontology.
We acknowledge core funds from EMBL-EBI, this work is also funded in part by the National Center for Biomedical Ontology, one of the National Centers for Biomedical Computing funded by NHGRI, the NHLBI, and the NIH Common Fund under grant U54-HG004028, the European Union funded CORBEL project (654248).
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Dietze H, Berardini T, Foulger R, Hill D, Lomax J, Osumi-Sutherland D, Roncaglia P, Mungall C. Termgenie - a web-application for pattern-based ontology class generation. J Biomed Semant. 2014; 5(1):48. [doi:10.1186/2041-1480-5-48].View ArticleGoogle Scholar
- Malone J, Stevens R. Measuring the level of activity in community built bio-ontologies. J Biomed Inform. 2013; 46(1):5–14. doi:10.1016/j.jbi.2012.04.002.View ArticleGoogle Scholar
- Peters B, Ruttenberg A, Greenbaum J, Courtot M, Brinkman R, Whetzel P, Schober D, Sansone SA, Scheuerman R, Rocca-Serra P. Overcoming the ontology enrichment bottleneck with quick term templates: 2009. Available from Nature Precedings 10.1038/npre.2009.3970.1.
- Gangemi A. Ontology design patterns for semantic web content. In: International Semantic Web Conference. Galway, Ireland: Springer Berlin Heidelberg: 2005. p. 262–76.Google Scholar
- Gangemi A, Presutti V. Ontology design patterns In: Staab S, Rudi Studer D, editors. Handbook on Ontologies. Springer: 2009. p. 221–43. International Handbooks on Information Systems.Google Scholar
- Aranguren ME, Antezana E, Kuiper M, Stevens R. Ontology Design Patterns for bio-ontologies: a case study on the Cell Cycle Ontology. BMC Bioinforma. 2008; 9(Suppl 5):1.View ArticleGoogle Scholar
- O’Connor MJ, Halaschek-Wiener C, Musen MA. Mapping master: A flexible approach for mapping spreadsheets to owl. In: International Semantic Web Conference (2). Springer: 2010. p. 194–208.Google Scholar
- Xiang Z, Zheng J, Lin Y, He Y. Ontorat: automatic generation of new ontology terms, annotations, and axioms based on ontology design patterns. J Biomed Semant. 2015; 6(1):4. [doi:10.1186/2041-1480-6-4].View ArticleGoogle Scholar
- Egaña M, Rector A, Stevens R, Antezana E. Applying Ontology Design Patterns in Bio-ontologies In: Gangemi A, Euzenat J, editors. EKAW 2008, LNCS 5268. Springer: 2008. p. 7–16.Google Scholar
- 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. [doi:10.1093/bioinformatics/btq099, http://bioinformatics.oxfordjournals.org/content/26/8/1112.full.pdf+html].View ArticleGoogle Scholar
- Noy NF, Shah NH, Whetzel PL, Dai B, Dorf M, Griffith N, Jonquet C, Rubin DL, Storey M-AA, Chute CG, Musen MA. Bioportal: ontologies and integrated data resources at the click of a mouse. Nucleic Acids Res. 2009; 37(Web Server issue):170–3. [doi:10.1093/nar/gkp440].View ArticleGoogle Scholar
- Kolesnikov N, Hastings E, Keays M, Melnichuk O, Tang YA, Williams E, Dylag M, Kurbatova N, Brandizi M, Burdett T, Megy K, Pilicheva E, Rustici G, Tikhonov A, Parkinson H, Petryszak R, Sarkans U, Brazma A. Arrayexpress update? simplifying data submissions. Nucleic Acids Res. 2015; 43(D1):1113–1116. [doi:http://dx.doi.org/10.1093/nar/gku1057, http://nar.oxfordjournals.org/content/43/D1/D1113.full.pdf+html].View ArticleGoogle Scholar
- Petryszak R, Keays M, Tang YA, Fonseca NA, Barrera E, Burdett T, Fullgrabe A, Fuentes AM-P, Jupp S, Koskinen S, Mannion O, Huerta L, Megy K, Snow C, Williams E, Barzine M, Hastings E, Weisser H, Wright J, Jaiswal P, Huber W, Choudhary J, Parkinson HE, Brazma A. Expression atlas update:an integrated database of gene and protein expression in humans, animals and plants. Nucleic Acids Res. 2015. [doi:10.1093/nar/gkv1045, http://nar.oxfordjournals.org/content/early/2015/10/19/nar.gkv1045.full.pdf+html].
- Welter D, MacArthur J, Morales J, Burdett T, Hall P, Junkins H, Klemm A, Flicek P, Manolio T, Hindorff L, Parkinson H. The nhgri gwas catalog, a curated resource of snp-trait associations. Nucleic Acids Res. 2014; 42(D1):1001–1006. [doi:10.1093/nar/gkt1229, http://nar.oxfordjournals.org/content/42/D1/D1001.full.pdf+html].View ArticleGoogle Scholar
- Vizcaíno J, Reisinger F, Côté R, Martens L. Pride and “database on demand” as valuable tools for computational proteomics In: Hamacher M, Eisenacher M, Stephan C, editors. Data Mining in Proteomics. Humana Press: 2011. p. 93–105. Methods in Molecular Biology.Google Scholar
- Malladi VS, Erickson DT, Podduturi NR, Rowe LD, Chan ET, Davidson JM, Hitz BC, Ho M, Lee BT, Miyasato S, Roe GR, Simison M, Sloan CA, Strattan JS, Tanaka F, Kent WJ, Cherry JM, Hong EL. Ontology application and use at the encode dcc. Database. 2015; 2015. [doi:10.1093/database/bav010, http://database.oxfordjournals.org/content/2015/bav010.full.pdf+html].
- Jupp S, Horridge M, Iannone L, Klein J, Owen S, Schanstra J, Wolstencroft K, Stevens R. Populous: a tool for building owl ontologies from templates. BMC Bioinforma. 2012; 13(Suppl 1):5. [doi:10.1186/1471-2105-13-S1-S5].View ArticleGoogle Scholar
- Shearer R, Motik B, Horrocks I. Hermit: A highly-efficient owl reasoner In: Dolbear C, Ruttenberg A, Sattler U, editors. OWLED. CEUR-WS.org: 2008. CEUR Workshop Proceedings.Google Scholar
- Consortium TGO. Gene ontology: Tool for the unification of biology. Nat Genet. 2000; 25:25–9.View ArticleGoogle Scholar
- Smith B, Ashburner M, Rosse C, Bard J, Bug W, Ceusters W, Goldberg LJ, Eilbeck K, Ireland A, Mungall CJ, Leontis N, Rocca-Serra P, Ruttenberg A, Sansone SA, Scheuermann RH, Shah N, Whetzel PL, Lewis S. The OBO Foundry: coordinated evolution of ontologies to support biomedical data integration. Nat Biotechnol. 2007; 25:1251–1255.View ArticleGoogle Scholar
- Wolstencroft K, Owen S, Horridge M, Krebs O, Mueller W, Snoep JL, du Preez F, Goble C. Rightfield: embedding ontology annotation in spreadsheets. Bioinformatics. 2011; 27(14):2021–022. doi:10.1093/bioinformatics/btr312.View ArticleGoogle Scholar
- Maguire E, Gonzalez-Beltran A, Whetzel PL, Sansone SA, Rocca-Serra P. Ontomaton: a bioportal powered ontology widget for google spreadsheets. Bioinformatics. 2013; 29(4):525–7. [doi:10.1093/bioinformatics/bts718, http://bioinformatics.oxfordjournals.org/content/29/4/525.full.pdf+html.View ArticleGoogle Scholar
- Horridge M, Bechhofer S. The owl api: A java api for owl ontologies. Semant web. 2011; 2(1):11–21.Google Scholar
- Horridge M, Drummond N, Goodwin J, Rector A, Stevens R, Wang HH. The Manchester OWL syntax. In: Proc. of the OWL Experiences and Directions Workshop (OWLED’06) at the ISWC’06: 2006.Google Scholar
- Iannone L, Rector AL, Stevens R. Embedding knowledge patterns into owl. In: ESWC. Springer: 2009. p. 218–32.Google Scholar
- Jupp S, Klein J, Schanstra J, Stevens R. Developing a kidney and urinary pathway knowledge base. J Biomed Semant. 2011; 2(Suppl 2):7.View ArticleGoogle Scholar