Bioschemas:

Abstract

The life sciences have a wealth of data resources with a wide range of overlapping content. Key repositories, such as UniProt for protein data or Entrez Gene for gene data, are well known and their content easily discovered through search engines. However, there is a long-tail of bespoke datasets with important content that are not so prominent in search results. Building on the success of Schema.org for making a wide range of structured web content more discoverable and interpretable, e.g. food recipes, the Bioschemas community (http://bioschemas.org) aim to make life sciences datasets more ndable by encouraging data providers to embed Schema.org markup in their resources.

Keywords

• Schema
• org
• metadata
• dataset descriptions
• data discovery

-

Schema.org provides a way to add semantic markup to web pages to enable those web pages to become more interpretable by the search engines that index them, and therefore to improve search results [ 2 ]. Schema.org markup describes types of information, which then have properties. For example, Recipe is a type for representing cooking recipes that has properties like cookTime, nutrition, and recipeIngredient for marking up the characteristics of the recipe. Schema.org markup is increasingly being applied to web pages as it boosts a site's ranking in search results [ 2 ]. Schema.org markup in web pages also enhances the search experience for end users; enabling them to make more informed decisions when deciding between two search results. For example, when searching for a recipe for potato salad the result snippets contain information such as cooking time and the number of calories (see Fig. 1). These have been extracted from the Schema.org markup of the underlying web pages and enable the user to make a decision without reading the whole web page. Another example of services being built over Schema.org markup include the content of the knowledge graphs of the search engines (also shown in Fig. 1).

The life sciences community have a wealth of data resources with a wide range of overlapping content. When gathering data about a particular gene or protein, scientists want the data to be aggregated from all available sources. Currently data from key repositories, such as UniProt (SIB/EBI) for information about proteins [ 3 ] or Gene (NIH) for information about genes [ 1 ], are well known and easily gathered. However, there is a long-tail of bespoke datasets with important content whose content are not so readily available. The Bioschemas community6 aim to make life sciences datasets more ndable by encouraging data providers to embed Schema.org markup in their resources. Thus enabling aggregation of content through a common approach that will enable novel applications. Previous work has added Biomedical terms7, but these are not su cient for the breadth of the life sciences community. The Bioschemas community are working in conjunction with the wider Schema.org community. 2

Bioschemas

Within the life sciences there is demand to discover more than just generic types like Dataset and Event. The Bioschemas community have identi ed a wide range of discovery use cases searching for di erent types of biological resources. 6 http://bioschemas.org (accessed Sept 2017) 7 https://health-lifesci.schema.org/ (accessed Sept 2017)

These include searching for data about a speci c biological entity such as a particular gene or protein, discovering a data repository to deposit experimental results, and identifying the storage location of speci c biological samples8. Currently biological types like genes, proteins, and samples are not represented in Schema.org. Bioschemas aims to engage with life science communities relying on existing community agreements to bring forward new biological types to Schema.org.

For any given entity type in Schema.org there are a large number of properties available, many inherited from parent types. For example, Dataset has two properties (distribution and includedInDataCatalog) but inherits 78 properties from CreativeWork and 11 from Thing. This is far more properties than can be realistically expected from resource providers, c.f. [ 2 ]. Additionally, this wide range of choice makes it di cult to develop tools to consume markup. To support tools developed to exploit Schema.org markup in web resources, it is bene cial if the markup is done in a consistent way, i.e. all resources describing a particularly type of entity provide the same set of properties.

The Bioschema speci cations are being developed in an example driven manner in a short timeframe { the ELIXIR Implementation Study runs for just one 8 Links to documents containing these use cases can be found on the Bioschemas website http://bioschemas.org/groups/ (accessed Sept 2017)

Gray et al. calendar year (2017). The Bioschema speci cations go beyond simply extending Schema.org with new types and properties for biological entities. As shown in Fig. 2, the Bioschemas speci cations layer provides additional constraints over the Schema.org model. These constraints capture (i) the minimal information properties agreed by the community which are mandatory (M), recommended (R), or optional (O), (ii) the cardinality of the property, i.e. whether it is expected to occur once or many times, and (iii) associated controlled vocabulary terms drawn from existing ontologies. Following from the experience of the wider Schema.org community [ 2 ], the Bioschemas speci cations aim to require just 6 properties for any resource type. These properties are being selected based on their ability to support indexing and snippet generation to enable a consumer of the search result to discover and distinguish between resources.

Future Work

The Bioschemas Implementation Study is currently in the development/testing phase of its lifecycle. To ensure the viability of the speci cations from the resource providers perspective, example deployments are being developed. At the same time, tools for consuming and exploiting the markup are also being developed. The outcome of both these development processes will feed into the nal revisions of the speci cations and proposed extensions to the core Schema.org vocabulary.

While the Bioschemas community has a primary focus on life sciences data, prominent members of the community are involved with the European Open Science Cloud project9 with the aim to adopt the Bioschemas approach of de ning community agreed Schema.org markup pro les in other scienti c disciplines.

Acknowledgements

The work presented here represents the contributions of the whole Bioschemas Community (http://bioschemas.org/people/). The current work is funded through an ELIXIR Implementation Study (https://www.elixir-europe.org/ activities/bioschemas) and the EU ELIXIR-EXCELERATE grant within the Research Infrastructures programme of Horizon 2020, grant agreement number 676559.

References

1. Brown et al, G.R.: Gene: a gene-centered information resource at ncbi . NAR 43 ( D1 ), D36{D42 ( 2015 ), http://dx.doi.org/10.1093/nar/gku1055

2. Guha , R.V. , Brickley , D. , Macbeth , S. : Big data makes common schemas even more necessary . CACM 59 ( 2 ) ( 2016 ), http://dx.doi.org/10.1145/2844544

3. The UniProt Consortium: UniProt: the universal protein knowledgebase . NAR 45 ( D1 ), D158{D169 ( 2017 ), http://dx.doi.org/10.1093/nar/gkw1099 9 European Open Science Cloud https://eoscpilot.eu/ accessed Sept 2017