Search

Life science data is evolving to be ever larger, more distributed, and more natively web-based. However, our collective handling of identifiers has lagged behind these advances. Diverse identifier issues (for instance “link rot” and “content drift”) have hampered our ability to integrate data and derive new knowledge from it. Optimizing web-based identifiers is harder than it appears and no single scheme is perfect: Identifiers are reused in different ways for different reasons, by different consumers. Moreover, digital entities (e.g., files), physical entities (e.g., biosamples), and descriptive entities (e.g., ‘mitosis’) have different requirements for identifiers. Nevertheless, there is substantial room for improvement throughout the life sciences and several other groups have been converging on identifier standards that are broadly applicable. Building on these efforts and drawing on our experience, we focus on the use case of large-scale data integration: we outline the identifier qualities and best practices that we feel are most important in this context. Specifically, we propose actions that providers of online databases (repositories, registries, and knowledgebases) should take when designing new identifiers or maintaining existing ones (Rules 1-9). In Rule 10, we conclude with guidance to data integrators and redistributors on how best to reference identifiers from these diverse sources. This article may also be useful to data generators and end users as it offers insight into the issues associated with data provision in a web environment. We call upon data providers to take a long-term view of their entities’ scope and lifecycle, and to consider existing identifier platforms and services. Rule 1. Use established identifiers Rule 2. Design identifiers for use by others Rule 3. Help local identifiers travel well: document Prefix and Namespace Rule 4. Opt for simple durable web resolution Rule 5. Avoid embedding meaning Rule 6. Make URIs clear and findable Rule 7. Implement a version management policy Rule 8. Do not re-assign or delete identifiers Rule 9. Document the identifiers you issue and use Rule 10. Reference responsibly

In the life sciences, problems with identifiers impede the flow and integrity of information. This is especially challenging within “synthesis research” disciplines such as systems biology, translational medicine, and ecology. Implementation-driven initiatives such as ELIXIR, BD2K, and others have therefore been actively working to understand and address underlying problems with identifiers. Good, global-scale, persistent identifier design is harder than it appears, and is essential for data to be Findable, Accessible, Interoperable, and Reusable (Data FAIRport principles). Here, we build on emerging conventions and existing general recommendations and summarise the identifier characteristics most important to optimising the utility of life-science data. We propose actions to take in the identifier ‘green field’ and offer guidance for using real-world identifiers from diverse sources.

This poster was presented at the 10th International Conference on Data Integration in the Life Sciences (DILS 2014, http://dils2014.inesc-id.pt/). Background Generally, data providers identify individual records within their datasets using an identifier or accession number. Commonly, those identifiers are only unique within the dataset they originate from. For example, ‘9606’ identifies “Homo sapiens” in the NCBI Taxonomy, but also identifies “Catha edulis” in the GRIN Taxonomy for Plants, the publication “Kohlhausen K. Das Offentliche Gesundheitswesen 1976 38(7):424-430” in PubMed, the “parathyroidhormone” in HGNC, “3-Fluorotoluene” in PubChem, etc. Additionally, the data is often distributed via multiple resources, using different URLs to access individual records. For example, the enzyme nomenclature information is available from at least 4 different sources: ExplorEnz (Trinity College, Dublin), KEGG Enzyme Database (KyotoUniversity Bioinformatics Center), ENZYME (Swiss Institute of Bioinformatics) and IntEnz (EMBL-European Bioinformatics Institute). In order to address those issues related to the increased proliferation of ambiguous identifiers and non-perennial access URLs, an effort was launched in 2006 to provide a system through which appropriate URIs (Uniform Resource Identifiers) could be generated, based on existing local record identifiers already assigned by the data providers (http://identifiers.org/registry) [1]. A resolving system (http://identifiers.org/) [2], was launched to support requests from the Semantic Web community to provide these as HTTP URIs. These URIs are directly incorporable in datasets and usable by Semantic Web applications. Software tools handling data using those URIs need little work to process them and display them in a meaningful way to the end user (these URIs can actually be used as they stand in web interfaces). Moreover, these URIs are free, and provide unique, perennial and location-independent identifiers. Here, the underlying Registry acts as the central storage repository for general information on the various datasets (termed collections), namespace information (unique short string identifying the collections), and lists the resources (physical locations from where data records can be retrieved). The infrastructure is already used very successfully by, for example, the computational modelling community, which requires the ability to perennially record cross-references and links to external data records, despite the ever changing nature of the location of information on the web. It is being continually improved to meet the growingneeds of new user communities. Results Identfiers.org is now running on multiple servers in 2 redundant data centres in London, which provide more reliable, robust and faster services to the community. The semantics of the URIs handled by the resolver have been enhanced by having 2 distinct types of URIs: one for identifying the entity concept, and one for identifying the information recorded by the Registry about it. This means that canonical URIs (which uniquely identify data entities and are of the form http://identifiers.org/[namespace]/[entity]) provide more directly usable information for end users (previously a list all possible resources wasprovided, now one default resource is displayed, with a link to the former page). This default resource is selected using an algorithm which uses various properties such as resource reliability (‘uptime’) and data ownership (designated ‘primary’ for resources maintained directly by the data provider) status of the resources. Registry records of data entities are identified by URIs of the form http://info.identifiers.org/[namespace]/[entity]. ThoseURIs can be used to identify and retrieve metadata provided by the Registry or access information in various formats, such as RDF/XML (via content negotiation). With the growing uptake of Linked Data in life sciences, exemplified by projects such as Bio2RDF [3] and the launch of the EBI RDF platform [4], new functionality and improvements to the infrastructure have been made. For example, there are now numerous URIs used to identify equivalent data records. These pose a challenge whenintegrating across heterogeneous datasets, even if all are encoded in RDF. In order to facilitate such integration, we have extended the services provided by Identifiers.org in order to serve a SPARQL compliant endpoint for URI schemes conversion. This allows the conversion of a URI from one scheme into a URI from a different one. Toallow such functionality, the Registry records various URI scheme formats within the Registry, which are used to generate alternative URIs, using the Identifiers.org URIs as a canonical form. This allows the system to handle modification of the entity identifier, such converting http://identifiers.org/go/GO:0006915 into http://purl.obolibrary.org/obo/GO_0006915. Recording the alternative URI scheme formats requires additional curation efforts, but provides accurate results and should prevent any false positive. With the growth of Linked Data and Semantic Web efforts, it has also become increasingly important not only to know where data may be accessed, but also the forms or formats in which it is available. Many resources now provide their datasets in a variety of formats on top of the standard HTML format. There are users who specificallyrequire data encoded in a particular format, such as RDF/XML, or JSON. To allow direct data access to records encoded in specific formats, the Registry data model has been extended to enable it to record the various formats provided by a resource and how to access them. This information is stored at the level of the individual resources that offer records for a data collection, since each resource may offer different formats for consumption. Additionally, Identifiers.org has been updated to allow access to this information via content negotiation, performedon a user request. ConclusionIdentifiers.org URIs are now a core element of numerous data management and provision infrastructures, such as the OpenPHACTS project [5] and the EBI RDF Platform. The extension of the information recorded in the underlying Registry, and the services provided, should greatly help in the integration of heterogeneous datasets. Theseincremental improvements to the infrastructure facilitate data integration independently of the URI scheme that may have been used originally, and allows the system to be more universally suitable for other use cases. AcknowledgementsThis work received support from the BBSRC (BB/K016946/1), IMI (OpenPhacts/115191) and EMBL. References1. Laibe, C., Le Novère, N.: MIRIAM Resources: tools to generate and resolve robust cross-references in Systems Biology. BMC Syst. Biol. 1:58 (2007)2. Juty, N., Le Novère, N., Laibe, C.: Identifiers.org and MIRIAM Registry: community resources to provide persistent identification. Nucleic Acids Res. 40(D1):D580-D586 (2011)3. Belleau, F., Nolin, M.A., Tourigny, N., Rigault, P., Morissette, J.: Bio2RDF: Towards a mashup to buildbioinformatics knowledge systems. J. Biomed. Inform. 41:706–716 (2008)4. Jupp, S., Malone, J., Bolleman, J., Brandizi, M., Davies, M., Garcia, L., Gaulton, A., Gehant, S., Laibe, C., Redaschi, N., Wimalaratne, S.M., Martin, M., Le Novère, N., Parkinson, H., Birney, E., Jenkinson, A.M.: The EBI RDF platform: linked open data for the life sciences. Bioinformatics 1–2 ( 2014)5. Gray, A., Groth, P., Loizou, A.: Applying linked data approaches to pharmacology: Architectural decisions and implementation. Semant. Web. 0:1–13 (2012)

Reactome (http://reactome.org/) is a public database dedicated to curated pathways. BioModels Database (http://www.ebi.ac.uk/biomodels/) is a public repository of computational models of biological processes. In this common talk we intend to present both resources, explain their differences, show their similarities, and illustrate their increasing interactions with current and future projects.

What are annotations in Systems Biology models? How Identifiers.org can help you for this matter?

This poster was presented at COMBINE 2012 (http://co.mbine.org/events/COMBINE_2012). Abstract We describe our work to provide the community with directly resolvable URIs through Identi ers.org, using information stored in the MIRIAM Registry. Identi ers.org o ffers an annotation and cross-referencing framework which provides perennial, unambiguous and resolvable identi ers, which fi t well with Linked Data initiative and Semantic Web vision.

Background Systems biology projects and omics technologies have led to a growing number of biochemical pathway models and reconstructions. However, the majority of these models are still created de novo, based on literature mining and the manual processing of pathway data. Results To increase the efficiency of model creation, the Path2Models project has automatically generated mathematical models from pathway representations using a suite of freely available software. Data sources include KEGG, BioCarta, MetaCyc and SABIO-RK. Depending on the source data, three types of models are provided: kinetic, logical and constraint-based. Models from over 2 600 organisms are encoded consistently in SBML, and are made freely available through BioModels Database at http://www.ebi.ac.uk/biomodels-main/path2models. Each model contains the list of participants, their interactions, the relevant mathematical constructs, and initial parameter values. Most models are also available as easy-to-understand graphical SBGN maps. Conclusions To date, the project has resulted in more than 140 000 freely available models. Such a resource can tremendously accelerate the development of mathematical models by providing initial starting models for simulation and analysis, which can be subsequently curated and further parameterized.

Background: Systems biology projects and omics technologies have led to a growing number of biochemical pathway reconstructions. However, mathematical models are still most often created de novo, based on reading the literature and processing pathway data manually. Results: To increase the efficiency with which such models can be created, we automatically generated mathematical models from pathway representations using a suite of freely available software. We produced models that combine data from KEGG PATHWAY, BioCarta, MetaCyc and SABIO-RK; According to the source data, three types of models are provided: kinetic, logical and constraint-based. All models are encoded using SBML Core and Qual packages, and available through BioModels Database. Each model contains the list of participants, the interactions, and the relevant mathematical constructs, but, in most cases, no meaningful parameter values. Most models are also available as easy to understand graphical SBGN maps. Conclusions: to date, the project has resulted in more than 140000 models freely available. We believe this resource can tremendously accelerate the development of mathematical models by providing initial starting points ready for parametrization., 29 pages, 8 figures