glycordf

Overview

GlycoRDF is a collaborative effort by the international glycomics community to create a standardized representation for glycomics data using Resource Description Framework (RDF) and Web Ontology Language (OWL). The ontology addresses the critical need for data integration across diverse glycomics databases, which historically have used incompatible formats and representations for glycan structures and associated metadata.

Motivation and Background

Over the past decades, numerous glycomics databases have been developed independently, each with unique data representations and interfaces. This fragmentation has created “disconnected islands” of valuable glycomics data, making it difficult to perform comprehensive queries across resources or to integrate glycan information with other life science data. GlycoRDF solves this problem by providing a common RDF standard that allows different databases to express their data in a unified, machine-readable format.

Ontology Structure

GlycoRDF is organized around five core concepts represented as classes:

1. Compound Class

Represents biological molecules, including glycans and glycoconjugates. Subclasses include:

• Saccharide: General glycan structures
• NGlycan: N-linked glycans
• OGlycan: O-linked glycans
• Glycoconjugate classes for molecules containing both glycan and non-glycan components (peptides, proteins, lipids)

Each glycan can be represented in multiple sequence formats including GlycoCT, LinearCode, IUPAC, and WURCS. Monosaccharide compositions and cross-references to other databases are also supported.

2. Citation Class

Captures literature references including articles, book chapters, and thesis documents. Reuses predicates from Dublin Core Metadata Initiative (DCMI) and Bibliographic Ontology for describing publication information. Supports references to PubMed and DOI.

3. Source Class

Describes the biological origin of glycan molecules with subclasses:

• SourceNatural: For glycans extracted from biological organisms, with detailed specification of species, organ, tissue, fluid, cell type, cell line, strains, life stage, and associated diseases
• SourceSynthetic: For chemically synthesized glycans

Uses existing dictionaries and ontologies such as UniProt Taxonomy for species information and Medical Subject Headings (MeSH) for tissue classification.

4. Evidence Class

Represents experimental data with subclasses for specific techniques:

• EvidenceNMR: Nuclear magnetic resonance data
• EvidenceLC: Liquid chromatography data
• EvidenceMS: Mass spectrometry data (reuses terms from HUPO-PSI Mass Spectrometry Ontology)

5. ReferencedCompound Class

Groups primary information to link glycan structures with publications, biological sources, and experimental evidence. This grouping enables complex queries such as “Which experimental data support that glycan X was found in organism Y?” or “Which papers report identification of sialic acid-containing structures by positive mode ESI-MS?”

Implementation and Adoption

GlycoRDF has been widely adopted by the glycomics community. Database providers that have implemented GlycoRDF exports include:

• CSDB (Carbohydrate Structure Database)
• GlycomeDB (integrated glycan structure database)
• MonosaccharideDB (monosaccharide nomenclature database)
• UniCarbKB (glycoprotein glycan structures database)
• GlycoEpitope (carbohydrate epitopes and antibodies database)
• GlycoNAVI (Japanese glycomics portal)
• GlycoProtDB (glycoprotein database)

These databases provide GlycoRDF data through various access methods including statically generated RDF files, web services generating RDF on-the-fly, and documented APIs.

Applications and Benefits

GlycoRDF enables:

1. Cross-database queries: Query multiple glycomics databases simultaneously using SPARQL
2. Data integration: Combine glycan data with protein, gene, and pathway information from other life science resources
3. Machine learning: Access standardized data for computational analysis and prediction
4. Data mash-ups: Create applications that integrate information from diverse sources
5. Semantic web interoperability: Link glycomics data to the broader semantic web ecosystem

Technical Features

• Extensible design: Can accommodate new data types and experimental techniques through subclassing
• Reuse of established ontologies: Leverages UniProt Core, Bibliographic Ontology, DCMI, and HUPO-PSI-MS
• Standardized dictionaries: Defines unique URIs for common glycomics terms (monosaccharide configurations, substituents, representation schemes, sequence formats)
• Flexible sequence representation: Supports multiple glycan sequence formats
• Cross-reference support: Enables mapping between database entries using owl:sameAs and rdfs:seeAlso predicates

Development and Maintenance

GlycoRDF was developed through collaborative efforts at the Japanese BioHackathon workshop in Toyama (2012) and an international Glyco-BioHackathon in Dalian, China (2013). The ontology is maintained on GitHub and is registered in the NCBO BioPortal for easy browsing and exploration.

Future Directions

The GlycoRDF initiative continues to work toward:

• Expanding the list of database providers implementing the standard
• Linking glycomics data with genomics and proteomics resources
• Establishing SPARQL endpoints and triplestores for direct querying
• Developing user-friendly web interfaces for non-technical researchers
• Supporting federated queries across multiple glycomics databases