Unit mission
SIB’s Knowledge Representation Unit interoperates and integrates diverse data stored in multiple locations and formats, and develops pioneering interfaces to quickly access and explore interconnected databases.
This increases the value of life science data by:
- maximizing data reuse by people and AI systems
- connecting knowledge within and across fields
- enabling a holistic view of complex biological systems
- generating more powerful insights than possible from individual databases.
The unit is co-led by Tarcisio Mendes de Farias and Ana-Claudia Sima and is part of the SIB Vital-IT Computational Biology group at the SIB Hub. Its members contribute to a wide variety of Swiss and international research projects involving academic, hospital, and industry partners, and collaborate across the SIB network to enhance bioinformatics tools, resources and discovery.
Data integration and interoperability are key enablers of scientific discoveries, especially as the quantity of biodata continues to grow.
Knowledge representation expertise
The team has extensive expertise in semantic web technologies for formally describing, structuring and connecting data in a meaningful way, as well as in building user-friendly tools for efficient retrieval of data and information across multiple databases.
- Ontology engineering: developing machine-readable ontologies (or vocabularies) and metadata standards for describing biodata and their relationships, and applying these to heterogenous databases from different sources, fields and/or locations.
- Knowledge graphs: structuring metadata and databases into integrated networks of linked data that seamlessly interconnect information stored in the different databases.
- Database querying: using SPARQL and other technical query languages for retrieving and compiling relevant information from large quantities of complex data from different sources.
- Semantic search and natural language processing: delivering user-friendly interfaces for fast and easy data access, retrieval, and compilation, by integrating knowledge representation with sophisticated keyword search and Large Language Model (LLM)-based generative AI technologies.
Integrating knowledge graphs with LLM-based AI tools allows users to quickly find relevant data and accurate, specialized information without computer science expertise.
Advancing knowledge discovery and preservation
The SIB Knowledge Representation Unit contributes to a variety of multi-site, cutting-edge data projects in Switzerland and internationally – many of which are funded by highly competitive grants, including from the Swiss National Science Foundation (SNSF), swissuniversities, The Loop Zurich, and Horizon Europe.
- Advancing the European Open Science Cloud (EOSC): The multi-partner EOSC EDEN, FIDELIS, and EOSC Data Commons projects will ensure the long-term preservation of Europe’s data assets and allow AI-based discovery of data and tools across the entire collection of European research outputs. FIDELIS also involves the SIB Biodata Resources and Training teams.
- Enabling AI-driven precision oncology: The AI Tumor Board project with Swiss hospitals is converting unstructured PDF guidelines for cancer care into structured, interoperable treatment procedures, and building an AI tool to analyse the procedures plus historical clinical data to predict the best treatment for individual patients.
- Fostering sustainable, climate-resilient agriculture: the European MICROBES4CLIMATE project is building integrated infrastructure to support research on interactions between soil microorganisms, plants and the environment. This multi-partner effort also includes SIB’s Environmental Bioinformatics group.
- Bridging data silos: The Digital Botanical Gardens Initiative, Earth Metabolome Initiative, and MetaboLinkAI project are connecting and curating vast quantities of heterogenous data on plant metabolic molecules and enabling AI-assisted querying of this metabolomics knowledge. These joint initiatives also involve other members of the Vital-IT Computational Biology group and the SIB Group led by Pierre-Marie Allard.
- Linking public and private databases: Coordinated by SIB, the European TRIPLE initiative is enabling an unprecedented level of interoperable data sharing between researchers from any science domain, including integrated searches across public-private research data. The multi-partner initiative additionally includes members of the Vital-IT Computational Biology group, Swiss-Prot knowledgebases group, and the SIB Group led by Christophe Dessimoz and Natasha Glover
The Knowledge Representation Unit also to developed a standard way for staff at BASF, a leading crop protection company, to find and access data produced and stored by different internal groups and public databases. - Keeping SIB Resources at the cutting edge: The unit’s work to develop, deploy and maintain highly connected knowledge graphs for SIB Resources contributed to Bgee, a gene expression database developed by the SIB Group led by Marc Robinson-Rechavi and Frederic Bastian, becoming an ELIXIR Recommended Interoperability Resource.
- Accelerating biological and bioinformatics discovery: The SIB Resource knowledge graphs also form one foundation of ExapsyGPT, a powerful generative AI tool that makes it faster and easier than ever to explore, retrieve and compile data and information to answer complex biological questions. ExpasyGPT is a collaboration between the Knowledge Representation Unit and other members of the Vital-IT Computational Biology group, the SIB Semantic Web of data focus group, and members of the Information Technology and Biodata Resources teams at the SIB Hub.
The Knowledge Representation Unit also develops methods and tools to automatically disseminate data from SIB Resources into non-SIB life science knowledgebases and public encyclopaedias such as Wikipedia.
Members
View Knowledge Representation Unit members here
