• Sergi Beltran, Centro Nacional de Analisis Genomico (CNAG), Spain
• Emidio Capriotti, University of Bologna, Italy
• Magda Chegkazi, ELIXIR-EBI, United Kingdom

The ELIXIR Rare Disease Community is dedicated to advancing a sustainable, collaborative, and interoperable research ecosystem to accelerate rare disease research across Europe and support its translation into healthcare. The Community established key service bundles including AMP4RD for molecular variant interpretation, Systems Biology Service Bundles for pathway and network-based analyses, and the ELIXIR Training Platform for Rare Diseases, an eLearning platform supporting training and capacity development. These services are powered by ELIXIR's interoperability resources and core data resources, which provide the foundation for FAIR data management, reproducible analyses, and cross-domain integration. Specifically, the Community utilises UniProt for the annotation of genetic variants, mapping amino acid changes to functional consequences and protein-level effects. Orphadata Science is used for linking resources from Orphanet, providing standardised machine-readable data on rare diseases and associated genes. IntAct enables protein-protein interaction network analysis to uncover molecular mechanisms underlying rare diseases. DisGeNET supports the retrieval of genes and variants associated with rare diseases, facilitating gene-disease association discovery. Looking ahead to 2026–2028, the Community will align with the European Health Data Space and the European Genomic Data Infrastructure, enhancing secure federated access via Beacon technology. Future activities focus on systematic FAIRification of existing tools rather than new tool development, leveraging the ELIXIR Tools Platform for benchmarking, and expanding training for clinical and research communities. Collaboration with ERDERA and GA4GH will ensure alignment and maximise impact, strengthening an integrated ecosystem supporting data-driven rare disease research and improved patient outcomes.

• Luc Mottin, BiTeM group (HES-SO) / SIB Text Mining group (SIB), Switzerland
• Julien Knafou, BiTeM group (HES-SO) / SIB Text Mining group (SIB), Switzerland
• Michael Baudis, University of Zurich (UZH) / SIB Text Mining group (SIB), Switzerland
• Anais Mottaz, BiTeM group (HES-SO) / SIB Text Mining group (SIB), Switzerland
• Venkata Satagopam, Luxembourg Centre for Systems Biomedicine (LCSB) & Université du Luxembourg (UNILU), Luxembourg
• Teresa D'Altri, Centre for Genomic Regulation (CRG), Spain
• Patrick Ruch, BiTeM group (HES-SO) / SIB Text Mining group (SIB), Switzerland

The semantic annotation of biomedical datasets remains a major challenge to support both data discovery and semantic interoperability. In this work, we present a k-Nearest Neighbors (kNN) approach for the automatic assignment of MeSH descriptors to studies from the Federated Human Data Portal (FHDportal) using MEDLINE as a knowledge base. The method relies on a simple pipeline: input text is vectorized, similar biomedical articles are retrieved, and MeSH terms are aggregated through a voting mechanism.
Two retrieval strategies were evaluated: a neural dense index using FAISS and a sparse search service based on the SIB Literature Services. The system was evaluated on EGA studies (~9,890 records) using cross-validation. Preprocessing included removal of EGA-related publications from the index and filtering of non-informative MeSH terms.
Performances were assessed using precision, recall, and F1-score at multiple cutoffs (top-5, top-10, top-15) under the assumption that an interactive service should not return more than 15-20 descriptors. Results show that the approach achieves a precision above expected inter-annotator agreement, with respectively P@0 and P@5 at 0.97 and 0.61. While the top returned descriptor is highly reliable, the precision at P@5 suggests that approximately 3 of 5 descriptors are relevant. Response time remains compatible with interactive deployment (~1.2 seconds on average).
These results suggest that a scalable classification framework can effectively support semantic enrichment of EGA metadata. The system, available through an API, is intended as a decision-support tool to support the curation workflows of EGA datasets.
Availability: https://fhdportal-dev.text-analytics.ch/docs#/

• Anaïs Mottaz, University of Applied Sciences Geneva HES-SO & Swiss Institute of Bioinformatics, Switzerland
• Valérie Barbié, Swiss Institute of Bioinformatics, Switzerland
• Michael Baudis, University of Zurich, Switzerland
• Tim Beck, University of Nottingham, United Kingdom
• Melissa Cline, UC Santa Cruz Genomics Institute, United States
• Friederike Ehrhart, Maastricht University, Netherlands
• Hindrik Kerstens, Princess Maxima Center, Netherlands
• Worawich Phornsiricharoenphant, University of Zurich, Switzerland
• Jordi Rambla, Centre de Regulacio Genomica, Spain
• Patrick Ruch, University of Applied Sciences Geneva HES-SO & Swiss Institute of Bioinformatics, Switzerland
• David Salgado, Institut Français de Bioinformatique (IFB-Core), France
• Venkata Satagopam, Luxembourg Centre For Systems Biomedicine (LCSB), Luxembourg
• Sergi Beltran, Centro Nacional de Analisis Genomico (CNAG), Spain
• Emidio Capriotti, University of Bologna, Italy

Genomic data are increasingly distributed across clinical, research and national infrastructures, making interoperability a key challenge for their effective reuse. This is particularly critical for rare disease, cancer and population genomics, where variant interpretation requires integrating both observational data and interpretative knowledge across sources.
GA4GH standards provide a common framework to address these challenges. The Variant Representation Specification (VRS) enables consistent, computable representations of genomic variants across formats and sources, while Categorical VRS (Cat-VRS) extends this to groups of related variants. The Beacon protocol supports federated discovery across distributed datasets without centralising sensitive data, with controlled access and adjustable granularity of query responses. Downstream analyses can be performed within Trusted Research Environments (TREs), where standards such as the Task Execution Service (TES) support federated execution by bringing computation to the data.
This work illustrates how these components can be articulated into a federated analysis pipeline, using a rare disease-oriented scenario as an illustrative example. Starting from a candidate variant, consistent representations enable linking observational occurrences with interpretative evidence from databases and literature, including related variants within the same functional class. Federated discovery through Beacon identifies relevant datasets, while subsequent analyses within TREs enable comparison of phenotypic profiles across cohorts.
The work highlights current efforts and remaining challenges in interoperability, including variant representation harmonisation, integration of heterogeneous data types and alignment between discovery and analysis layers.

• Carissa Bleker, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Maja Zagorščak, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Mojca Juteršek, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Sergej Praček, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Anže Županič, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Kristina Gruden, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Bruno Sá, Centre of Biological Engineering (CEB), University of Minho, Portugal
• Augustin Luna, CBB, National Library of Medicine, United States
• Yağmur Doğay, Cell Biology-Inspired Tissue Engineering (cBITE), MERLN, Netherlands
• Jan Zrimec, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Adrien Rougny, LCSB, University of Luxembourg, Luxembourg
• Ugur Dogrusoz, Computer Engineering, Bilkent University, Turkey
• Marek Ostaszewski, LCSB, University of Luxembourg, Luxembourg
• Matti van Welzen, Department of Systems Biology and Bioinformatics, University of Rostock, Germany
• Sebastian Lobentanzer, Helmholtz Munich, Germany

Understanding plant stress responses requires integrating complex, dispersed knowledge on molecular interactions into queryable, reusable resources. Stress Knowledge Map (SKM; skm.nib.si) addresses this by consolidating curated information on protein-protein interactions, transcriptional regulation, small RNA-transcript interactions, and enzymatic transformations in Arabidopsis thaliana. An additional workflow integrating multiple lines of evidence supports translation of this information to crop species including potato, tomato, and fruit trees. Together with associated analysis and modelling tools, SKM supports systematic contextualisation of multi-omic data and the development of dynamic systems biology models of stress signalling.
ELIXIR community activities have been central to SKM's growth and dissemination. The implementation study "Increasing Plant data findability and reuse beyond ELIXIR" improved the accessibility and interoperability of plant knowledge resources across the plant sciences. At the de.NBI BioHackathon Germany 2023, SKM was indexed in FAIDARE, increasing its findability within the ELIXIR plant data infrastructure. An ELIXIR Plant Sciences Community staff exchange extended SKM further, enabling a BioChatter-based prototype for LLM-accessibility to lower barriers for experimental biologists.
These developments revealed a broader need: standardised, reusable infrastructure for transforming systems biology models (in SBGN-ML and SBML formats) into human- and AI-accessible knowledge graphs. This motivated the ELIXIR Systems Biology Community's project at the 4th BioHackathon Germany (December 2025), which produced an initial common labelled property graph schema grounded in a standard ontology (SBO), SBGN and SBML BioCypher adapters, and initial support for SBML export - laying the groundwork for a community-wide framework and future AI integration across systems biology.

• Miguel Rocha, University of Minho, Portugal
• Carissa Bleker, National Institute of Biology, Slovenia
• Vitor Martins dos Santos, Wageningen University and Research, Netherlands
• Rahuman Sheriff, BioModels, EMBL-EBI, United Kingdom
• John Hancock, University of Ljubljana, Slovenia

The ELIXIR Systems Biology Community has contributed to making Systems Biology (SB) research more Findable, Accessible, Interoperable, Reusable (FAIR). Through the SYBEL project, we mapped connections between researchers and essential tools/data resources, demonstrating that access mechanisms are as vital as the resources themselves for enabling reproducible research.
Building on this foundation, we are synthesizing FAIR and CURE (Credible, Understandable, Reproducible, Extensible) principles into actionable guidelines for the community. This translates into curated FAIRsharing collections documenting standards and databases, bio.tools collections cataloging essential software, and a RDMKit page providing practical implementation guidance. These resources directly leverage ELIXIR Recommended Interoperability Resources and Core Data Resources, exemplifying how Community activities enhance the value of ELIXIR.
Looking forward, we are exploring how artificial intelligence (AI) and knowledge-graphs (KGs) can transform SB workflows. AI-assisted curation can accelerate development and annotation of models and datasets, improving efficiency and consistency. KGs enhance data integration across heterogeneous sources, making complex data more accessible. For modeling workflows, AI tools can ease accessibility to datasets, suggest appropriate formalisms, validate model structure, propose mechanistic hypotheses grounded in literature, and contribute to enhanced interpretability and explainability of SB models. In a broader sense, AI-enabled data-driven models provide a synergistic complement to fundamental knowledge-based SB models.
This evolution from merely exploring FAIR resources to AI-enhanced models/data resources represents the Community's commitment to reducing barriers in SB research. By combining robust data management practices with emerging AI capabilities, we aim to accelerate discovery, while maintaining the rigor essential for reproducible SB.

• Jan Zrimec, National Institute of Biology, Slovenia
• Sandra Correa, Bioinformatics, Institute of Biochemistry and Biology, University of Potsdam, Germany
• Maja Zagorščak, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Carissa Bleker, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Marko Petek, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Katja Stare, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Kristina Gruden, Department of Biotechnology and Systems Biology, National Institute of Biology, Slovenia
• Zoran Nikoloski, Bioinformatics, Institute of Biochemistry and Biology, University of Potsdam, Germany

Plants enter a defensive state in response to environmental stresses or pathogens, rerouting many resources from their primary growth objective and thus performing well below their maximum potential. Understanding such trade-offs can help us develop improved crop varieties with increased stress tolerance, improving yields and quality. As potato is among the most important food crops and a model tuber crop species, it is of interest to establish it also as a model system for studying growth-defence trade-offs.
To investigate growth-defence trade-offs in the context of metabolism, we developed a compartmentalised genome-scale metabolic model of potato metabolism, Potato-GEM. It contains information on the stoichiometry of over 7,000 biochemical reactions, connecting them to the genes encoding the enzymes that catalyse them. We also curated and expanded the Potato-GEM secondary metabolism, capturing its full extent according to the MetaCyc database. The potato leaf biomass was quantified using in-house and published experimental data.
We employed the model to explore growth-defence trade-offs using transcriptomics data of biotic stress responses to the pathogens Potato virus Y and pest Colorado potato beetle. This revealed the general principles of stress induced growth-defence trade-offs and enabled the comparison of micro- and macro-biotic stress response mechanisms. We thus demonstrate how the curated model can be used to investigate multi-reaction trade-offs across primary and secondary metabolism, by integrating and interpreting omics data. This provides the basis for a metabolic perspective on trade-offs and biotic interactions typical across many plant species.

• Jerven Bolleman, Swiss Institute of Bioinformatics, Switzerland
• Gos Micklem, University of Cambridge, United Kingdom
• Julia Koblitz, Leibniz-Institut DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Germany
• Pierre Larmande, AgroLD, France
• Karel Berka, Palacký University Olomouc, Czechia
• Sebastien Moretti, SIB Swiss Institute of Bioinformatics, Switzerland
• Jakub Galgonek, Institute of Organic Chemistry and Biochemistry of the CAS, Czechia
• Adrian Altenhoff, Swiss Institute of Bioinformatics, Switzerland
• Dmitry Kuznetsov, Swiss Institute of Bioinformatics, Switzerland
• Pierre-Andre Michel, Swiss Institute of Bioinformatics, Switzerland
• Deepak Unni, SIB Swiss Institute of Bioinformatics, United States
• Sabine Oesterle, SIB, Switzerland
• Nicole Redaschi, SIB Swiss Institute of Bioinformatics, Switzerland
• Alan Bridge, SIB Swiss Institute of Bioinformatics, Switzerland
• Tarcisio Mendes, Swiss Institute of Bioinformatics, Switzerland

We would like to demonstrate how interoperability between ELIXIR Core Data, Interoperability and Services of 9 ELIXIR nodes, is possible using the already deployed and supported Semantic Web based solutions. These resources connected using FAIR respecting technologies enable scientific research and interoperate with Global Core Biodata Resources, EU infrastructure and NFDI4Chem today. This demonstrates that FAIR data has no borders and that interoperability is not just a dream but a practical reality. TESS and training data are also a part of this interoperability success story, and due to its use of BioSchemas markup are part of the wider ELIXIR Semantic Web community.

Core Data Interoperability Service
OMABrowser (CH) Identifiers.org (EMBL-EBI) WikiPathways.org (NL)
BGee (CH) humanmine (UK) idsm/sachem (CZ)
OrthoDB (CH) BridgeDb (NL) MolMeDB (CZ)
UniProt (CH) InterMine (UK)
Cellosaurus (CH) AgroLD (FR)
Rhea (CH) datacatalogue (LU)
BRENDA (DE) HAMAP (CH)
LipidMaps (UK) SIBiLS (CH)
ChEBI (EMBL-EBI) SwissLipids (CH)
CHEMBL (EMBL-EBI)
Bacdive (DE)

• Mihail Anton, ELIXIR Europe, United Kingdom
• Finn Bacall, The University of Manchester, United Kingdom
• Bérénice Batut, Institut Français de Bioinformatique, France
• Salvador Capella-Gutiérrez, Barcelona Supercomputing Center (BSC), Spain
• Frederik Coppens, Ghent University, Belgium
• José Mª Fernández, Barcelona Supercomputing Center, Spain
• Alban Gaignard, CNRS, France
• Claire Garrard, ELIXIR Europe, United Kingdom
• Josep Ll. Gelpi, Barcelona Supercomputing Center (BSC), Spain
• Carole Goble, The University of Manchester, United Kingdom
• Björn Grüning, University of Freiburg, Germany
• Ove Johan Ragnar Gustafsson, Australian Biocommons, Australia
• Jennifer Harrow, ELIXIR, United Kingdom
• Alireza Heidari, University of Freiburg, Germany
• Hans Ienasescu, Technical University of Denmark, Denmark
• Arash Kadkhodaei Elyaderani, University of Freiburg, Germany
• Matus Kalas, University of Bergen, 5020 Bergen, Norway, Norway
• Peter Maccallum, ELIXIR Europe, United Kingdom
• Steven Manos, Australian BioCommons, Australia
• Ana Mendes, Southern Denmark University, Denmark
• Kota Miura, Nikon Imaging Center, University of Heidelberg, Germany
• Steffen Möller, Rostock University Medical Center, Germany
• Stuart Owen, The University of Manchester, United Kingdom
• Magnus Palmblad, LUMC, Netherlands
• Perrine Paul-Gilloteaux, Université de Nantes, France
• Yasset Perrez-Riverol, European Bioinformatics Institute, United Kingdom
• Hedi Peterson, University of Tartu, Estonia
• Manthos Pithoulias, ELIXIR Europe, United Kingdom
• Dmitri Repchevsky, Barcelona Supercomputing Center (BSC), Spain
• Claire Rioualen, Institut Français de Bioinformatique, France
• Olivier Sallou, IRISA / University de Rennes 1, France
• Veit Schwämmle, University of Southern Denmark, Ødense, Denmark
• Mariia Steeghs-Turchina, LUMC, Netherlands
• Stian Soiland-Reyes, The University of Manchester, United Kingdom
• Marco-Antonio Tangaro, ma.tangaro@ibiom.cnr.it, Italy
• Jonathan Tedds, ELIXIR Europe, United Kingdom
• Andreas Tille, Debian, Germany
• Federico Zambelli, Dept. of Biosciences, University of Milan, Italy
• Oleg Zharkov, University of Freiburg, Germany
• Hervé Ménager, Institut Pasteur; Institut Français de Bioinformatique, France

Research software underpins contemporary science, yet the metadata needed to discover, evaluate, deploy, and reuse software remains fragmented across registries, packaging systems, workflow platforms, and execution environments. This talk introduces the ELIXIR Research Software Ecosystem (RSEc), an open, federated, and version-controlled metadata commons designed to reduce that fragmentation without imposing a single metadata standard. Instead, RSEc synchronises, validates, and cross-links heterogeneous metadata from multiple upstream services, creating a shared infrastructure that supports interoperability while respecting community-specific schemas and governance models.
I will outline the architectural principles behind RSEc, including its central public Git-based repository, continuous integration workflows for automated import and validation, explicit equivalence mappings across repositories, and export into machine-actionable formats such as JSON-LD and RDF. I will also show how this infrastructure already supports practical downstream use cases: unified software discovery through the RSEc Atlas, enriched metadata delivery to services such as ToolFinder and WorkflowHub, and stronger integration with Galaxy, BioContainers, and other components of the ELIXIR tools landscape.
Beyond the technical implementation, the talk will argue for RSEc as a sustainability mechanism for research software metadata. By centralising curation effort, improving consistency across platforms, and making metadata openly reusable, the ecosystem benefits developers, users, and curators alike. The broader claim is that a federated metadata backbone is essential if research software is to become more discoverable, FAIR, and reusable at ecosystem scale. Finally, I will discuss current limitations, including dependence on upstream metadata quality, the continuing need for community-led curation and governance, and equivalence mapping across services.

• Valeria Di Cola, SIB Swiss Institute of Bioinformatics, Switzerland
• Patricia M. Palagi, SIB Swiss Institute of Bioinformatics , Switzerland
• Anne-Françoise Adam-Blondon, Unité de Recherche Génomique Info (URGI); Institut Français de Bioinformatique (IFB-core), France
• Anthony Bretaudeau, Institut de Recherche en Informatique et Systèmes Aléatoires, France
• Robert Waterhouse, SIB Swiss Institute of Bioinformatics, Switzerland

Learning Paths (LPs) serve as curated sequences of learning topics designed to guide learners through the mastery of specific subjects. Within ELIXIR, the Learning Path Focus Group collaborates across communities to standardize the development of these structured curricula. Here we outline the methodology and results the ELIXIR Biodiversity Community used to establish an initial set of three Learning Paths developed under the Implementation Study.
The development process used a systematic approach to map skills and stakeholder roles within the ELIXIR Biodiversity Community. Through a series of face-to-face and online consultations (2023-2024), we benchmarked existing LPs in Research Data Management and Plant Sciences to identify core competencies and domains. We successfully defined and populated three primary domains:
Eukaryotic Genome Sequence Assembly and Annotation
BiiGLE: Species annotation of images and videos
PLAZI: Accessing digital taxonomic literature
For each path, we mapped target audiences, entry points, and comprehensive learning outcomes. By indexing available training materials - ranging from synchronous short courses to asynchronous tutorials - we utilized the TeSS (Training e-Support System) interface to visualize these pathways. The Learning Paths also served as a diagnostic tool to identify training gaps and set priorities for the development of new training materials. The collaborative approach, combining internal analysis with external community input, ensures that forthcoming training resources will be both relevant and impactful. These LPs provide a scalable framework for biodiversity training, ensuring that the community's evolving needs are met through continuous, structured knowledge transfer.

• Becky Upton, President, Pistoia Alliance, United Kingdom
• César Velasco, Science & Innovation Director, AstraZeneca Global Hub, Spain
• Peter Maccallum, Chief Technical Officer, ELIXIR, United Kingdom

This panel will discuss how Europe can accelerate responsible AI-driven health innovation by better connecting life-science data resources, advanced computing infrastructures, and public–private collaboration.

A central question is how to ensure that Europe’s major investment in AI factories and antennas translates into real impact for life-science and health applications. The discussion will focus on the role of trusted, high-value biological and biomedical datasets, domain expertise, interoperable standards, benchmarking, and secure access to sensitive data.

The panel will also address practical roadblocks: what is still missing to make AI factories genuinely useful for life science, and how can ELIXIR, computing infrastructures, industry, and public institutions work together to close these gaps?