In distinction to MCPeval, which focuses on the inner operational conduct of Galaxy’s native agent system, this framework addresses a distinct scientific query: can an AI agent – whether or not Galaxy-native or externally developed – accurately interpret bioinformatics task specifications, work together with Galaxy instruments and datasets, execute the mandatory analytical procedures, and produce scientifically valid outputs? To determine GIA, the infrastructure for provision of tools and corresponding software program dependencies in Galaxy (e.g., Bioconda, Grüning et al., Nature Methods 2018) needed to be extended by necessary software packages for picture evaluation. LabID adopts the RO-Crate commonplace for workflow exchange: imported workflow crates are validated, versioned, and linked to their source repository, while workflow runs, including execution parameters, enter and output datasets, can be exported as Workflow Run RO-Crate packages. We acquired very optimistic suggestions on the first roadmap process and plan to improve it in future iterations by gathering input from the broader group on the roadmap, defining actions more concretely, and shaping items into extra uniform scope. By combining annotation, prediction and interactive exploration in a single open framework, the Scop3P Toolkit expands the accessibility of ELIXIR Belgium protein resources for the broader proteomics, intrinsically disordered protein (IDP) and 3D bioinformatics communities.
Together, these features illustrate a broader sample: Galaxy’s reproducibility infrastructure is not overhead for agents to bypass-it is the suggestions loop that makes agent-assisted science reliable by design. The key design choice is that this isn’t a generic chatbot making scientific guesses. We will even focus on the worth of open coaching assets, reproducible workflows, and community-curated examples for improving mannequin high quality and making automated generation more dependable. Its structure helps the combination of alternative fashions and document-processing elements, making it appropriate for speedy adaptation as AI model capabilities evolve and serving to general data re-use. This work builds on the transition to structured tool request artifacts and a unified job execution model. New work introduces robustness to make job completion resilient to those failure scenarios, reducing the operational burden of working Galaxy across many distributed compute sites. Each site brings its personal storage methods, distinctive system configurations, and failure modes. We’ve developed a unified deployment framework that brings all HPC Pulsar endpoints under a common configuration and management technique. Previously, every remote compute site was deployed and configured independently, leading to operational complexity and configuration drift. These external sources may be native or distant filesystems, require totally different community protocols, cloud object stores, repository platforms, or area-particular scientific databases.
The German Network for Bioinformatics Infrastructure (de.NBI) is a nationwide, educational, non-profit analysis infrastructure supporting information-driven life sciences in Germany and beyond. Beyond supporting learners, the GTA additionally helps strengthen the Galaxy training community. On this presentation, we are going to present a complete overview of the technical and organizational pillars supporting UseGalaxy Canada. Galaxy’s role extends beyond a technical instrument-it embodies a new paradigm for reproducible science, community-pushed innovation, and open knowledge practices in geosciences. Since its launch in 2024, the initiative has undergone a rigorous “ramp-up” phase, scaling its production infrastructure and increasing its core technical staff. AI-assisted development also helped make the migration tractable for a small team by dealing with repetitive content analysis, preprocessing logic, and many of the edge circumstances unfold throughout thousands of pages. This work highlights each the challenges and design concerns involved in adapting complicated, AI-assisted and cross-study meta-analysis pipelines for Galaxy, and goals to contribute reusable instruments and workflows again to the group. On this talk, we’ll showcase the challenges of processing image data with multiple dimensions of different types, and the corresponding options provided by the GIA auxiliary library. This work describes a coordinated set of contributions to the Galaxy Hub and Galaxy Instances initiative to address these challenges.
We evaluation the roadmap quarterly, reflecting on actions which can be accomplished or in progress to make sure timely work and might forecast ample bandwidth for future work. Within the accompanying demonstration, we present the reference implementation of the History Graph API and its interactive visualization, displaying how it unifies provenance access across visualization, automation, and AI-pushed evaluation in Galaxy. The Galaxy ecosystem continues to evolve, not just in computation and visualization, however in how analyses are represented, queried, and acted upon. To attain this, we now have mixed Galaxy with the CKAN information management platform in order to supply a seamless atmosphere for running disparate area science models for exploring power resilience. The pipeline demonstrates how direct integration of Galaxy workflows with repository infrastructure can automate provenance seize and lengthy-term archival of scRNA-seq analyses. The first framework, MCPeval, is a live integration framework developed to evaluate the reliability and correctness of Galaxy’s native AI agent infrastructure. Thus, the Galaxy neighborhood is growing a dual evaluation framework: assessing the reliability of Galaxy’s native AI agents while understanding how successfully exterior agents perform analyses within Galaxy.