DOGSURF: Database Optimized for Graph Search Using Robust Federation - Abstract The Translator project has made substantial progress in the area of biomedical data integration and question answering, evolving from a feasibility study to its current development phase. This proposal builds on prior progress made collaboratively between the Exploring Agent and Service Provider teams, and between those teams and the rest of the Translator Consortium. Our proposal, entitled DOGSURF: Database Optimized for Graph Search Using Robust Federation will further advance the project into the Performance Phase, focusing on architectural standardization and centralization, improved performance and result quality, and expanded functionality. The proposal is organized around three Specific Aims: 1. Expanding the BioPack Framework: This aim involves enhancing BioPack, a modular framework for querying the Translator Knowledge Graph (KG) and improving result scoring and organization. Collaboration with the DOGSLED team will focus on the Retriever and Shepherd components to boost performance and functionality. 2. Enhancing the BioThings SDK and BioThings Explorer: Leveraging BioPack, the project will extend the BioThings SDK and improve the BioThings Explorer. This includes enhancing answer quality, introducing new query types, and improving performance through integration with BioPack. 3. Building High-Performance Central Components: The final aim focuses on developing a unified backend infrastructure and access interface. This includes consolidating data ingestion backends, unifying services like Annotator and NodeNorm, and creating a centralized interface for efficient access across the Translator ecosystem. User engagement is central to the proposal, with a plan to collect feedback, prioritize features, and increase usage by the research community. Collaborations with external bioinformatics cores will further expand Translator's impact. The project plan provides detailed annual milestones for each Specific Aim for five years. Key abbreviations ARA: Autonomous Relay Agent – Translator components responsible for reasoning BSDK: BioThing SDK – a framework developed by our team to efficiently create high-performance APIs BTE: BioThings Explorer – an ARA developed by our team for reasoning on a federated KG DOGSLED: Data, Ontologies, and Graphs to Support Learning and Enhance Development – the name of another team with whom the BioPack vision and plan was developed DOGSURF: Database Optimized for Graph Search Using Robust Federation – the name of the project described in this proposal ITRB: Information Resources Technology Branch – the branch of NCATS responsible for provisioning computational resources KP: Knowledge Provider – Translator components responsible for providing access to information from knowledge sources KS: knowledge source – a generic term for a resource that provides useful and relevant data to Translator SDK: Software Development Kit – a set of programmatic tools and libraries to efficiently create and extend software applications
