Autonomous Relay System
Overview
The Autonomous Relay System (ARS) serves as the central query routing and orchestration service for the NCATS Biomedical Data Translator program. It acts as an intelligent relay that accepts biomedical questions, distributes them to appropriate knowledge providers (KPs) and autonomous reasoning agents (ARAs), and aggregates the results into unified responses.
Key Features
- Query Orchestration: Routes biomedical queries to multiple knowledge providers and reasoning agents
- Intelligent Distribution: Selects appropriate resources based on query type and available data
- Result Aggregation: Combines responses from multiple sources into coherent answers
- API Gateway: Provides a unified interface for accessing Translator resources
- Asynchronous Processing: Handles long-running queries with callback mechanisms
- Load Balancing: Distributes queries efficiently across available services
- Query Tracking: Monitors query status and provides real-time updates
Architecture
Components
- Query Interface: Accepts TRAPI (Translator Reasoner API) formatted queries
- Routing Logic: Determines which KPs and ARAs should process each query
- Message Queue: Manages asynchronous query distribution and response collection
- Result Merger: Aggregates and ranks results from multiple sources
- Status Monitor: Tracks query progress and service health
Workflow
- Client submits biomedical question via TRAPI format
- ARS analyzes query structure and requirements
- Query is distributed to relevant KPs and ARAs
- Services process query and return results
- ARS aggregates responses, removing duplicates and ranking results
- Unified response returned to client
Query Types
ARS supports various biomedical query patterns:
- Drug-disease associations
- Gene-phenotype relationships
- Mechanistic pathways
- Drug repurposing hypotheses
- Multi-hop reasoning chains
- Creative mode queries (exploratory)
Applications
- Translational Research: Connecting basic science to clinical applications
- Drug Discovery: Identifying therapeutic candidates for diseases
- Hypothesis Generation: Discovering novel biomedical relationships
- Knowledge Integration: Unified access to diverse biomedical data sources
- Clinical Decision Support: Evidence-based insights for patient care
Integration
ARS connects to the Translator ecosystem:
- Knowledge Providers (KPs): Specialized databases exposing biomedical assertions
- Autonomous Reasoning Agents (ARAs): Services performing inference and reasoning
- User Interfaces: Web applications and tools for end users
- External Systems: APIs for programmatic access
Technical Specifications
- API Standard: TRAPI (Translator Reasoner API)
- Message Format: JSON-based knowledge graphs
- Ontology: Biolink Model for semantic standardization
- Deployment: Cloud-based microservices architecture
- Authentication: API keys for rate limiting and tracking
This resource has the Information Resource identifier: infores:ars
Access
- Production ARS: https://ars.transltr.io/
- Development ARS: https://ars-dev.transltr.io/
- Documentation: https://github.com/NCATSTranslator/Translator-All/wiki/Autonomous-Relay-System-(ARS)
- API Specification: https://smart-api.info/ui/
For more information about the NCATS Biomedical Data Translator program, visit https://ncats.nih.gov/translator