arax

ARAX is a modular, open-source software platform designed for reasoning over large biomedical knowledge graphs. It provides tools to assemble, query, and integrate heterogeneous data sources by representing information as a network of concepts (nodes) and relationships (edges). ARAX is used to perform automated reasoning, generate hypotheses, and retrieve explanations for complex biomedical questions.

A core feature of ARAX is its support for the Translator Reasoner API (TRAPI), a standard interface for sending requests to knowledge providers and receiving structured results. Through TRAPI, ARAX coordinates multiple reasoning services, executes complex queries, and returns results that include supporting evidence and provenance. This architecture enables interoperability across diverse data sources and enables the composition of multi-step reasoning workflows.

The ARAX architecture comprises a central reasoning engine and a suite of modular components. Key components include a Node Synonymizer to unify concept identifiers, edge scoring and filtering to assess relationship confidence, and path-finding modules to discover connecting routes between entities. The platform supports plug-ins and adapters to query local graphs or remote endpoints, apply domain-specific constraints, and format results for downstream analysis or user display.

Applications of ARAX span several areas of biomedicine, including drug repurposing, disease mechanism exploration, and biomarker or target discovery. Researchers can generate and test hypotheses by tracing connections among genes, proteins, compounds, diseases, and phenotypes, with results that include reasoning steps and justifications.

ARAX is an ongoing, open-source project with active development and a community around TRAPI. It is used by researchers and developers to enable interoperable knowledge-graph reasoning in translational science and related fields.