RNA-KG v2.0: An RNA-centered Knowledge Graph with Properties

2025-08-10

AI Summary

Existing RNA knowledge graphs suffer from insufficient contextual information and limited data coverage. Method: We construct RNA-KG v2.0—the first RNA-centric knowledge graph supporting experimental context modeling—by integrating 91 open data sources and ~100 million manually curated RNA interactions (including both coding and non-coding RNAs). We systematically incorporate standardized contextual attributes (e.g., tissue, cell line, disease state) and achieve semantic unification via RDF triples aligned with multi-source ontologies (OBO, NCBI, RNAcentral, Ensembl). We further propose a context-aware link prediction method that jointly leverages topological structure and semantic embeddings. Contribution/Results: RNA-KG v2.0 enriches nodes with sequence data, synonyms, and functional annotations, supports complex SPARQL queries, and significantly improves RNA functional inference accuracy. It serves as a scalable, context-aware knowledge infrastructure for RNA mechanistic studies and precision medicine research.

Technology Category

• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Knowledge Representation and Reasoning: Knowledge Representation Languages
• Machine Learning: Graph-based Machine Learning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

RNA-KG is a recently developed knowledge graph that integrates the interactions involving coding and non-coding RNA molecules extracted from public data sources. It can be used to support the classification of new molecules, identify new interactions through the use of link prediction methods, and reveal hidden patterns among the represented entities. In this paper, we propose RNA-KG v2.0, a new release of RNA-KG that integrates around 100M manually curated interactions sourced from 91 linked open data repositories and ontologies. Relationships are characterized by standardized properties that capture the specific context (e.g., cell line, tissue, pathological state) in which they have been identified. In addition, the nodes are enriched with detailed attributes, such as descriptions, synonyms, and molecular sequences sourced from platforms such as OBO ontologies, NCBI repositories, RNAcentral, and Ensembl. The enhanced repository enables the expression of advanced queries that take into account the context in which the experiments were conducted. It also supports downstream applications in RNA research, including "context-aware" link prediction techniques that combine both topological and semantic information.

Problem

• Integrates RNA interactions from public data sources
• Supports classification and link prediction of RNA molecules
• Enables advanced context-aware queries in RNA research

Innovation

• Integrates 100M curated interactions from 91 sources
• Enriches nodes with detailed molecular attributes
• Supports context-aware link prediction techniques