CROssBAR

CROssBAR integrates extensive biomedical data into a unified NoSQL database and generates information-rich knowledge graphs augmented with deep-learning-based relationship predictions to support systems-level analysis of biological and biomedical datasets.

Key Features:

• Data integration: Integrates extensive biomedical data from multiple sources into a unified NoSQL database.
• NoSQL storage: Consolidates heterogeneous datasets within a NoSQL database for cross-resource connectivity.
• Deep-learning-based relationship prediction: Uses deep learning to predict relationships among numerous biomedical entities and to enrich integrated data.
• Knowledge graph construction: Builds information-rich, heterogeneous knowledge graphs (KGs) that incorporate both existing and predicted biomedical relationships.
• Machine and deep learning techniques: Leverages machine learning and deep learning approaches to generate relation predictions and enhance dataset connectivity.
• Support for large-scale analysis: Enables systemic analysis of large-scale biological and biomedical datasets across fragmented resources.

Scientific Applications:

• COVID-19-specific knowledge graphs: Construction of COVID-19 KGs integrating virus and host genes/proteins, interactions, pathways, phenotypes, other diseases, and known and newly predicted drugs/compounds.
• Virus–host interaction analysis: Systems-level evaluation of virus-host protein interactions and molecular mechanisms.
• Phenotype and disease mechanism studies: Investigation of phenotypic implications and relationships among pathways and diseases.
• Therapeutic hypothesis generation: Support for identifying potential therapeutic interventions and drug/compound candidates via integrated and predicted relationships.

Methodology:

Integration of biomedical data from multiple sources into a unified NoSQL database; deep-learning-based prediction of relationships among biomedical entities; construction of heterogeneous knowledge graphs incorporating existing and predicted relationships; application of machine and deep learning techniques to generate relation predictions.