Coronavirus Network Explorer

Coronavirus Network Explorer maps SARS-CoV-2 viral proteins to host biological functions, diseases, and pathways using a machine learning-enabled knowledge graph to support mechanistic hypothesis generation and drug target identification.

Key Features:

• Knowledge graph integration: Uses a large-scale knowledge graph derived from extensive biomedical literature to connect viral proteins with host entities and evidence.
• Machine learning and gene embeddings: Employs machine learning with gene embeddings that utilize causal gene expression signatures curated from the literature to relate genes to biological functions.
• Directional effect analysis: Explicitly distinguishes activation versus inhibition effects when linking genes, functions, and pathways.
• Network construction: Constructs 70 networks linking SARS-CoV-2 viral proteins to biological functions, diseases, and pathways that reflect viral biology, clinical observations, and comorbidities associated with COVID-19.
• Network visualizations: Produces network visualizations that expose the underlying experimental evidence supporting each connection.
• Drug repurposing and enrichment: Identifies existing drugs targeting genes within the networks and reports significant enrichment of those drugs among agents in COVID-19 clinical trials, suggesting potential mechanisms of action.

Scientific Applications:

• Host–virus interaction mapping: Enables exploration of mechanistic links between SARS-CoV-2 proteins and host biological processes, diseases, and pathways.
• Drug target identification: Supports identification of host genes as candidate drug targets for repurposing existing drugs against COVID-19.
• Hypothesis generation for pathogenesis: Facilitates generation of mechanistic hypotheses related to immunological, virological, and pathological aspects observed in infected patients.

Methodology:

Integrates a large-scale biomedical literature-derived knowledge graph with machine learning using gene embeddings based on causal gene expression signatures; distinguishes directional effects (activation/inhibition); constructs 70 SARS-CoV-2–host networks; and identifies drugs targeting network genes with enrichment analysis against drugs in COVID-19 clinical trials.