AIdrug2cov

AIdrug2cov applies network representation learning to identify candidate drugs, targets, and mechanisms that mitigate immune imbalance and inflammatory cytokine responses in COVID-19.

Key Features:

• Network Representation Learning: Employs a deep bidirectional Transformer encoder to learn low-dimensional vector representations from heterogeneous biological networks.
• Target and Drug Identification: Identifies 40 potential targets and 24 high-confidence drugs interacting with tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6).
• Mechanistic Insights: Provides mechanistic links indicating that chloroquine and hydroxychloroquine may reduce fatality via inhibition of inflammatory cytokines in addition to antiviral activity.
• Performance Evaluation: Demonstrates superior performance in comparative analyses against five state-of-the-art network representation approaches across pharmacological applications.
• Focus on Immune Imbalance and ARDS: Targets immune imbalance associated with acute respiratory distress syndrome (ARDS) in COVID-19.

Scientific Applications:

• Drug mechanism discovery: Elucidates mechanisms of action for candidate drugs in COVID-19, emphasizing anti-inflammatory effects.
• Drug repurposing prioritization: Prioritizes high-confidence repurposed drugs for mitigating excessive inflammatory responses mediated by TNF-α and IL-6.
• Target identification: Identifies potential protein targets implicated in immune imbalance and ARDS.

Methodology:

Uses a deep bidirectional Transformer encoder network representation approach to learn low-dimensional vector representations from heterogeneous networks.