LinkExplorer

LinkExplorer predicts and explains links in large biomedical knowledge graphs to support hypothesis generation and discovery of associations among biological entities.

Key Features:

• Link Prediction: Employs machine learning algorithms tailored for link prediction and integrates the rule-based link prediction engine SAFRAN for link inference.
• Explainability: Provides rule-based explanations that clarify why specific link predictions are made.
• Knowledge Graph Evaluation: Evaluated across multiple large biomedical knowledge graphs and compared against other explainable and black-box algorithms to assess performance.
• Hypothesis Generation: Produces candidate links between entities to facilitate formulation of testable scientific hypotheses.

Scientific Applications:

• Hypothesis Generation: Predicts potential links to generate testable hypotheses about relationships among biological entities.
• Disease Mechanism Discovery: Uncovers novel associations that can inform investigations into disease mechanisms.
• Drug Target Identification: Identifies candidate relationships relevant to discovery and prioritization of drug targets.
• Therapeutic Strategy Exploration: Reveals associations that can inform development and evaluation of therapeutic strategies.

Methodology:

Integration of the rule-based link prediction engine SAFRAN and machine learning algorithms for link prediction, with evaluations and comparisons performed on multiple large biomedical knowledge graphs against other explainable and black-box approaches.