Deep Functional Synthesis (DeepSyn)

Deep Functional Synthesis (DeepSyn) infers gene functions by integrating literature-derived information, a large knowledge graph, and machine learning to provide context-specific functional enrichment in disease and drug settings.

Key Features:

• Contextual Inference: Dynamically infers gene functions conditioned on specific disease and drug contexts.
• Knowledge Graph Utilization: Employs a knowledge graph comprising 3,048,803 associations among genes, diseases, drugs, and functions.
• Machine Learning Integration: Applies machine learning algorithms to data networks and the knowledge graph to predict gene functions.
• Literature and Data Network Integration: Integrates literature-derived information and data networks to inform the knowledge graph and functional inference.
• Performance: Reports area under the curve (AUC) values ranging from 0.74 to 0.96 across evaluated applications.

Scientific Applications:

• Drug Target Identification: Identifies potential drug targets by analyzing gene functional roles within disease and therapeutic contexts.
• Gene Set Functional Enrichment: Provides context-aware functional enrichment of gene sets beyond static curated databases.
• Disease Gene Prediction: Predicts genes associated with specific diseases to support genetic research and precision medicine.

Methodology:

Constructs a comprehensive knowledge graph interlinking genes, diseases, drugs, and functions and applies machine learning algorithms to dynamically infer gene functions based on experimental disease and drug context.