PrGeFNE

PrGeFNE predicts disease-associated genes by embedding and propagating heterogeneous biological networks to identify genetic contributors to disease mechanisms.

Key Features:

• Integration of Multi-Source Biological Data: Constructs a heterogeneous network from phenotype-disease associations, disease-gene links, protein-protein interactions, and gene-GO (Gene Ontology) associations.
• Fast Network Embedding Algorithm: Uses a fast network embedding algorithm to extract low-dimensional representations of nodes that preserve network structure.
• Dual-Layer Heterogeneous Network Reconstruction: Reconstructs a dual-layer heterogeneous network from node embeddings to represent layered associations across data types.
• Network Propagation for Gene Prediction: Applies network propagation on the reconstructed network to predict potential disease-related genes.
• Validation and Performance Evaluation: Evaluates predictions using cross-validation and newly added-association validation methods and reports performance relative to state-of-the-art algorithms in predicting disease-gene links.

Scientific Applications:

• Disease-gene prioritization: Prioritizes candidate disease genes by integrating phenotype-disease associations, disease-gene links, protein-protein interactions, and gene-GO associations.
• Elucidation of disease mechanisms: Facilitates analysis of genetic contributions to disease mechanisms through network-based embeddings and propagation.

Methodology:

Constructs a heterogeneous network from phenotype-disease associations, disease-gene links, protein-protein interactions, and gene-GO (Gene Ontology) associations; applies a fast network embedding algorithm to obtain low-dimensional node representations; reconstructs a dual-layer heterogeneous network from embeddings; applies network propagation for disease-gene prediction; evaluates performance via cross-validation and newly added-association validation.