An extensive analysis of disease-gene associations using network integration and fast kernel-based gene...

Area of Science:

• Network medicine
• Computational biology
• Genomics

Background:

• Gene prioritization is crucial for discovering disease-associated genes using functional relationships.
• Existing methods often use single data sources, limiting discovery potential.
• Systematic evaluation of network integration's impact on gene prioritization is lacking.

Purpose of the Study:

• To conduct an extensive analysis of gene-disease associations across various diseases.
• To systematically compare different network integration methods for gene prioritization.
• To evaluate the quantitative impact of integrating multiple functional networks.

Main Methods:

• Collected nine diverse functional gene networks.
• Applied unweighted and weighted network integration techniques.
• Utilized guilt-by-association, random walk, and kernelized score functions for gene prioritization across 708 MeSH diseases.

Main Results:

• Network integration and kernelized score functions improved average AUC from 0.82 to 0.89.
• Weighted integration significantly outperformed unweighted integration (p < 0.01).
• Provided top-ranked candidate genes for each MeSH disease.

Conclusions:

• Network integration is essential for enhancing gene prioritization performance.
• Kernelized score functions, using local and global network topology, further improve ranking accuracy.
• The study identifies novel candidate genes for biomedical research.