Disease candidate gene identification and prioritization using protein interaction networks.

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Current disease gene identification relies heavily on functional annotations, which have limited coverage.
• Protein-protein interaction network (PPIN) analysis presents an alternative approach for candidate gene prioritization.

Purpose of the Study:

• To develop and evaluate a novel candidate gene prioritization method based entirely on PPIN analyses.
• To compare the effectiveness of PPIN-based prioritization against other gene features and annotations.

Main Methods:

• Application of extended PageRank and HITS algorithms, and the K-Step Markov method to prioritize disease candidate genes.
• Utilized a training-test schema with known disease-related genes as seeds for large-scale cross-validation.
• Evaluated and compared the performance of network-based methods using AUC values.

Main Results:

• The PageRank, HITS, and K-Step Markov methods achieved comparable AUC values, indicating similar performance under optimized settings.
• PPIN-based candidate gene prioritization outperformed all other individual gene features and annotations in direct comparisons.
• Demonstrated the successful application of social and web network analysis techniques to prioritize disease candidate genes.

Conclusions:

• PPIN analysis provides a robust method for disease candidate gene prioritization, especially when functional annotation coverage is insufficient.
• Network-based prioritization methods, adapted from social and web network analysis, are effective for identifying disease-related genes.