Identification of human disease genes from interactome network using graphlet interaction.

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Identifying disease-related genes is crucial for understanding and treating conditions like cancer and cardiovascular disease.
• Protein-protein interaction networks are utilized for disease gene identification, assuming related genes cluster within the network.

Purpose of the Study:

• To introduce a new network analysis measure, graphlet interaction, for improved disease gene identification.
• To develop a scoring system based on graphlet interaction to rank candidate disease genes.

Main Methods:

• Proposed a novel 'graphlet interaction' measure analyzing relationships between network nodes, comprising 28 isomers.
• Designed a network property-based score utilizing graphlet interaction for disease gene identification.
• Evaluated the approach using leave-one-out cross-validation and compared it with existing algorithms.

Main Results:

• Disease genes showed significantly more graphlet interaction isomers in interactome networks than random genes.
• The proposed scoring system achieved 90% precision at 10% recall, outperforming random walk, Endeavour, and neighborhood-based methods.
• Successfully predicted new disease genes for four common diseases, with many predictions validated by independent research.

Conclusions:

• Graphlet interaction is an effective tool for analyzing disease gene network properties.
• The scoring system based on graphlet interaction offers higher precision in identifying disease genes.