A phase synchronization clustering algorithm for identifying interesting groups of genes from cell cycle expression...

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Previous studies indicate a percentage of genes are cell cycle regulated.
• Existing methods for identifying cell cycle genes include various analyses of expression data.
• This study hypothesizes cell cycle genes function as oscillating systems with distinct rhythms.

Purpose of the Study:

• To apply multivariate phase synchronization theory for clustering cell cycle-specific genome-wide expression data.
• To identify groups of genes involved in specific biological processes based on their expression patterns.
• To develop a novel clustering approach for gene expression data.

Main Methods:

• Utilized the modified Kuramoto model, a phase governing equation for globally coupled oscillators.
• Simulated gene expression signals with inherent rhythms to create test datasets.
• Applied a novel algorithm to analyze simulated and real yeast cell cycle expression data.

Main Results:

• The algorithm successfully identified simulated expression signals belonging to the same oscillating process.
• Clustering of yeast cell cycle data revealed gene groups with shared Gene Ontology terms and known biological interactions.
• The proposed method identified gene clusters not discoverable by traditional clustering algorithms using Euclidean distance or linear correlation.

Conclusions:

• Multivariate phase synchronization is effective for grouping genes with shared biological processes and interactions from cell cycle expression data.
• Larger gene clusters identified by the algorithm tend to contain more known biological interactions.
• Cell cycle gene expression patterns can be understood as collective synchronization phenomena.
• The developed algorithm identifies significant gene groups missed by conventional clustering techniques.