Clustering samples characterized by time course gene expression profiles using the mixture of state space models

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Analyzing time-course gene expression data is crucial for understanding dynamic biological processes.
• Existing methods may struggle with the complexity and heterogeneity of longitudinal gene expression profiles.

Purpose of the Study:

• To develop a novel computational method for classifying samples based on time-course gene expression patterns.
• To enable automated clustering, gene discrimination, and coefficient estimation from longitudinal expression data.

Main Methods:

• Utilizing a mixture of state space models (MSSM) to capture temporal dependencies in gene expression.
• Applying the MSSM for sample clustering based on expression dynamics.
• Implementing algorithms for automatic identification of differentially expressed genes across clusters.
• Estimating cluster-specific restricted autoregressive coefficients.

Main Results:

• Successfully clustered 53 multiple sclerosis patients based on their longitudinal gene expression profiles.
• Demonstrated the capability of the MSSM to identify temporal patterns and discriminating genes.
• Provided cluster-specific autoregressive coefficient estimations.

Conclusions:

• The proposed MSSM-based method offers a robust framework for analyzing time-course gene expression data.
• This approach facilitates a deeper understanding of biological responses, such as in patients undergoing interferon beta therapy.
• The method aids in identifying patient subgroups and their underlying molecular dynamics.