Clustering of time-course gene expression data using a mixed-effects model with B-splines

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Time-course gene expression data are crucial for understanding dynamic biological systems and gene regulatory networks.
• Mixed-effects models using B-splines address time dependency and noise in microarray data.
• This work extends these models for gene expression analysis and clustering.

Purpose of the Study:

• To explore a mixed-effects model within a mixture model framework for time-course gene expression data analysis.
• To cluster genes based on their temporal expression profiles.
• To provide smooth estimates of gene expression trajectories.

Main Methods:

• Fitting a mixture model using a Expectation-Maximization (EM) algorithm within a mixed-effects model framework.
• Utilizing B-splines to model gene expression over time.
• Clustering genes based on smooth estimates of their expression trajectories.

Main Results:

• The method successfully clusters noisy gene expression curves, differentiating clusters by curve shape or peak timing.
• Smooth mean gene expression curves were obtained for each cluster.
• Analysis of yeast cell cycle and human fibroblast serum response data revealed biologically relevant clustering and patterns.

Conclusions:

• The proposed mixed-effects model effectively analyzes time-course gene expression data and clusters genes.
• The method provides accurate gene expression trajectory estimates and identifies distinct biological patterns.
• The approach is validated on both simulated and real biological datasets.