Bayesian profile regression for clustering analysis involving a longitudinal response and explanatory variables

Area of Science:

• Genomics
• Bioinformatics
• Statistical Modeling

Background:

• Identifying co-regulated genes with shared functions is crucial in modern genomics.
• Bayesian profile regression is a semi-supervised method for clustering data based on a response variable.
• Existing methods handle univariate outcomes, limiting broader applications.

Purpose of the Study:

• To extend Bayesian profile regression for longitudinal (multivariate continuous) outcomes.
• To introduce PReMiuMlongi, an updated R package for profile regression analysis.
• To apply the extended model to identify co-regulated gene groups in yeast.

Main Methods:

• Developed an extension of Bayesian profile regression for longitudinal data.
• Incorporated multivariate normal and Gaussian process regression response models.
• Utilized PReMiuMlongi R package for analysis and simulation studies.

Main Results:

• Successfully applied the model to budding yeast data from the Saccharomyces cerevisiae cell cycle.
• Identified four distinct groups of co-regulated genes based on their expression trajectories.
• Linked these gene groups to specific transcriptional factors involved in co-regulation.

Conclusions:

• The extended Bayesian profile regression model effectively analyzes longitudinal gene expression data.
• PReMiuMlongi facilitates the discovery of co-regulated gene sets and their regulatory mechanisms.
• This approach enhances our understanding of gene regulation during the yeast cell cycle.