A growth curve model with fractional polynomials for analysing incomplete time-course data in microarray gene...

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Analyzing time-course gene expression data from microarrays is complex due to the heterogeneity of regulatory responses.
• Manually characterizing individual gene expression patterns over time is infeasible for large-scale experiments.

Purpose of the Study:

• To develop an automated method for capturing diverse time-dependent gene expression patterns in microarray experiments.
• To efficiently handle missing observations in time-course gene expression data.
• To identify transcriptional responses to chemical irritants in human epidermis.

Main Methods:

• A growth curve model incorporating fractional polynomials was developed to automatically model time-course expression patterns.
• The procedure selects the best-fitting fractional polynomial model for each gene from a set of predefined power terms.
• The model was validated through simulations and applied to human in vivo irritated epidermis data with missing observations.

Main Results:

• The proposed method successfully identified various nonlinear time-course gene expression trajectories.
• The model demonstrated effectiveness in handling missing data points within the time-course experiments.
• The approach facilitated the investigation of time-dependent transcriptional responses to a chemical irritant.

Conclusions:

• The integration of growth curves with fractional polynomials offers a flexible approach for modeling diverse time-course gene expression patterns.
• This method provides robust model selection and significant gene identification strategies for microarray time-course experiments, even with missing data.