Characterizing Gene Expressions Based on Their Temporal Observations

Jiuzhou Song,Hong-Bin Fang,Kangmin Duan

doi:10.1155/2009/357937

Abstract

Temporal gene expression data are of particular interest to researchers as they contain rich information in characterization of gene function and have been widely used in biomedical studies. However, extracting information and identifying efficient treatment effects without loss of temporal information are still in problem. In this paper, we propose a method of classifying temporal gene expression curves in which individual expression trajectory is modeled as longitudinal data with changeable variance and covariance structure. The method, mainly based on generalized mixed model, is illustrated by a dense temporal gene expression data in bacteria. We aimed at evaluating gene effects and treatments. The power and time points of measurements are also characterized via the longitudinal mixed model. The results indicated that the proposed methodology is promising for the analysis of temporal gene expression data, and that it could be generally applicable to other high-throughput temporal gene expression analyses.

Highlights

The high-throughput gene expression techniques, such as oligonucleotide and DNA microarray, serial analysis gene expression (SAGE) make it possible to quickly generate huge amount of time series data on gene expression under various conditions [1,2,3,4,5], and have been widely applied in biomedical studies
There are correlation genes at different time points, and the correlation structure cannot be ignored in analysis
The expression trajectories of the genes change over time for all of the genes, and at a certain time point, the change rate for each gene is different from other time point and from that of other genes

Summary

Introduction

The high-throughput gene expression techniques, such as oligonucleotide and DNA microarray, serial analysis gene expression (SAGE) make it possible to quickly generate huge amount of time series data on gene expression under various conditions [1,2,3,4,5], and have been widely applied in biomedical studies. Using the difference at two or very few time points to understand changes has some fundamental limitations It tells us nothing about each gene’s trajectory, and does not consider “overall” difference, nor does it allow studying evolution difference. For these such data with observations at very few time points, the current widely used analysis methods are various clustering methods, fold expression changes, ANOVA [6,7,8,9], and recently the hidden Markov chain models (Yuan and Kendziorski 2006). Some genetic information may be lost using fold change analysis, and difficulties arise when genes having a bigger folds change in one expression experiment have different performance in multiple arrays or different experiments.

Methods

Results

Conclusion