Identifying Periodically Expressed Genes Using Microarray Time-course Expression Data

Microarray experiments are widely used for gene profiling in different cell lines, various tissues, and conditions. High throughput microarray technologies have made it possible to study the regulation of genes in cell cycle. As the genes which are significantly periordically expressed during the cell cycle play some very important roles in the cell cycle regulation, a new challenge to statisticians for analyzing the microarray experiments is to identify which are statistically significantly periodically expressed genes.Most of the existing methods of periodical gene identification base on complicated mathematical models but ignore the underlying features in periodically expressed genes. Different from the complicated models, a novel but effective method which bases on the periodic peak trait of periodically expressed gens is suggested in this article. The novel method can deal with the disturbance in microarray series data and obtain more accurate identification of periodically expressed genes.A comparison between our novel method and other three models is made by applying these methods to three different time-course expression data, and the efficacy of different methods is checked by using two benchmark data sets. The results demonstrate that our method outperform other methods in the analysis of expression data with disturbances.Finally, the cluster and functional analysis is applied to identifying periodically expressed genes. The results of cluster analysis suggest that the pattern of gene expression closely correlate with the cell stage in which the gene works. And the functional analysis reflects the functional abundance of the periodically expressed gene expression. The result of functional analysis is propitious to deeply understand the molecular mechanism of the cell cycle.