Narrow Results

Microarray technology allows the monitoring of expression levels for thousands of genes simultaneously. In time-course microarray experiments in which gene expression is monitored over time, we are interested in clustering genes that show similar temporal profiles and identifying genes that show a pre-specified candidate profile. Unfortunately, many traditional clustering methods used for analyzing microarray data do not effectively detect temporal profiles for the time-course microarray data. We propose a rank-based clustering analysis for the time-course microarray data.

Our clustering method consists of two steps: the first step discretizes the expression data into groups and then transform them into the rank data, the second step performs the rank-based clustering analysis. Our testing procedure uses the bootstrap samples to select the genes that show similar patterns for the candidate profiles. Simulation study is performed to evaluate the performance of the proposed rank-based method. The results are illustrated with the breast cancer data and the Arabidopsis cold stress data.