A Moderated Lawley-Hotelling--Type Trace Test for High-Dimensional Longitudinal Data: Its Application in Detecting Temporal Patterns in Time Course Microarray Data

• Activity Number: 634
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Jemila Seid Hamid*+ and Joseph Beyene and Dietrich von Rosen
• Companies: Dalla Lana School of Public Health and Macmaster University and Swedish University of Agricultural Sciences
• Keywords: growth curve model ; high dimensional data ; moderation ; wighted sum of chi-square variables ; Satterthwaite's approximation

Abstract

The growth curve model is a GMANOVA model which is particularly useful in analyzing short time series data. It has many applications in medical and biological research. Recently, the model is used in ranking genes in time course microarray experiments. In this study, we consider a trace test we proposed in a previous work and extend it to high-dimensional time course data by using a moderated version of the sample covariance matrix. It is shown that the distribution under the null hypothesis can be represented as weighted sum of central chi-square random variables and Satterthwaite's approximation is used to provide p-values. Under the alternative, a weighted sum of non-central chi-square variables is used to describe the distribution. We use a simulation study to assess the performance of our method and illustrate the results using a publicly available time course microarray data.