Gene Set Analysis with Correlated Gene Expressions

• Activity Number: 422
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: SSC
• Author(s): Qiaohao Zhu*+ and Keumhee Chough Carriere
• Companies: University of Alberta and University of Alberta
• Keywords: Gene Set Analysis ; SAM-GS ; Correlation ; Bootstrap Method ; Permutation Test

Abstract

Gene Set Analysis methods are developed to evaluate differential expressions of pre-defined set of genes. We developed new statistical techniques dealing with correlated gene expressions. We first modified the method for two-sample testing, SAM-GS, by incorporating the covariance matrix in the combined gene-set test statistics, with covariance matrix estimated from the data. We then considered the general regression models, and proposed a gene-set level test statistics by combining individual gene test statistics and incorporating the correlation matrix of the individual gene statistics. We proposed to estimate the correlation matrix among individual gene test statistics using the bootstrap method. We used the permutation method to obtain the p-value for the gene-set test statistics. Results from simulation studies comparing our methods with other methods are provided.