Dimension Reduction in SNP Data Analysis

• Activity Number: 449
• Type: Contributed
• Date/Time: Wednesday, August 6, 2008 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Adarsh Joshi*+ and Jianhua Hu and Valen E. Johnson
• Companies: The University of Texas M.D. Anderson Cancer Center and The University of Texas M.D. Anderson Cancer Center and The University of Texas M.D. Anderson Cancer Center
• Keywords: Single nucleotide polymorphisms ; Latent Class Models ; Pseudo-Bayes Factors ; MCMC ; Dimension Reduction ; Statistical Genetics

Abstract

Genomewide SNP data typically consists of hundreds of thousands of SNP markers. Analyzing these genomewide data sets is, therefore, not computationally convenient. Our goal is to reduce a discrete SNP dataset to a more computationally manageable data set without losing relevant information. A Latent Structure Model is applied to segments of SNP data. If the model fits well, then the values of the SNPs within that segment are replaced by the corresponding classification from the latent class model. In this way, we reduce the dimension of the data set. Subsequent analyses can thus be applied to the latent class variables instead of the higher dimensional SNP data.