|
Abstract:
|
As the second wave of the genome wide association study (GWAS), approaches of detecting association with multiple markers on a candidate region and disease are gaining an attention. Due to the linkage disequilibrium among markers, methods that are using few hidden components among markers such as principal components analysis (PCA) or fast Fourier transform (FFT) decomposition show a great performance to detect the association among multiple markers and a disease. However, the hidden components from PCA or FFT are derived only from covariates. In this study, we will derive hidden components by utilizing both response and covariates via generalized partial least squares (PLS) approach to detect an association between multiple markers and a disease. In the midst of the study, we will compare and discuss several versions of existing generalized PLS approaches.
|
