CUR Matrix Decompositions Method for Joint Analysis of Multiple Phenotypes

Abstract

There are many methods that can be applied to reduce the dimension of large data sets such as principal components analysis (PCA). Many of these methods are linear combinations of up to all of the data points, which is typically quite hard to interpret. Recently, CUR matrix decompositions was developed as an alternative dimensionality reduction technique, which has been applied to genetic and internet data. CUR decompositions are low-rank matrix decompositions that are expressed in terms of a small number of actual columns and/or actual rows of the data matrix. CUR decompositions are interpretable because they are constructed from actual data. In this report, we apply the CUR decompositions method for joint analysis of multiple phenotypes in association studies to identify the association between genetic variants and phenotypes. We perform simulation studies to compare the power of multivariate analysis of variance (MANOVA), trait-based association test that uses an extended simes procedure (TATES), and joint models of multiple phenotype (MultiPhen) with using CUR and those without using CUR.