A Fully Automated Parallel-Processing R Package for High-Dimensional Multiple-Phenotype Analysis Considering Population Structure

Abstract

A typical genome-wide association study is conducted through a single-phenotype analysis of the correlation between each phenotype and genotype one at a time. Alternatively, a multiple-phenotype analysis of the correlation between multiple phenotypes and a genotype often has many advantages over single-phenotype analysis. For example, statistical power in the association test may be increased in a multiple-phenotype analysis and thus may detect small effects that cannot be identified in a single-phenotype analysis. Of the several multiple-phenotype analytical methods that have been proposed, generalized analysis of molecular variance for mixed-model analysis (GAMMA) is used to analyze many phenotypes simultaneously while considering the population structure. This method shows higher accuracy than the other methods. However, GAMMA has not been widely used because no automated and user-friendly software is available; this is also the case with most other multiple-phenotype analysis methods. In addition, the lack of a parallel-processing option, which is essential in a genome-wide-association-studies analysis, is also prevalent in GAMMA. In this study, we propose an easy-to-use R package for GAMMA called GAMMA Renew (GAMMAR) that performs multiple-phenotype analysis using parallel processing. We evaluate GAMMAR using a recently published yeast dataset to locate trans-regulatory hotspots.