In this study, we used simulated data to compare the performance of six multivariate genome-wide association methods (MV-PLINK, MV-SNPTEST, MultiPhen, MV-BIMBAM, PCHAT and TATES) and standard univariate analysis, univariate PCA, and meta-analysis of univariate analyses. Our results showed that there is not a single method that performed best under all simulation scenarios. However, all six multivariate methods resulted in a higher power than UV analysis, even when only one of the traits was associated with the QTL. UV-MA only outperformed all methods when all traits were associated with the QTL and the genetic correlation was positive.
