(PCHAT) [4], and a Trait-based Association Test that uses Extended Simes procedure (TATES) [15]. These can be classified into direct, indirect and univariate-based methods (Figure 1). MV-SNPTEST, MultiPhen and MV-BIMBAM are direct MV methods, in which the effects of the genetic variant are modeled directly on the traits without changing the general format and nature of the trait data. MV-SNPTEST [20] and MV-BIMBAM [22] are both based on a Bayesian multivariate regression analysis, but MV-BIMBAM additionally partitions the traits into three groups: 1) traits that are unaffected by the genetic variant, 2) traits that are directly affected by the genetic variant, and 3) traits that are indirectly affected by the genetic variant through directly affected traits. MultiPhen identifies the linear combination of traits most associated with each genetic variant by applying a reversed ordinal regression, such that genotype (allele count) is regressed on a collection of traits [12]. MV-PLINK [7], PCHAT [4] and UV-PCA are indirect methods based on a reduction of the trait dimension. In MV-PLINK the association between a set of traits and a genetic variant is assessed using canonical correlation analysis. Specifically, the linear combination of traits that maximizes the covariance between the genetic variant and all
