TATES cannot be used directly to test specific hypotheses concerning the underlying genotype-phenotype model. However, as TATES outputs the p-values from the original univariate GWAS analyses along with TATES' trait-based p-values, further inspection of the pattern of significant univariate tests that drive the significant TATES p-value can be informative. For instance, if a significant TATES p-value is driven by an association with only one of the multiple phenotypes, then the true genotype-phenotype model is unlikely to be a 1-factor model with the GV effect on the factor. The more these phenotype-specific GV effects are observed, the stronger the indication that the trait under study is genetically heterogeneous. This, again, implies that multivariate approaches, in which the heterogeneity is accommodated, are more likely to reveal the genetic architecture of that trait than the standard approach based on univariate composite scores. Furthermore, if one aspires to actually test specific hypotheses concerning the underlying genotype-phenotype model, TATES can be used as a filter to reduce the number of SNPs to a computationally manageable set. The exact location and role of the selected SNPs may then be studied in detail in appropriate multivariate models [4]–[5].
