the observed correlational structure between the phenotypes. Here we show that TATES has correct false positive (type-I error) rate, and that TATES picks up both phenotype-specific genetic effects as well as genetic effects that are common to multiple correlated phenotypes. Through extensive simulations, probing a wide variety of genotype-phenotype models, we demonstrate under which circumstances TATES outperforms analyses based on sum scores and MANOVA/MultiPhen with respect to the statistical power to detect causal GVs.
