Four multi-trait GWAS methods with per trait interpretation.

1.5 million variants were simulated with z-scores for four traits with 10% of variants being truly associated. The first column lists which trait has a genetic effect. The second column is the number of variants simulated under this specific model. The third column is the genetic effect size of the variant. The remaining columns are split by trait where the performance of the four methods are shown for each trait. These 16 columns present the number of variants identified as associated by each method for the specific trait. MTAG uses p-values, ASSET uses the optimal subset, while PAT and HIPO use the m-value framework to provide per trait associations.