Similar to simulation scenarios with two out of three traits associated to the QTL, UV-PCA and UV-MA showed comparable power for all scenarios with rE>0. Again, for scenarios with rE = 0, UV-PCA performed better in case of a negative genetic correlation and UV-MA performed better in case of a positive genetic correlation. UV-MA outperformed all methods when rE was 0 and the genetic correlation was positive; for rE = 0.3 and a positive genetic correlation, UV-PCA and UV-MA performed best of all methods. Power of UV-MA was the same as that of UV analysis for all scenarios with a negative genetic correlation.
