FIGURE 4.

Rejection rate (number of p‐values <.05 across 2000 simulations) in testing the treatment effect using log‐ratio RDA. In (A), the fold change is increased for several levels of ${\sigma }_a$ under independence of the treatment with the library size ($\gamma =0$). The type 1 error is controlled, but the power is reduced as ${\sigma }_a$ increases. In (B), there is no treatment effect (fold change is 1, $b=0$), but there is an increasing correlation between treatment and library size (set with $\gamma \ge 0$) for three levels of ${\sigma }_a$. The type 1 error is controlled for $\gamma =0$, but increases for higher values of $\gamma$ and ${\sigma }_a$