Figure 1

Sample size estimation without experimental pilot data using the PPCA model. In each panel is the estimated FDR (solid red lines) as well as the 10^th and 90^th percentiles (dashed red lines). A horizontal dashed black line is the target FDR at 5%. (A) The sample size $\widehat{n}$ is estimated to be 30 with 15 samples in each treatment group. (B-D) show the effect of varying the proportion of significant bins over a range of sample sizes.