Figure 7.

Power estimates when focusing on detecting effects in rare clusters (<1% mean frequency across samples)

Power calculations for detecting effects in rare clusters across different ranges of dataset sizes at fold changes of 1.5, 2, or 4 (same fold changes as shown in Figures 6B–6D). Each tile represents power estimation from n = 100 simulations. Simulations were performed in a realistic context and frequency fold changes are induced only in original clusters that exhibited <1% mean frequency across samples.