Figure 10

QQNP with simulation envelopes (based on 1,000 random samples from a normal population) for residuals of nine datasets from Hughes et al. [18] after prominent outlier removal. Two-sided 99.9% cut-off and robust measure of scale (median absolute deviation) for residuals were used to remove outliers. Channel intensity values were log(base 2)-transformed and normalized. The envelopes are depicted as dashed red lines.