Standard error and distribution of log fold changes within conditions on MAQC dataset. (a) The scatter plots depict the standard errors of the null distribution L
. The neighborhood of each gene i contains 50 genes with similar expression strength. The green line is fitted by using R function "lowess". (b) There are three curves (green, blue, and red) representing the null distributions of log fold changes estimated from genes with the normalized read counts of around 8, 20, 500, respectively. The black curve represents the null distribution estimated from all genes regardless of their expression levels. Each dashed line indicates the 90% quantile of the respective distribution.