Figure 3

Split cell expression comparisons: fitting modeling predictions to observed data. A. Simulations of microarray expression data comparing two halves of the same cell were performed for a range of transcript abundance distributions and compared to observed data. The experimental log intensity ratio distribution (broad green curve) for two halves of a neural stem cell is shown, and superimposed on this are data for (i) the three transcript distributions that are closest to the real data (blue-green points), and the three worst fit distributions for (ii) high μ values (yellow-brown), and (iii) low μ values (red-pink). X-axis: log intensity ratio (data divided into 0.10 log unit bins); y-axis: proportion of genes in each bin. B. Model transcript distributions for the best and worst fit cases, colour coded as in A. The distributions that fit the observed data best (blue-green) segregate from the poorer fit distributions. X-axis: transcript copy number for a gene; y-axis: the number of genes with a given transcript copy number. The legend shows the independent parameters determining each model distribution: G = active genes, T = total transcripts, μ = scale value of log-log-normal distribution.