Figure 4

Fold change error estimation from observable parameters. a) Large estimated fold changes were found to entail increasing absolute error in the fold change estimate. This can be explained from the observation that large fold changes are the result of one sample (either treated or control) being present in low amounts which in turn increases the influence of noise and poor labeling efficiency. For low degrees of added noise (noise ∈ [0.01 0.10]) the relationship is strikingly linear, for higher noise levels the fold change error is seen to increase more rapidly. b) The contour plot shows which fold change error can be expected (contour lines) for a given Error/Signal-ratio and estimated labeling efficiency (x- and y-axes). Since the two latter parameters are estimated by the method for experimentally observed spectra as well, the plot can be used to look up what would be the expected fold change error given an estimated labeling efficiency and an estimated error/signal ratio.