Figure 2

Background-subtracted intensities vs. Langmuir model. Background-subtracted intensities (numbers, probe numbering follows Affymetrix convention) plotted as functions of RT log(K/K₀) = ΔG + RT log α for six different probe sets in the experiments of Ref. [2]. As additional information we also plot the raw fluorescent intensities (crosses), so that the effect of background subtraction can be viewed by comparing the numbers with their corresponding crosses on the same x value. The background level is probe dependent, and was calculated using the algorithm of Ref. [13]. Typically, background subtraction has little effect on the higher intensity data, while its effect is more important for lower intensities. The lines shown are plots of Eq. (2) for three different values of the concentrations. A perfect agreement with the ILM would imply the alignment of all background-subtracted intensities (numbers) along one single line which would identify a single concentration for the whole probe set. Apart from few outliers the background-subtracted intensities (numbers) follow well the prediction of the Langmuir model (the raw fluorescent intensities (crosses) are not expected to follow the model). The solid line corresponds to the median value of the concentration, whose value, in picomolar (see text), is reported in the upper left corner of each graph. The two dashed lines are obtained from the median absolute deviation of the logarithm of the concentrations and measure the dispersion in the values of the concentrations within each probe set.