Supporting Information Washing scaling of microarray expression

1. Hook analysis of the washing experiment of Skvortsov et al. To illustrate the benefits of the hook approach we transform the intensity-data of a second washing experiment published recently by Skvortzov et al. [1] into hook-graphs (see Supplementary Figure 1). In this particular experiment the authors labelled only a small number of 18-20 transcripts each of which is interrogated by one 'spiked probe set' to filter out the effect of specific hybridization. With this limited number of labelled transcripts the smoothing algorithm of the standard hook method partly fails because it takes running averages over the intensities of about one hundred probe sets, a number that largely exceeds the number of labelled transcripts. As a consequence, the obtained hook curves lack the S-, sat-and as-ranges because of the small number of strongly expressed probe sets (see part a of Supplementary Figure 1 and Figure 7 for the definition of the hybridization ranges). In part b of Supplementary Figure 1 we therefore plot the individual probe-set level data without smoothing. The obtained data clouds are well described by theoretical hook curves. The spiked probe sets (solid symbols) essentially accumulate in the sat-range due to their relatively strong specific hybridization with the spikes. Contrarily, the probe sets not interrogating the targets (so-called 'empty' probe sets) are mostly found in the N-hybridization range owing to cross hybridization with the spikes (see the open symbols). The smoothed hook curves essentially reflect the behaviour of the N-range which is most strongly affected by washing. The 'washing-cycle experiment' of Skvortsov et al. uses a modified fluidic script similar to that applied to chip A in our study: Particularly, it scans the array once before and two-times after stringent washes (see ref. [1] for details). The respective hook data undergo virtually the same changes as in our study (compare part a and b of Supplementary Figure 1 with Figure 8). Namely, the N-range shifts markedly towards smaller values paralleled by the increase of the height of the hook curve. Repeated washing is much less effective than the first washing step. Note that the results also agree quantitatively with ours: the observed shift of the N-range, δβ N ~0.8, agrees in both experiments despite the different chip types used by us (Human HG-U133plus2 array) and Skvortsov et al. (Drosophila DG-1 array). In a second 'hybridization time experiment' Skvortsov et al. hybridized two microarrays differently before washing, one …