Figure 10

3'/5'-bias of two replicated hybridizations A (left part) and B (right part) of RNA of different quality on the rat genome array RAE-230: The graph above, in the middle and below show the total log-averaged mean of the PM- and MM-intensities, Σ, taken over all 11 probe-pairs of each set, the hook curves and the signal distributions, respectively, as a function of the sub-mean, Σ_sub, averaged over subsets of the first four probes of a probe set (probes no. 1–4) closer to the 5'-end of the transcripts and the last four probes (no. 8–11) nearer to the 3'-end of the transcripts. The 5'- and 3'-biased sub-means virtually agree in the N-hybridization range whereas upon specific hybridization the 3'-biased sub-mean exceeds that of the 5'-biased one owing to 3'-biased amplification of RNA (upper panel, see arrows, the factors indicate the fold changes of the 3'-end relative to the 5'-end). The dimensions of the different hooks calculated using either all probes (1–11) or the biased subsets roughly agree each with another showing that all probes follow virtually the same hybridization law (middle panel). The higher yield for RNA fragments near the 3'-end of the transcripts gives rise to larger decay constants λ if one plots the signal-density as a function of the Σ_sub-coordinate of the respective subsets of probes (see lower part, φ is the respective mean negative logarithmic expression index). The 3'/5'-bias is larger for sample A shown in the left column of the figure. Note that the width of the respective hooks and the fraction of absent probes (42% versus 54%, see figure) increase upon decreasing RNA-quality (compare A with B). The open circles in the middle panel indicate the positions of the GADPH-probe sets used typically for 3'/5'-hybridization control. The left one in each hook refers to the 5'-biased set and the right one to the 3'-biased set.