Figure 3From: Algorithm-driven Artifacts in median polish summarization of Microarray dataInverse correlation between probeset tendency to yield identical expression values and internal probeset consistency, measured as a probe linear model R2. The x axis shows a fit to a linear model of any given probeset across 3707 Arabidopsis microarrays, using probeset sample means as explanatory variable. On the y axis the fraction of 3 sample subsets yielding 3 identical arrays for a given probeset is shown (10000 randomly picked groups were selected).Back to article page