Skip to main content

Advertisement

Figure 3 | BMC Genomics

Figure 3

From: Autocorrelation analysis reveals widespread spatial biases in microarray experiments

Figure 3

Widespread autocorrelation patterns in microarray studies. Autocorrelation analysis of individual microarrays from the Spellman et al., 1998 [27] α-factor (A), Cho et al., 1998 [24] (B; performed on Affymetrix microarrays), our unpublished cell cycle data (C), Hardwick et al., 1999 [44] (D), and Posas et al., 2000 [45] (E) datasets. The autocorrelation values are represented by a colorcode instead of a curve as in figure 1, and each individual experiment is represented by a seperate column in the plots. Periodic autocorrelations can be observed in most individual microarray experiments in these datasets. However, the variability of both the magnitude and the actual periods within a given dataset indicates that this effect occurs in a stochastic, rather than systematic manner. Note that in most microarray designs, both a two-gene period as well as at least one additional characteristic period could be observed in different or even the same experiment. (F) Autocorrelation analysis of 2005 yeast microarray experiments. Significant autocorrelation periodicities are manifested by values that are visually different from zero, showing that most experiments exhibit periodic autocorrelations. (G) A set of 340 experiments from a single microarray printing source are completely devoid of autocorrelation signals.

Back to article page