Figure 2

Data normalisation by linear regression. y_n is the Cy3 signal from each probe from array n;

x is the mean Cy3 signal from each probe across all 80 arrays.

is the linear corrected Cy3 signal from each probe from array n given by:

= Y_n-((α + βx)- x) where α and β are the intercept and slope of the linear regression of y_n on x.

is the Loess corrected Cy3 signal from each probe from array n given by:

where α and β are the set of intercepts and slopes of the Loess linear regression of y_n on x, dividing the range of x into 10 equal segments.

Panels a) and b) show data for n = 1. c), d) and e) show data for n = 3. The dashed line shows x = y; the solid line shows the fitted linear regression. The thick solid line shows the fitted LOESS regression.