I have around 50 arrays each with 4 hybridization blocks. On each array is a control block. Specifically the control is a known genome which is hybridized against the annotated probe set on the chip. The control block gives a estimate on the number of false positives/negatives and is used to asses the validity of the experiment. Arrays can be stripped and re-used up to four times but the signal strength fades with each re- use. Eventually I would like to combine all the arrays 150 unknown and 50 controls into one dataset. Is the following possible ? Take out the 50 controls and normalize, then use the results from the normalization to adjust the 150 remaining unknown arrays. At the moment quantile normalization gives good results. Has anyone done something similar, any thoughts ? Chris.