The intensities were also normalized across arrays using a median absolute deviation scaling method [ 22].
As a first step the Cy3 and Cy5 intensities without background subtraction were converted to M (=log2 [red channel/green channel]) and A (=log2 [red channel*green channel]/2) format using the median intensity values for both channels.
The data were renormalized using the median intensity on each array.
Tumor-to-normal ratios of the fluorescently labeled antibodies were calculated with ImageJ using the median intensity scores.
Microarray data was integrated into the network using the median intensity value as the node weights in the network.
The corresponding scatter plots were tested for linearity as described previously [ 20], and normalized by global normalization using chip median intensity.
Cy-dye emission signals were scaled in QuantArray using the median intensity of each channel, and any visible hybridization artifacts were flagged and recorded as absent during data filtering and analysis.
The non-signal background was determined using the median intensity value plus one standard deviation for the blank positions and negative controls, and the maximum value for all arrays was set as the "no signal" value.
For the pooled reference, we randomly selected 100 female genotype data and used the median intensity values of each data point as a pooled reference value.
Normalization of the data was performed using a global median normalization excluding features with intensity in either channel lower than 200 and the 10% most intense and least intense reporters.
