Data was gathered from the scanned image using Genepix (Axon Instruments). A grid containing a series of circles corresponding to the spots of the array was laid over the image, and the circles manually adjusted to each spot, a process known as "masking" or "gridding". The intensity of each pixel within the circle was calculated for each channel, termed "foreground" or "feature". Background for each spot was measured in a two pixel area outside of the circle. For each spot on the microarray, the mean pixel intensity of both foreground and background was multiplied by the number of pixels in the spot. The background was then subtracted from the foreground.
For each microarray hybridization, the Cy5 channel was normalized such that the total Cy5 signal equals the total Cy3 signal, a process necessary to avoid inaccurate measurements that would occur if one channel were consistently brighter than the other. The Cy3/Cy5 ratio for each gene was calculated.
For direct experiments (i.e. A versus B), the average of the log2 of the expression ratio for each repeat is used.
For indirect experiments (i.e. A/ref versus B/ref), the difference between the logs of the expression ratios are used (log2(A/ref) &endash; log2(B/ref)).