Microarray data were imported into Genespring using Agilent's two-color 'Enhanced FE' import scenario which included 'Per Spot: Divide by control channel' and 'Per Chip: Normalize to 50th percentile' normalization steps.
Data from these remaining features were normalized using the per chip normalize to 50th percentile and the per gene normalize to median parameters in Genespring version 7.2 software, as described above.
Median percentile normalization was performed utilizing the "normalize to median or percentile" option in GeneSpring GX 7.3 (Silicon Genetics, Redwood City, CA).
After normalization (signals below 0.01 were taken as 0.01; per chip: normalize to 50th percentile; per gene: normalize to specific sample), probesets representing intergenic regions and control genes were removed, as well as nonchanging genes (between 0.667- and 1.334-fold 1.334-fold
The cancer-specific signal across all probes was normalized as a ratio to baseline using Normalize to Baseline Tool in PGS, where baseline data corresponded to normal human lymphocyte DNA.
The second normalization adjusts the total signal of each chip to a standard value ("normalize to 50th percentile") determined by the median of all the reliable values on the chip; this renders the output of each chip comparable with that of every other chip in the study.
