Figure 3 illustrates the average False Change for sample targets hybridized to three replicate HG-U133A arrays across the different spiked target concentrations at each site. Therefore, for each scanner type at each site, the False Change listed is an average from 12 samples. Error bars represent one standard deviation.
The specification for HG-U133 probe arrays is a False Change of less than 1%. The average False Change across all sites is well below 1% and is statistically indistinguishable between the GeneChip Scanner 3000 and GeneArray 2500. A possible explanation for the slight variation in False Change across the sites (for example, Figure 3, Site 1 versus Site 2) is due to differences in probe array processing, including hybridizing probe arrays on different days, using different operators, using different fluidics stations and wash buffers, etc.
Accuracy of Target Spike Detection Comparisons
The results discussed previously utilized the entire probe set population for analysis and thus, were considered global metrics measuring instrument performance. To further compare instrument performance, additional analyses (discussed next) focus on the cRNA spikes described for the alpha and beta test plans.
The use of target spikes at concentrations of 0.00 pM, 0.75 pM, 1.50 pM, and 3.00 pM in a complex biological material provided a calibrated measure of instrument performance at the limits of detection of the system. The transcripts at 1.50 pM correlate to a frequency of 1 transcript in 100,000, and are considered a reliable measu