Performance Data & Analysis
In both the internal and external test plans, the performance of each scanner was assessed on the basis of both single array and comparison analyses generated within the MAS software. Analysis output files (.chp files) were created at Affymetrix by globally scaling all probe sets on image data files (.dat files) to a target intensity of 250. Data sets were generated to assess overall Signal correlation, global percent Present calls, and False Change between replicates. In addition, spiked transcript Detection call accuracy, Change call accuracy, and Signal Log Ratio accuracy between scanner types were evaluated.
Data from each set of scanners were tested for statistical significance, where appropriate, by applying paired Students t-tests4 for both the HG-U95Av2 and HG-U133A probe arrays. The null hypothesis was defined as the GeneArray 2500 performance significantly exceeded the GeneChip Scanner 3000 performance at a p>0.95 confidence level. Therefore, p-values less than or equal to (≤) 0.95 were interpreted as contradicting the null hypothesis; p-values approaching 0.5 were interpreted as equivalent scanner performance; and p-values <0.05 were interpreted as significantly better performance of the GeneChip Scanner 3000.
Results & Discussion
Taken together, all array data comparing the GeneChip Scanner 3000 to the GeneArray 2500, from both internal alpha testing and external beta testing, indicate that there is no significant performance difference between the two scanner types. Both scanner types produce high-quality data for Affymetrix GeneChip expression analysis.
Results from both HG-U95Av2 and HG-U133A probe arrays were independent of the scann