In addition to the analyses performed using all of the arrays, we tested the effects of number of arrays on both true and false positives. To mimic smaller experiments testing the effects of interferon, we generated smaller virtual experiments from the MAS5 data by randomly selecting samples without replacement from within each treatment group to generate sample sizes of 3, 4, 5, 6, 7, and 8 arrays per group. 1000 permutations were completed for each sample size, and the results were tallied and averaged as described above.
