Probe sets for which expression was significantly different between exposed and unexposed individuals were identified using a standard paired t-test, and a recently developed bootstrapping technique to provide a critical value adjusted to provide a 5% familywise error rate (FWER), the standard value used in the literature.
RMA analysis of the data using paired t-statistics, bootstrapping, and a 5% FWER indicated that 2,129 probe sets were significantly different in people exposed to high levels of benzene compared with matched unexposed subjects.
Although the results will differ based on both the type of processing (e.g., RMA) and adjustment for multiple testing (e.g., FWER with bootstrapping), our results showed that, in the six pairs examined, a potentially large number (> 2,100) of probe sets were (statistically) differentially expressed in the benzene-exposed subjects compared with the control, unexposed subjects.