Improved significance test for DNA microarray data: temporal effects of shear stress on endothelial genes

Statistical methods for identifying differentially expressed genes from microarray data are evolving. We developed a test for the statistical significance of differential expression as a function of time. When applied to microarray data obtained from endothelial cells exposed to shearing for different durations, the new multi-group test (G-test) identified three times as many genes as the one-way ANOVA at the same significance level. Using simulated data, we showed that this increase in sensitivity was achieved without sacrificing specificity. Several genes known to respond to shear stress by Northern blotting were identified by the G-test at P less than or equal to 0.01 (but not by ANOVA), with similar temporal patterns. The validity and utility of the G-test were further supported by the examination of a few more example genes in relation to the present knowledge of their regulatory mechanisms. This new significance test may have broad application for the analysis of gene-expression studies and, in fact, to other biological studies in general.