Inflammation and Immune System Activation in Aging: A Mathematical Approach

Nikas, Jason B.

In greater detail, to assess statistical significance, I used to assess statistical significance, I used the following three different and independent methods. 1) ROC curve analysis. I performed ROC curve analysis on all gene variables in order to assess their discriminating power with respect to the two groups (old vs. young), and with respect to this method, I set statistical significance at ROC AUC $ 0.920. 2) Fold Change. For all gene variables, I defined fold change (FC) as the mean expression value of the old subjects over the mean expression value of the young subjects, and I set statistical Y) . 3) P-value. I used the independent t-Test for parametric gene variables (both normality and homogeneity of variance conditions were met); the Aspin-Welch unequal-variance test (AW) for gene variables that met the normality condition but not the homogeneity of variance condition; and the Mann-Whitney U test (MW) for the non-parametric gene variables, i.e., for those variables that i) the normality condition was not met or ii) the normality and the homogeneity of variance conditions were not met. Taking into account that there are 54,675 probe sets (including those of the housekeeping genes) in the Affymetrix HG-U133 Plus 2.0 chip, and using the Bonferroni correction, I set the significance level for the entire study at a 5 9.15 3 10 27 . Therefore, in order for any variable to be deemed significant according to the P-value method, the following condition must be met: P , a. Regarding the Mann-Whitney U test (MW), since none of the non-parametric variables had any sets of ties (a subject from one group having the same expression value as a subject from the other group), I used the exact probability for all MW tests.