Inflammation and Immune System Activation in Aging: A Mathematical Approach

Nikas, Jason B.

Statistical methods. I first assessed the quality of the data by examining the expression of all housekeeping genes by all 40 subjects (25 old and 15 young). With regard to the housekeeping genes, there was no statistically significant differential expression between the two groups (Supplementary Table 7 ). As can be seen by the results of K-Means clustering analysis in Fig. 1a , the two groups cannot be discriminated based on the expression of the housekeeping genes. Having, thus, established the quality of the data, I investigated for any differential expression among all gene variables using three different and independent methods. 1) Using a methodology that I have developed and introduced previously 38-42 , 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. Table 8 . In order to minimize the number of false negatives in the case of the third method 43, 44 , for the final selection of significant variables, I imposed the condition that if a given gene variable met the significance criteria of all three methods, or those of the first method and those of only one of the other two methods, it would be deemed significant. Excluding multiplicities (different transcripts that corresponded to the same genes), thirty six genes made up the final list of the most significantly differentially expressed genes between the two groups, as assessed by the aforementioned three different and independent methods of statistical significance (Table 1) .