Data analysis was performed using the GeneSpring array analysis software package (GeneSpring GX 7.31, Agilent Corporation, Carmel, IN). Before analysis, a qualified data set was obtained by performing a bi-level quality control. First, at the replicate level, principal component analysis (PCA) and hierarchical cluster analysis were performed to identify outlier samples from within each biological replicate group that fell distal to cohort replicates comprising each treatment group. Secondly, filtering at the gene level was performed, removing those genes with signal to noise ratios (S/N) below 3.0 in 6/12 arrays per group. This quality analysis resulted in the removal of 9 total outlier arrays from the experiment and produced a qualified gene list of 12,655 genes (out of 27,000 total genes) used for all subsequent analyses. Statistical comparisons were performed using Wilcoxon-Mann-Whitney nonparametric analysis with Benjamini and Hochberg False Discovery Rate multiple testing correction. Gene signature pattern recognition was performed using Quality Threshold (QT) clustering algorithm with minimum similarity set to 0.95 and minimum cluster size set to 10 genes. Genes comprising each cluster were grouped into functionally significant clusters and
