Methodology for application of EPSA. A. For each perturbation in the training compendium, replicate log2 expression of treated vs. untreated cells were analyzed. B. For each perturbation, Significance Analysis of Microarrays (SAM)  software was applied to identify genes with significantly altered expression. Median values among the replicates of log2(treated/untreated) were used to represent genes in a signature. C. These values were correlated with test profiles. A false discovery rate was calculated for each perturbation, for the average level of correlation with disease profiles. D. For those signatures with a statistically significant level of correlation, survival analysis was performed using the EPSA score, or the degree of signature correlation, as a factor influencing survival. Survival analysis was carried out using the survival package in R.