Identifying association with a gene expression signature using openSESAME. a) Raw CEL files are normalized to obtain gene-level expression measurements. Here, red and blue indicate genes with high or low levels of absolute expression, respectively. b) Within each experiment, expression measurements for each gene are z-score normalized to produce relative expression levels. Here, red and blue indicate high or low expression relative to the other samples in the experiment. c) For each sample in each experiment, a Wilcoxon test is performed between the relative expression values of the "up" and "down" gene sets defined by the query signature, yielding a signature association (SA) score that varies from positive (yellow) to negative (magenta). d) A Kolmogorov-Smirnov (top) or Fisher's exact (bottom) test is performed to determine whether the distribution of SA scores in each experiment is significantly different from the distribution of SA scores across all samples in the repository.