Enrichment test for differentially expressed HRS genes in MBLCL. A graphic rendition of the computation of the KS scores for HRS genes that are expressed at low (A) and at high (B) levels in MLBCL is shown. The bottom panel for each signature indicates with vertical bars the location of the HRS genes within the ranked set of 15 000 differentially expressed genes in MLBCL and DLBCL (with the MLBCL genes to the left and the DLBCL genes to the right). The top panel shows the assignment of “rewards” and “penalties” to the overall KS score as the list of 15 000 ranked genes is scanned from the MLBCL end (left) to the DLBCL end (right). Every “hit” (ie, the encounter of an HRS gene during the scan) increases the KS score, and every “miss” (the encounter of a non-HRS gene) decreases the score, resulting in the indicated stepwise curve. The final score corresponds to the highest value (on the y-axis) in the plot. High enrichment would correspond to a steep climb upward to the left, whereas lack of enrichment would correspond to a lack of clear upward trend in the curve. The KS statistic indicates that genes reported to have low levels of expression in HRS cell lines are similarly decreased in MLBCL (p_max = .012, p_min < .001). In contrast, the similarity between genes with higher levels of expression in HRS cell lines and primary MLBCL was less significant (p_max = .213, p_min = .007), likely reflecting the importance of tumor microenvironment to primary MLBCL and cHL signatures⁴⁵  and the absence of these features in the signatures of isolated HRS cell lines maintained in vitro.¹⁹