Gene expression profiles of pediatric acute leukemia with MLL chimeric fusion genes. (A) Multidimensional scaling plot generated using unsupervised principle components analysis with a combined data set containing 130 AML cases, 132 ALL cases,⁴¹  and 5 additional T-lineage ALL (T-ALL) cases that contain MLL chimeric fusion genes. A variation filter was applied to remove any probe sets that showed minimal variation in expression across this data set, and the analysis was performed with the remaining 17 944 probe sets. Each case is represented by a colored sphere, with AML cases indicated by blue, B-progenitor lineage ALL (B-ALL) by yellow, and T-ALL by green. Acute leukemia cases cluster based on lineage. (B) The same PCA analysis as shown in panel A, except that cases that contain an MLL chimeric fusion gene are indicated in red. The cases containing the MLL chimeric fusion gene continue to cluster according to lineage. (C) Multidimensional scaling plot generated using the supervised learning algorithm, discriminants analysis with variance (DAV) with the expression data from the 267 acute leukemia samples generated using the 17 944 probe sets that passed the variation filter. Cases are color coded as described for panel B. Cases with an MLL chimeric fusion gene (in red) can be separated in gene space from the leukemias that lack this genetic lesion. (D) Expression profile of the top 50–ranked MLL discriminating genes. The probe set number and gene symbol for the discriminating genes are listed on the right. The normalized expression level for each gene is represented by color using the scale shown.