All subject samples were obtained by member COG institutions after written consent was obtained from the parents/guardians of minors upon enrolling in the trial. The study was overseen by the Institutional Review Board at Fred Hutchinson Cancer Research Center (protocol 1642, IR file no. 5236). Data on selected clinical (for example, age, presenting hematological indices and cytogenetic classification) and molecular (for example, KIT, RAS genes, NPM, WT1, CEBPA and IDH1 mutations and FLT3-ITD allelic ratios) features were clinically available before genomic analyses and are included in the clinical data file available at the TARGET data matrix. 177 cases from the adult de novo AML TCGA data set3 (link) were selected for analysis after exclusion of those with French-American-British (FAB) system M3 morphology (n = 20) or BCR-ABL1 gene fusion (n = 3), as these subtypes are not represented in the COG–TARGET AML cohort. The age distributions for the TARGET WGS discovery group and the TCGA cohort are outlined in Supplementary Table 3. DNA and RNA were extracted from Ficoll-enriched, viably cryopreserved samples from the COG biorepository using the AllPrep Extraction Kit (Qiagen). Nucleic acids were quantified by NanoDrop (Thermo Scientific). RNA samples were tested for quality and integrity using the Agilent 2100 Bioanalyzer (Agilent Technologies). The integrity of DNA samples was confirmed by visualization on a 0.8% agarose gel.