Gene expression profile and clustering of NPM1-mutated AML. (A) Differentially expressed and predictive genes in NPM1e12 AML (n = 125) compared with control AML (n = 305) using EPCY. Overexpressed (orange) and underexpressed (blue) genes are defined by Log 2 Fold Change (L2F) >|0.5| and NPM1e12 predictive power is defined by Matthews Correlation Coefficient (MCC) >0.2. Selected genes are labeled, and HOXA/B genes are colored separately. (B) Uniform manifold approximation and projection (UMAP) performed on the complete log transformed Leucegene cohort (n = 430) using the most predictive genes from supplemental Table 2 followed by principal component analysis and colored by Louvain-based clusters (left)¹⁹  or mutational/cytogenetic group (right). Barplot shows the proportion of genetic groups per cluster; the number of samples per group is indicated in parentheses. (C) UMAP colored by NPM1 mutation type.