Gene shaving of the leukemia data. (a) The projection score (black) and the gene shaving gap statistic (red) as functions of the cardinality of the variable subset. (b) Heatmap of the most informative variable subset by the projection score, consisting of 691 variables. Red - ALL, green - AML. (c) Heatmap of the most informative variable subset by the gene shaving gap statistic, consisting of 336 variables. The hierarchical clusterings in panels (b) and (c) are created using Euclidean distances and average linkage. The figures in (b) and (c) were generated using Qlucore Omics Explorer 2.2 (Qlucore AB, Lund, Sweden).