FIG 3.

Cluster-based diversity scoring reveals strong differences between healthy individuals and patients with cancer. In our analysis, using data from Zheng et al,²¹ we evaluated our ability to score significant differences in cluster diversity across healthy and acute myeloid leukemia (AML) samples pre– and post–bone marrow transplant (BMT). (A) As a first indicator of between-sample or between-condition differences, we used the Kolmogorov-Smirnov (KS) distance for discrete probability mass functions. (B and C) Little difference was found in the KS distance within condition differences, except in (D) post-BMT samples. (E-G) Between-condition differences were larger when comparing pooled samples across conditions. (H) We calculated a general diversity index (GDI) ^qD to quantify diversity across orders of diversity q. (I) For all orders of diversity measure, patients with AML (pre- and post-BMT) had a higher diversity index compared with healthy individuals (two samples per condition), which suggests that GDI can be used as a metric for stratification.