Toward development of antibody-based panels for predicting TKI responses. (A) A custom 38-antibody panel was designed for mass cytometry to detect the cell abundance gradients and GE profiles uncovered by our single-cell map. (B) The mean expression of indicated markers was assessed within the Lin⁻CD34⁺CD38⁻ population of control and CML samples. (C) CD34⁺ HSPCs from the CyTOF data set was subjected to standard gating schemes to identify 6 HSPC populations: HSC, MPP, LMPP, CMP, megakaryocyte-erythroid progenitor (MEP), and GMP. MEP proportions were compared across CML groups A through C. (D) A flow cytometry panel was optimized to measure the frequency of ERPs across control and CML Groups A through C. An independent cohort of 6 patients of group A was used for this analysis, as well as the same set of group B and group C samples. (E) CD34⁻ cells from the CyTOF data set were subjected to standard biaxial gating to identify NK cells (far right). The total abundance of NK cells with the CD34⁻ population of samples and the mean expression of HLA-DR and CD7 within the NK cell population are shown. (F) A flow cytometry panel was optimized (supplemental Figure 13A) to measure the frequency of adaptive-like NK cells identified as CD57⁺NKG2C(KLRC2)⁺ cells within CD56^dim CD16^bright NK cell population. (G) A flow cytometry panel was optimized (supplemental Figure 13B) to measure the frequency of the NKG2A(KLRC1)⁺ NK subset within the CD56^dim CD16^bright NK cell population. Statistical tests for all data are shown in Figure 6; 2-tailed t test. Supplemental Figures 11-13 and supplemental Tables 25 and 26 are linked with the data in Figure 6.