Extended Data Fig. 1: SWIFT-seq identifies rare CTCs that reflect BM biology and burden.
a) Heatmap of V(D)JC gene (y-axis) usage in all tumors (n = 75) (x-axis). Tiles are colored orange when the gene is used in the corresponding tumor’s clonotype. b) UMAP embedding of plasma cells (n = 1,292,479) colored by patient ID (top) or disease stage (bottom). c) Heatmap of correlation coefficients between BM tumor cells and CTCs in patients with at least 30 tumor cells in each compartment (n = 51). Correlation coefficients were computed using two-sided Spearman’s tests. d) Scatter plot of BM biopsy (BMBx) plasma cell infiltration (%) (x-axis) and the proportion of CTCs out of all plasma cells (y-axis) in patients (n = 71). For patients with two tumors (n = 3), secondary tumors were excluded. Two patients with MGUS had no BM infiltration information. A line (black) was fit using the “lm” method (95% confidence interval in light orange). Patients were colored by disease stage. The correlation coefficient and p-value were computed using a two-sided Pearson’s test (r = 0.56, p = 3.4e-07).
