Stephen Quake and colleagues report a method for reconstructing the clonal architecture of pediatric B cell acute lymphoblastic leukemia (B-ALL) samples using a microfluidics platform and novel computational approach (Proc. Natl. Acad. Sci. USA doi:10.1073/pnas.1420822111; 25 November 2014). The authors performed exome sequencing on bone marrow and matched saliva samples collected during remission from six children diagnosed with B-ALL. Mutations in regions of increased tumor heterogeneity in the exome data were confirmed through targeted sequencing. The authors found 46 mutations per individual, on average, but a wide variability between individuals. They carried out whole-genome amplification for 1,479 single cells from the 6 children to separate mutations into specific clones and reconstruct the clonal evolution of the sample using, in parallel, a model-based probabilistic algorithm and data-driven clonal clustering. These methods, which were designed to filter out low-quality single-cell sequences, were validated using simulation tests. Among their findings, the authors determined that the majority of structural alterations occurred before point mutations and found that KRAS mutations occurred at later stages of disease. Evaluation of interclonal IgH sequences showed that clones within the same individual might arrest at different stages in B cell differentiation.