Fig. 6

Runtime efficiency and scalability of the six DA testing methods. A Runtime of the six DA testing methods on the four real single-cell datasets. The error bars reflect the standard deviation of runtime across various target DA cell populations, DA ratios, and random seeds. The execution times were measured on the nodes of a cluster with Intel Xeon E5-2680 v3 CPUs and 256GB RAM. B Relative runtime growth ratio of the six DA testing methods on the single-cell datasets as a function of an increasing number of cells (4k, 10k, 15k, 30k, 50k, and 100k). The runtime of the smallest dataset was used to normalize the runtimes of the larger datasets