Extended Data Fig. 2: CellRank 2 scales to large cell numbers.
From: CellRank 2: unified fate mapping in multiview single-cell data
a. Runtime (left) and peak memory consumption (right) to compute fate probabilities with CellRank (orange) and CellRank 2 (blue). Both methods were run on subsets of a reprogramming dataset containing over 100, 000 cells94. Box plots indicate the median (center line), interquartile range (hinges), and whiskers at 1.5x interquartile range (N = 10 runs each). b. The CellRank 2 adaptation of CytoTRACE scales to a mouse organogenesis atlas of 1.3 million cells31, whereas CytoTRACE fails above 80, 000 cells. Box plots indicate the median (center line), interquartile range (hinges), and 1.5x interquartile range (whiskers) (50, 000 cells, original: N = 6 runs; 60, 000 cells, original: N = 8 runs; 80, 000 cells, original: N = 9 runs; otherwise: N = 10 runs). c. Runtime for calculating macrostates and fate probabilities using the RealTimeKernel with (brown) and without (green) thresholded transition matrix. Box plots indicate the median (center line), interquartile range (hinges), and 1.5x interquartile range (whiskers) (N = 10 runs each).
