Fig. 2: Benchmark of STORIES on three large datasets.
From: STORIES: learning cell fate landscapes from spatial transcriptomics using optimal transport
a, Visual representation of the three datasets in our benchmark. From left to right: an axolotl brain regeneration dataset, zebrafish development and mouse development. Slices in each dataset are split into a train set (blue), an early test set (orange) and a late test set (green). b, Wasserstein distance between predicted and ground-truth gene expression in early test sets (orange) and late test sets (green) across the three datasets. Scores are reported for n = 10 initialization seeds. In the box plots, the center line, box limits and whiskers denote the median, upper and lower quartiles and 1.5 times the interquartile range, respectively. c, Cell-type transition accuracy on the mouse development dataset in the early test set and late test set. Scores are reported for n = 10 initialization seeds. In the box plots, the center line, box limits and whiskers denote the median, upper and lower quartiles, and 1.5 times the interquartile range, respectively. d, Visual representation of the linear OT matching between the gene expression predicted by STORIES (top) or PRESCIENT (bottom) and the target gene expression measured at the next time point for adaxial cells of the zebrafish development dataset. The left slice displays the positions of cells belonging to that cell population, and the right slice displays the cells they are matched with at the following time point. e, Same plot as in d, but for somite cells of the zebrafish development dataset.
