The paper introduces STORIES (SpatioTemporal Omics eneRgIES), a novel computational method for inferring cell fate trajectories from time-series spatial transcriptomics data, which captures gene expression within a tissue's physical context. STORIES is built upon an extension of Optimal Transport theory called Fused Gromov–Wasserstein (FGW) to learn a spatially informed differentiation potential, effectively modeling cellular dynamics on a Waddington-like epigenetic landscape. The method's core innovation is its use of the quadratic FGW term to account for spatial coherence without requiring rigid alignment of tissue slices across time points, a challenge faced by existing methods. Benchmarked against state-of-the-art tools using large Stereo-seq atlases from axolotl regeneration and mouse development, STORIES demonstrates superior performance in predicting cell states and recovering known biological pathways and putative driver genes. The authors present STORIES as an open-source Python package that integrates with existing single-cell analysis ecosystems.

References:

• Huizing G J, Samaran J, Capocefalo D, et al. STORIES: learning cell fate landscapes from spatial transcriptomics using optimal transport[J]. Nature Methods, 2025: 1-10.